diff --git "a/alpha-prompt_v1/checkpoint-28576/trainer_state.json" "b/alpha-prompt_v1/checkpoint-28576/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/alpha-prompt_v1/checkpoint-28576/trainer_state.json"
@@ -0,0 +1,172611 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999766709436604,
+  "eval_steps": 200,
+  "global_step": 28576,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 3.7758,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 3.7954,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 3.7898,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 3.791,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 3.878,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 3.8261,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7e-05,
+      "loss": 3.8312,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 3.8548,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-05,
+      "loss": 3.7137,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 3.5687,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011000000000000002,
+      "loss": 3.5224,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012,
+      "loss": 3.3281,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 3.161,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014,
+      "loss": 3.1596,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 3.1166,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016,
+      "loss": 3.0319,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017,
+      "loss": 2.9977,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018,
+      "loss": 2.8154,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019,
+      "loss": 2.7827,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 2.8516,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999939483335,
+      "loss": 2.7275,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999757933332,
+      "loss": 2.6713,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999455350001,
+      "loss": 2.6723,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999903173334,
+      "loss": 2.5739,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999998487083356,
+      "loss": 2.747,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999997821400057,
+      "loss": 2.6164,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999997034683448,
+      "loss": 2.5405,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999996126933542,
+      "loss": 2.5678,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999509815035,
+      "loss": 2.477,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999993948333883,
+      "loss": 2.5899,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999267748415,
+      "loss": 2.5148,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999991285601175,
+      "loss": 2.4777,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998977268497,
+      "loss": 2.5334,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999988138735554,
+      "loss": 2.3585,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999986383752947,
+      "loss": 2.3501,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999984507737168,
+      "loss": 2.38,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999982510688247,
+      "loss": 2.3845,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199999803926062,
+      "loss": 2.4353,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999978153491057,
+      "loss": 2.4507,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999975793342842,
+      "loss": 2.2851,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999973312161587,
+      "loss": 2.4644,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999997070994732,
+      "loss": 2.2418,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999967986700074,
+      "loss": 2.3765,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999996514241988,
+      "loss": 2.3221,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999962177106776,
+      "loss": 2.4146,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999995909076079,
+      "loss": 2.2785,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999955883381968,
+      "loss": 2.3752,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999952554970347,
+      "loss": 2.442,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999949105525963,
+      "loss": 2.3037,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999945535048862,
+      "loss": 2.4021,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999941843539085,
+      "loss": 2.4293,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999938030996675,
+      "loss": 2.3343,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999934097421687,
+      "loss": 2.275,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999930042814156,
+      "loss": 2.3579,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999992586717414,
+      "loss": 2.3144,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999921570501688,
+      "loss": 2.2176,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999917152796847,
+      "loss": 2.3017,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999912614059677,
+      "loss": 2.3317,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999990795429023,
+      "loss": 2.3865,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999990317348856,
+      "loss": 2.3168,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999989827165473,
+      "loss": 2.2029,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999893248788796,
+      "loss": 2.3537,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999888104890818,
+      "loss": 2.3237,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999882839960863,
+      "loss": 2.4037,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999987745399899,
+      "loss": 2.2743,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999871947005263,
+      "loss": 2.289,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999866318979757,
+      "loss": 2.2029,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999860569922532,
+      "loss": 2.2701,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999854699833658,
+      "loss": 2.2414,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999848708713208,
+      "loss": 2.3094,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999984259656126,
+      "loss": 2.18,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999836363377877,
+      "loss": 2.1799,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999983000916314,
+      "loss": 2.2362,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999982353391713,
+      "loss": 2.268,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999981693763992,
+      "loss": 2.2659,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999810220331588,
+      "loss": 2.2002,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999803381992224,
+      "loss": 2.1722,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199997964226219,
+      "loss": 2.1506,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999789342220708,
+      "loss": 2.1809,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999978214078873,
+      "loss": 2.2809,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999774818326056,
+      "loss": 2.2151,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999976737483277,
+      "loss": 2.1926,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999975981030897,
+      "loss": 2.2353,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999975212475474,
+      "loss": 2.0853,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999744318170172,
+      "loss": 2.1206,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999736390555368,
+      "loss": 2.2225,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999728341910423,
+      "loss": 2.1565,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999720172235425,
+      "loss": 2.2583,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999711881530483,
+      "loss": 2.1692,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999703469795695,
+      "loss": 2.0796,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999694937031156,
+      "loss": 2.1963,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999968628323698,
+      "loss": 2.1811,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999677508413265,
+      "loss": 2.1725,
+      "step": 93
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999668612560116,
+      "loss": 2.1964,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999965959567765,
+      "loss": 2.0809,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999650457765963,
+      "loss": 2.2315,
+      "step": 96
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999641198825173,
+      "loss": 2.2633,
+      "step": 97
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999631818855393,
+      "loss": 2.1876,
+      "step": 98
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999622317856734,
+      "loss": 2.1973,
+      "step": 99
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999612695829316,
+      "loss": 2.195,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999602952773247,
+      "loss": 2.2909,
+      "step": 101
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999959308868865,
+      "loss": 2.1511,
+      "step": 102
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999583103575642,
+      "loss": 2.195,
+      "step": 103
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999572997434344,
+      "loss": 2.0458,
+      "step": 104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999562770264882,
+      "loss": 2.147,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999552422067377,
+      "loss": 2.1316,
+      "step": 106
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999541952841956,
+      "loss": 2.1652,
+      "step": 107
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999531362588743,
+      "loss": 2.2374,
+      "step": 108
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999520651307868,
+      "loss": 2.1093,
+      "step": 109
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999950981899946,
+      "loss": 2.0743,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999949886566365,
+      "loss": 2.0776,
+      "step": 111
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999487791300572,
+      "loss": 2.167,
+      "step": 112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999476595910357,
+      "loss": 1.9666,
+      "step": 113
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999465279493147,
+      "loss": 1.9724,
+      "step": 114
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999945384204907,
+      "loss": 2.1261,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999944228357827,
+      "loss": 2.1665,
+      "step": 116
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999430604080885,
+      "loss": 2.1211,
+      "step": 117
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999941880355706,
+      "loss": 2.0932,
+      "step": 118
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999940688200693,
+      "loss": 2.151,
+      "step": 119
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999394839430648,
+      "loss": 2.123,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999382675828353,
+      "loss": 2.1598,
+      "step": 121
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199993703912002,
+      "loss": 2.0637,
+      "step": 122
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999935798554633,
+      "loss": 2.0844,
+      "step": 123
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999345458866893,
+      "loss": 2.0496,
+      "step": 124
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999332811162048,
+      "loss": 2.1444,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999320042431943,
+      "loss": 2.0019,
+      "step": 126
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999307152676732,
+      "loss": 2.0492,
+      "step": 127
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999294141896572,
+      "loss": 2.014,
+      "step": 128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999281010091622,
+      "loss": 2.0702,
+      "step": 129
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999267757262043,
+      "loss": 2.2244,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999254383407987,
+      "loss": 2.0803,
+      "step": 131
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999924088852962,
+      "loss": 2.1006,
+      "step": 132
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999227272627108,
+      "loss": 1.9807,
+      "step": 133
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999213535700614,
+      "loss": 2.1726,
+      "step": 134
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999199677750307,
+      "loss": 1.9621,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999918569877635,
+      "loss": 2.1452,
+      "step": 136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999171598778913,
+      "loss": 2.0792,
+      "step": 137
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999915737775817,
+      "loss": 2.0909,
+      "step": 138
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999143035714288,
+      "loss": 2.1541,
+      "step": 139
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999128572647447,
+      "loss": 2.0567,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999911398855782,
+      "loss": 2.1154,
+      "step": 141
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999909928344558,
+      "loss": 2.0285,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999084457310903,
+      "loss": 2.087,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999906951015398,
+      "loss": 1.9863,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999905444197498,
+      "loss": 1.9535,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999039252774093,
+      "loss": 1.9562,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999023942551496,
+      "loss": 2.0672,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999008511307378,
+      "loss": 2.0068,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999899295904193,
+      "loss": 2.067,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998977285755332,
+      "loss": 1.9957,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998961491447782,
+      "loss": 2.0257,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998945576119463,
+      "loss": 2.0212,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998929539770573,
+      "loss": 2.0337,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998913382401304,
+      "loss": 2.0843,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998897104011852,
+      "loss": 2.0952,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998880704602415,
+      "loss": 2.0291,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998864184173193,
+      "loss": 2.0475,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998847542724378,
+      "loss": 2.0223,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998830780256182,
+      "loss": 2.1662,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199988138967688,
+      "loss": 1.9596,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998796892262442,
+      "loss": 2.0822,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999877976673731,
+      "loss": 2.0138,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998762520193611,
+      "loss": 2.0572,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998745152631554,
+      "loss": 2.0095,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998727664051355,
+      "loss": 2.1748,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998710054453217,
+      "loss": 2.1394,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998692323837362,
+      "loss": 2.0004,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998674472203996,
+      "loss": 2.0097,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998656499553338,
+      "loss": 2.0392,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998638405885607,
+      "loss": 1.9423,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998620191201023,
+      "loss": 1.9983,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998601855499804,
+      "loss": 1.9226,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998583398782172,
+      "loss": 2.1775,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998564821048352,
+      "loss": 1.9279,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998546122298568,
+      "loss": 2.0429,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998527302533044,
+      "loss": 1.9846,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999850836175201,
+      "loss": 1.9362,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199984892999557,
+      "loss": 1.9731,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998470117144336,
+      "loss": 1.9784,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998450813318156,
+      "loss": 2.0241,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999843138847739,
+      "loss": 1.9179,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998411842622277,
+      "loss": 2.1113,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999839217575305,
+      "loss": 1.8587,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999837238786995,
+      "loss": 1.8519,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998352478973214,
+      "loss": 1.9495,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998332449063083,
+      "loss": 1.9097,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998312298139803,
+      "loss": 1.9739,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998292026203614,
+      "loss": 1.9311,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999827163325476,
+      "loss": 2.0319,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998251119293493,
+      "loss": 1.9876,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998230484320062,
+      "loss": 2.1169,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999820972833471,
+      "loss": 1.9311,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999818885133769,
+      "loss": 1.8186,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999816785332926,
+      "loss": 1.8765,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998146734309668,
+      "loss": 2.0021,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998125494279174,
+      "loss": 1.9363,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998104133238034,
+      "loss": 1.9467,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998082651186502,
+      "loss": 2.0427,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998061048124845,
+      "loss": 1.9089,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998039324053323,
+      "loss": 1.9065,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.9888688325881958,
+      "eval_runtime": 852.9661,
+      "eval_samples_per_second": 28.207,
+      "eval_steps_per_second": 14.104,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019998017478972191,
+      "loss": 1.944,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997995512881724,
+      "loss": 1.9667,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997973425782184,
+      "loss": 1.9624,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997951217673836,
+      "loss": 2.0128,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997928888556952,
+      "loss": 1.8815,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199979064384318,
+      "loss": 1.9884,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997883867298652,
+      "loss": 1.9044,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997861175157782,
+      "loss": 1.7887,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997838362009464,
+      "loss": 2.1397,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997815427853977,
+      "loss": 1.9701,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997792372691592,
+      "loss": 1.9973,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997769196522595,
+      "loss": 2.0358,
+      "step": 212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997745899347265,
+      "loss": 2.028,
+      "step": 213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999772248116588,
+      "loss": 1.9116,
+      "step": 214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999769894197873,
+      "loss": 2.0898,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999767528178609,
+      "loss": 1.9648,
+      "step": 216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997651500588254,
+      "loss": 1.9886,
+      "step": 217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999762759838551,
+      "loss": 1.9428,
+      "step": 218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997603575178146,
+      "loss": 1.8897,
+      "step": 219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999757943096645,
+      "loss": 1.9284,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997555165750716,
+      "loss": 2.0449,
+      "step": 221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997530779531243,
+      "loss": 1.9207,
+      "step": 222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997506272308318,
+      "loss": 1.9627,
+      "step": 223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999748164408224,
+      "loss": 1.9884,
+      "step": 224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999745689485331,
+      "loss": 1.9745,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997432024621822,
+      "loss": 2.1469,
+      "step": 226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997407033388084,
+      "loss": 1.9816,
+      "step": 227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997381921152394,
+      "loss": 2.0786,
+      "step": 228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997356687915057,
+      "loss": 1.9505,
+      "step": 229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997331333676378,
+      "loss": 2.0119,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997305858436665,
+      "loss": 1.8978,
+      "step": 231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997280262196228,
+      "loss": 1.9639,
+      "step": 232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997254544955372,
+      "loss": 1.9334,
+      "step": 233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999722870671441,
+      "loss": 2.04,
+      "step": 234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997202747473655,
+      "loss": 1.9726,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997176667233422,
+      "loss": 1.9538,
+      "step": 236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997150465994026,
+      "loss": 1.9116,
+      "step": 237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997124143755787,
+      "loss": 1.9431,
+      "step": 238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997097700519015,
+      "loss": 2.0044,
+      "step": 239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999707113628404,
+      "loss": 1.9959,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999704445105118,
+      "loss": 1.9369,
+      "step": 241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019997017644820758,
+      "loss": 1.7944,
+      "step": 242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996990717593096,
+      "loss": 2.0183,
+      "step": 243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996963669368522,
+      "loss": 1.8304,
+      "step": 244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996936500147366,
+      "loss": 1.9687,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999690920992995,
+      "loss": 2.0874,
+      "step": 246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996881798716611,
+      "loss": 1.951,
+      "step": 247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996854266507677,
+      "loss": 1.9732,
+      "step": 248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996826613303482,
+      "loss": 1.9077,
+      "step": 249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996798839104366,
+      "loss": 1.9507,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996770943910653,
+      "loss": 1.9264,
+      "step": 251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996742927722692,
+      "loss": 1.9296,
+      "step": 252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996714790540817,
+      "loss": 1.9627,
+      "step": 253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996686532365368,
+      "loss": 1.9399,
+      "step": 254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999665815319669,
+      "loss": 1.9108,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996629653035126,
+      "loss": 1.9512,
+      "step": 256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996601031881017,
+      "loss": 1.9906,
+      "step": 257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996572289734715,
+      "loss": 1.9619,
+      "step": 258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996543426596564,
+      "loss": 2.0583,
+      "step": 259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996514442466916,
+      "loss": 1.8957,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996485337346118,
+      "loss": 1.853,
+      "step": 261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996456111234527,
+      "loss": 1.946,
+      "step": 262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999642676413249,
+      "loss": 1.9591,
+      "step": 263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996397296040376,
+      "loss": 1.899,
+      "step": 264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996367706958524,
+      "loss": 2.0258,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199963379968873,
+      "loss": 1.9416,
+      "step": 266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996308165827066,
+      "loss": 1.9891,
+      "step": 267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996278213778182,
+      "loss": 1.8309,
+      "step": 268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999624814074101,
+      "loss": 1.9528,
+      "step": 269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999621794671591,
+      "loss": 2.0661,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996187631703255,
+      "loss": 2.0169,
+      "step": 271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996157195703405,
+      "loss": 1.7462,
+      "step": 272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996126638716734,
+      "loss": 1.8926,
+      "step": 273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996095960743608,
+      "loss": 1.9738,
+      "step": 274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199960651617844,
+      "loss": 1.7457,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999603424183948,
+      "loss": 2.083,
+      "step": 276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019996003200909227,
+      "loss": 1.9792,
+      "step": 277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999597203899401,
+      "loss": 1.8101,
+      "step": 278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995940756094215,
+      "loss": 1.9171,
+      "step": 279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999590935221021,
+      "loss": 1.9194,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999587782734239,
+      "loss": 1.9457,
+      "step": 281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999584618149112,
+      "loss": 1.7139,
+      "step": 282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995814414656792,
+      "loss": 1.9099,
+      "step": 283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999578252683979,
+      "loss": 1.9807,
+      "step": 284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995750518040498,
+      "loss": 1.9616,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995718388259305,
+      "loss": 1.909,
+      "step": 286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199956861374966,
+      "loss": 1.9358,
+      "step": 287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995653765752773,
+      "loss": 1.8919,
+      "step": 288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995621273028214,
+      "loss": 1.9592,
+      "step": 289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995588659323316,
+      "loss": 1.7866,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995555924638478,
+      "loss": 1.8337,
+      "step": 291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995523068974094,
+      "loss": 1.9129,
+      "step": 292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995490092330562,
+      "loss": 1.8571,
+      "step": 293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999545699470828,
+      "loss": 1.7356,
+      "step": 294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995423776107647,
+      "loss": 1.9385,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995390436529066,
+      "loss": 1.7429,
+      "step": 296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995356975972944,
+      "loss": 1.841,
+      "step": 297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995323394439684,
+      "loss": 1.9021,
+      "step": 298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995289691929693,
+      "loss": 1.7861,
+      "step": 299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995255868443374,
+      "loss": 1.7302,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995221923981143,
+      "loss": 1.8806,
+      "step": 301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995187858543408,
+      "loss": 1.9947,
+      "step": 302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999515367213058,
+      "loss": 1.8789,
+      "step": 303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995119364743077,
+      "loss": 1.8958,
+      "step": 304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999508493638131,
+      "loss": 2.019,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995050387045695,
+      "loss": 1.8476,
+      "step": 306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019995015716736653,
+      "loss": 1.9116,
+      "step": 307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994980925454607,
+      "loss": 1.6843,
+      "step": 308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999494601319997,
+      "loss": 1.8986,
+      "step": 309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999491097997317,
+      "loss": 1.9055,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999487582577463,
+      "loss": 1.9191,
+      "step": 311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994840550604774,
+      "loss": 1.9084,
+      "step": 312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999480515446403,
+      "loss": 2.061,
+      "step": 313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994769637352827,
+      "loss": 2.0198,
+      "step": 314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994733999271596,
+      "loss": 1.9255,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994698240220765,
+      "loss": 1.7521,
+      "step": 316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994662360200768,
+      "loss": 1.8283,
+      "step": 317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994626359212038,
+      "loss": 1.9359,
+      "step": 318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994590237255015,
+      "loss": 1.7781,
+      "step": 319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994553994330134,
+      "loss": 1.8472,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994517630437833,
+      "loss": 1.7692,
+      "step": 321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999448114557855,
+      "loss": 1.8144,
+      "step": 322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999444453975273,
+      "loss": 1.9698,
+      "step": 323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999440781296082,
+      "loss": 1.8078,
+      "step": 324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994370965203254,
+      "loss": 1.8788,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994333996480485,
+      "loss": 2.021,
+      "step": 326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999429690679296,
+      "loss": 1.8789,
+      "step": 327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994259696141126,
+      "loss": 1.7956,
+      "step": 328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994222364525435,
+      "loss": 2.0209,
+      "step": 329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994184911946338,
+      "loss": 1.8677,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994147338404288,
+      "loss": 1.8809,
+      "step": 331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999410964389974,
+      "loss": 1.811,
+      "step": 332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994071828433152,
+      "loss": 1.7658,
+      "step": 333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019994033892004978,
+      "loss": 1.9991,
+      "step": 334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999399583461568,
+      "loss": 1.8293,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993957656265718,
+      "loss": 1.7935,
+      "step": 336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993919356955554,
+      "loss": 1.9222,
+      "step": 337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999388093668565,
+      "loss": 1.7614,
+      "step": 338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993842395456477,
+      "loss": 1.8036,
+      "step": 339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993803733268494,
+      "loss": 1.9162,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993764950122173,
+      "loss": 1.7528,
+      "step": 341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999372604601798,
+      "loss": 1.9387,
+      "step": 342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993687020956392,
+      "loss": 1.7121,
+      "step": 343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993647874937874,
+      "loss": 1.7666,
+      "step": 344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993608607962905,
+      "loss": 1.6791,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999356922003196,
+      "loss": 1.6852,
+      "step": 346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993529711145514,
+      "loss": 1.7768,
+      "step": 347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993490081304045,
+      "loss": 1.879,
+      "step": 348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999345033050803,
+      "loss": 1.8973,
+      "step": 349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993410458757956,
+      "loss": 1.7864,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993370466054304,
+      "loss": 1.7576,
+      "step": 351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993330352397558,
+      "loss": 1.8737,
+      "step": 352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199932901177882,
+      "loss": 1.9492,
+      "step": 353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993249762226717,
+      "loss": 1.856,
+      "step": 354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993209285713607,
+      "loss": 1.835,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993168688249346,
+      "loss": 1.8941,
+      "step": 356
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993127969834435,
+      "loss": 1.8955,
+      "step": 357
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993087130469364,
+      "loss": 1.7771,
+      "step": 358
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999304617015463,
+      "loss": 1.723,
+      "step": 359
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019993005088890723,
+      "loss": 1.666,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999296388667814,
+      "loss": 1.7359,
+      "step": 361
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992922563517388,
+      "loss": 1.8893,
+      "step": 362
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992881119408956,
+      "loss": 1.7899,
+      "step": 363
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999283955435336,
+      "loss": 1.8028,
+      "step": 364
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992797868351089,
+      "loss": 1.9153,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992756061402655,
+      "loss": 1.9205,
+      "step": 366
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992714133508561,
+      "loss": 1.7963,
+      "step": 367
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999267208466932,
+      "loss": 1.9166,
+      "step": 368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992629914885433,
+      "loss": 1.8214,
+      "step": 369
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992587624157414,
+      "loss": 1.8386,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992545212485776,
+      "loss": 1.8599,
+      "step": 371
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992502679871033,
+      "loss": 1.7366,
+      "step": 372
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992460026313692,
+      "loss": 1.7519,
+      "step": 373
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992417251814282,
+      "loss": 1.9059,
+      "step": 374
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992374356373312,
+      "loss": 1.8657,
+      "step": 375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992331339991303,
+      "loss": 1.6957,
+      "step": 376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992288202668777,
+      "loss": 1.9064,
+      "step": 377
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999224494440625,
+      "loss": 1.878,
+      "step": 378
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992201565204254,
+      "loss": 1.7465,
+      "step": 379
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999215806506331,
+      "loss": 1.7703,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999211444398395,
+      "loss": 1.6892,
+      "step": 381
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999207070196669,
+      "loss": 1.7814,
+      "step": 382
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019992026839012067,
+      "loss": 1.8346,
+      "step": 383
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019991982855120614,
+      "loss": 1.757,
+      "step": 384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019991938750292862,
+      "loss": 1.827,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019991894524529343,
+      "loss": 1.6542,
+      "step": 386
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019991850177830588,
+      "loss": 1.7841,
+      "step": 387
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019991805710197142,
+      "loss": 1.8239,
+      "step": 388
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999176112162954,
+      "loss": 1.9053,
+      "step": 389
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019991716412128322,
+      "loss": 1.7905,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999167158169403,
+      "loss": 1.7313,
+      "step": 391
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199916266303272,
+      "loss": 1.8284,
+      "step": 392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019991581558028384,
+      "loss": 1.9175,
+      "step": 393
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019991536364798125,
+      "loss": 1.8694,
+      "step": 394
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999149105063697,
+      "loss": 1.8067,
+      "step": 395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019991445615545465,
+      "loss": 1.7929,
+      "step": 396
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019991400059524167,
+      "loss": 1.7875,
+      "step": 397
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019991354382573615,
+      "loss": 1.7609,
+      "step": 398
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019991308584694375,
+      "loss": 1.9315,
+      "step": 399
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019991262665886993,
+      "loss": 1.764,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.8015391826629639,
+      "eval_runtime": 852.1256,
+      "eval_samples_per_second": 28.235,
+      "eval_steps_per_second": 14.118,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999121662615203,
+      "loss": 1.7666,
+      "step": 401
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999117046549004,
+      "loss": 1.8973,
+      "step": 402
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999112418390158,
+      "loss": 1.926,
+      "step": 403
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999107778138721,
+      "loss": 1.8158,
+      "step": 404
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199910312579475,
+      "loss": 1.8269,
+      "step": 405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019990984613583006,
+      "loss": 1.8637,
+      "step": 406
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019990937848294292,
+      "loss": 1.8409,
+      "step": 407
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019990890962081926,
+      "loss": 1.761,
+      "step": 408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019990843954946478,
+      "loss": 1.9516,
+      "step": 409
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019990796826888508,
+      "loss": 1.9012,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019990749577908598,
+      "loss": 1.8334,
+      "step": 411
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019990702208007313,
+      "loss": 1.7128,
+      "step": 412
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019990654717185225,
+      "loss": 1.9252,
+      "step": 413
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019990607105442915,
+      "loss": 1.8302,
+      "step": 414
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019990559372780954,
+      "loss": 1.7996,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019990511519199923,
+      "loss": 1.8547,
+      "step": 416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199904635447004,
+      "loss": 1.8575,
+      "step": 417
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019990415449282965,
+      "loss": 1.687,
+      "step": 418
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199903672329482,
+      "loss": 1.782,
+      "step": 419
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019990318895696688,
+      "loss": 1.763,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019990270437529018,
+      "loss": 1.7777,
+      "step": 421
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019990221858445773,
+      "loss": 1.7363,
+      "step": 422
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999017315844754,
+      "loss": 1.6896,
+      "step": 423
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999012433753491,
+      "loss": 1.8067,
+      "step": 424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019990075395708473,
+      "loss": 1.7487,
+      "step": 425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019990026332968825,
+      "loss": 1.7469,
+      "step": 426
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019989977149316554,
+      "loss": 1.7135,
+      "step": 427
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001998992784475226,
+      "loss": 1.7856,
+      "step": 428
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998987841927654,
+      "loss": 1.8239,
+      "step": 429
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998982887288999,
+      "loss": 1.6931,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989779205593209,
+      "loss": 1.803,
+      "step": 431
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989729417386798,
+      "loss": 1.6568,
+      "step": 432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989679508271361,
+      "loss": 1.828,
+      "step": 433
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989629478247504,
+      "loss": 1.8274,
+      "step": 434
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998957932731583,
+      "loss": 1.786,
+      "step": 435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989529055476946,
+      "loss": 1.9052,
+      "step": 436
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998947866273146,
+      "loss": 1.9547,
+      "step": 437
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989428149079982,
+      "loss": 1.8066,
+      "step": 438
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989377514523126,
+      "loss": 1.72,
+      "step": 439
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000199893267590615,
+      "loss": 1.7774,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989275882695726,
+      "loss": 1.7855,
+      "step": 441
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989224885426412,
+      "loss": 1.9148,
+      "step": 442
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989173767254179,
+      "loss": 1.7854,
+      "step": 443
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989122528179645,
+      "loss": 1.7884,
+      "step": 444
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998907116820343,
+      "loss": 1.8088,
+      "step": 445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019989019687326155,
+      "loss": 1.8205,
+      "step": 446
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988968085548448,
+      "loss": 1.9315,
+      "step": 447
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988916362870924,
+      "loss": 1.8591,
+      "step": 448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988864519294216,
+      "loss": 1.8569,
+      "step": 449
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988812554818953,
+      "loss": 1.9802,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988760469445756,
+      "loss": 1.6164,
+      "step": 451
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988708263175262,
+      "loss": 1.5703,
+      "step": 452
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988655936008102,
+      "loss": 1.837,
+      "step": 453
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998860348794491,
+      "loss": 1.831,
+      "step": 454
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988550918986317,
+      "loss": 1.6483,
+      "step": 455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998849822913296,
+      "loss": 1.8301,
+      "step": 456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988445418385483,
+      "loss": 1.7487,
+      "step": 457
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988392486744517,
+      "loss": 1.8153,
+      "step": 458
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998833943421071,
+      "loss": 1.8274,
+      "step": 459
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988286260784696,
+      "loss": 1.6922,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988232966467123,
+      "loss": 1.6237,
+      "step": 461
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988179551258637,
+      "loss": 1.7134,
+      "step": 462
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988126015159884,
+      "loss": 1.8381,
+      "step": 463
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998807235817151,
+      "loss": 1.6303,
+      "step": 464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019988018580294168,
+      "loss": 1.8835,
+      "step": 465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019987964681528504,
+      "loss": 1.8446,
+      "step": 466
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019987910661875173,
+      "loss": 1.7553,
+      "step": 467
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998785652133483,
+      "loss": 1.7919,
+      "step": 468
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019987802259908128,
+      "loss": 1.7873,
+      "step": 469
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019987747877595727,
+      "loss": 1.7256,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998769337439828,
+      "loss": 1.6815,
+      "step": 471
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019987638750316452,
+      "loss": 1.7005,
+      "step": 472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000199875840053509,
+      "loss": 1.8814,
+      "step": 473
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998752913950229,
+      "loss": 1.7467,
+      "step": 474
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019987474152771285,
+      "loss": 1.8733,
+      "step": 475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998741904515855,
+      "loss": 1.7961,
+      "step": 476
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998736381666475,
+      "loss": 1.7295,
+      "step": 477
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019987308467290556,
+      "loss": 1.73,
+      "step": 478
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998725299703664,
+      "loss": 1.8154,
+      "step": 479
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019987197405903667,
+      "loss": 1.7869,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019987141693892315,
+      "loss": 1.8531,
+      "step": 481
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019987085861003257,
+      "loss": 1.6744,
+      "step": 482
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019987029907237168,
+      "loss": 1.9387,
+      "step": 483
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019986973832594726,
+      "loss": 1.711,
+      "step": 484
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019986917637076612,
+      "loss": 1.7568,
+      "step": 485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019986861320683502,
+      "loss": 1.7723,
+      "step": 486
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019986804883416077,
+      "loss": 1.7661,
+      "step": 487
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019986748325275024,
+      "loss": 1.7865,
+      "step": 488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019986691646261027,
+      "loss": 1.8472,
+      "step": 489
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998663484637477,
+      "loss": 1.6882,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019986577925616944,
+      "loss": 1.8372,
+      "step": 491
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019986520883988232,
+      "loss": 1.7334,
+      "step": 492
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019986463721489327,
+      "loss": 1.8774,
+      "step": 493
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019986406438120925,
+      "loss": 1.9131,
+      "step": 494
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019986349033883713,
+      "loss": 1.7944,
+      "step": 495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998629150877839,
+      "loss": 1.7146,
+      "step": 496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019986233862805654,
+      "loss": 1.861,
+      "step": 497
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019986176095966194,
+      "loss": 1.73,
+      "step": 498
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019986118208260717,
+      "loss": 1.899,
+      "step": 499
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019986060199689922,
+      "loss": 1.8272,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019986002070254512,
+      "loss": 1.7971,
+      "step": 501
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019985943819955185,
+      "loss": 1.6402,
+      "step": 502
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019985885448792654,
+      "loss": 1.8299,
+      "step": 503
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998582695676762,
+      "loss": 1.7223,
+      "step": 504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019985768343880794,
+      "loss": 1.8196,
+      "step": 505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998570961013288,
+      "loss": 1.6767,
+      "step": 506
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019985650755524595,
+      "loss": 1.8557,
+      "step": 507
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998559178005665,
+      "loss": 1.6043,
+      "step": 508
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019985532683729763,
+      "loss": 1.8455,
+      "step": 509
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019985473466544639,
+      "loss": 1.7814,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019985414128502,
+      "loss": 1.8217,
+      "step": 511
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019985354669602563,
+      "loss": 1.8071,
+      "step": 512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998529508984705,
+      "loss": 1.8467,
+      "step": 513
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998523538923618,
+      "loss": 1.8648,
+      "step": 514
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019985175567770676,
+      "loss": 1.6727,
+      "step": 515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019985115625451266,
+      "loss": 1.777,
+      "step": 516
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998505556227867,
+      "loss": 1.7031,
+      "step": 517
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019984995378253618,
+      "loss": 1.7819,
+      "step": 518
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019984935073376835,
+      "loss": 1.6425,
+      "step": 519
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019984874647649057,
+      "loss": 1.7842,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019984814101071008,
+      "loss": 1.7929,
+      "step": 521
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019984753433643427,
+      "loss": 1.7111,
+      "step": 522
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019984692645367042,
+      "loss": 1.5817,
+      "step": 523
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019984631736242597,
+      "loss": 1.8182,
+      "step": 524
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019984570706270822,
+      "loss": 1.8774,
+      "step": 525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019984509555452461,
+      "loss": 1.6988,
+      "step": 526
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998444828378825,
+      "loss": 1.7143,
+      "step": 527
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019984386891278933,
+      "loss": 1.7935,
+      "step": 528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998432537792525,
+      "loss": 1.6587,
+      "step": 529
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998426374372795,
+      "loss": 1.658,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019984201988687775,
+      "loss": 1.7545,
+      "step": 531
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019984140112805474,
+      "loss": 1.7021,
+      "step": 532
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019984078116081798,
+      "loss": 1.6885,
+      "step": 533
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019984015998517494,
+      "loss": 1.5913,
+      "step": 534
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019983953760113315,
+      "loss": 1.7834,
+      "step": 535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019983891400870014,
+      "loss": 1.6866,
+      "step": 536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019983828920788348,
+      "loss": 1.7182,
+      "step": 537
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019983766319869067,
+      "loss": 1.608,
+      "step": 538
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998370359811294,
+      "loss": 1.7012,
+      "step": 539
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019983640755520714,
+      "loss": 1.7889,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019983577792093157,
+      "loss": 1.6929,
+      "step": 541
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998351470783103,
+      "loss": 1.6166,
+      "step": 542
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019983451502735092,
+      "loss": 1.8339,
+      "step": 543
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019983388176806116,
+      "loss": 1.8181,
+      "step": 544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998332473004486,
+      "loss": 1.7042,
+      "step": 545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019983261162452102,
+      "loss": 1.8883,
+      "step": 546
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019983197474028597,
+      "loss": 1.792,
+      "step": 547
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998313366477513,
+      "loss": 1.7922,
+      "step": 548
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019983069734692463,
+      "loss": 1.7334,
+      "step": 549
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019983005683781377,
+      "loss": 1.698,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019982941512042642,
+      "loss": 1.8199,
+      "step": 551
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998287721947704,
+      "loss": 1.7954,
+      "step": 552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019982812806085346,
+      "loss": 1.7765,
+      "step": 553
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019982748271868338,
+      "loss": 1.7966,
+      "step": 554
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019982683616826802,
+      "loss": 1.7617,
+      "step": 555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019982618840961515,
+      "loss": 1.871,
+      "step": 556
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019982553944273264,
+      "loss": 1.5259,
+      "step": 557
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019982488926762831,
+      "loss": 1.6822,
+      "step": 558
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019982423788431007,
+      "loss": 1.7232,
+      "step": 559
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019982358529278585,
+      "loss": 1.8846,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019982293149306343,
+      "loss": 1.6727,
+      "step": 561
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019982227648515083,
+      "loss": 1.7664,
+      "step": 562
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019982162026905588,
+      "loss": 1.6636,
+      "step": 563
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998209628447866,
+      "loss": 1.7143,
+      "step": 564
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019982030421235092,
+      "loss": 1.7212,
+      "step": 565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019981964437175683,
+      "loss": 1.6606,
+      "step": 566
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019981898332301228,
+      "loss": 1.6258,
+      "step": 567
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019981832106612528,
+      "loss": 1.6665,
+      "step": 568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019981765760110386,
+      "loss": 1.6843,
+      "step": 569
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019981699292795607,
+      "loss": 1.6702,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998163270466899,
+      "loss": 1.7277,
+      "step": 571
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019981565995731346,
+      "loss": 1.6348,
+      "step": 572
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019981499165983482,
+      "loss": 1.5822,
+      "step": 573
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019981432215426202,
+      "loss": 1.628,
+      "step": 574
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998136514406032,
+      "loss": 1.6691,
+      "step": 575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998129795188665,
+      "loss": 1.6881,
+      "step": 576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019981230638906,
+      "loss": 1.6989,
+      "step": 577
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019981163205119186,
+      "loss": 1.7896,
+      "step": 578
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019981095650527028,
+      "loss": 1.5593,
+      "step": 579
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019981027975130342,
+      "loss": 1.5862,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019980960178929947,
+      "loss": 1.8094,
+      "step": 581
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019980892261926657,
+      "loss": 1.7799,
+      "step": 582
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019980824224121303,
+      "loss": 1.7546,
+      "step": 583
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019980756065514707,
+      "loss": 1.6407,
+      "step": 584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019980687786107693,
+      "loss": 1.761,
+      "step": 585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019980619385901084,
+      "loss": 1.8104,
+      "step": 586
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001998055086489571,
+      "loss": 1.6515,
+      "step": 587
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019980482223092404,
+      "loss": 1.7343,
+      "step": 588
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019980413460491992,
+      "loss": 1.5467,
+      "step": 589
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019980344577095308,
+      "loss": 1.7256,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019980275572903185,
+      "loss": 1.7352,
+      "step": 591
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019980206447916458,
+      "loss": 1.621,
+      "step": 592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019980137202135967,
+      "loss": 1.6919,
+      "step": 593
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019980067835562547,
+      "loss": 1.6409,
+      "step": 594
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019979998348197038,
+      "loss": 1.7146,
+      "step": 595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019979928740040278,
+      "loss": 1.8506,
+      "step": 596
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019979859011093115,
+      "loss": 1.6419,
+      "step": 597
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019979789161356392,
+      "loss": 1.7103,
+      "step": 598
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019979719190830951,
+      "loss": 1.6468,
+      "step": 599
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019979649099517642,
+      "loss": 1.6177,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 1.677674651145935,
+      "eval_runtime": 855.2981,
+      "eval_samples_per_second": 28.131,
+      "eval_steps_per_second": 14.065,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019979578887417313,
+      "loss": 1.6717,
+      "step": 601
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019979508554530812,
+      "loss": 1.6836,
+      "step": 602
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997943810085899,
+      "loss": 1.716,
+      "step": 603
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019979367526402704,
+      "loss": 1.7325,
+      "step": 604
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019979296831162802,
+      "loss": 1.7099,
+      "step": 605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019979226015140147,
+      "loss": 1.6586,
+      "step": 606
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997915507833559,
+      "loss": 1.6463,
+      "step": 607
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019979084020749988,
+      "loss": 1.5771,
+      "step": 608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997901284238421,
+      "loss": 1.7918,
+      "step": 609
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997894154323911,
+      "loss": 1.7063,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997887012331555,
+      "loss": 1.7194,
+      "step": 611
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019978798582614401,
+      "loss": 1.6628,
+      "step": 612
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019978726921136527,
+      "loss": 1.5552,
+      "step": 613
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019978655138882788,
+      "loss": 1.6591,
+      "step": 614
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019978583235854064,
+      "loss": 1.7087,
+      "step": 615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019978511212051216,
+      "loss": 1.7003,
+      "step": 616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997843906747512,
+      "loss": 1.7112,
+      "step": 617
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019978366802126649,
+      "loss": 1.6883,
+      "step": 618
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997829441600668,
+      "loss": 1.8531,
+      "step": 619
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019978221909116084,
+      "loss": 1.5988,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997814928145574,
+      "loss": 1.7019,
+      "step": 621
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997807653302653,
+      "loss": 1.6299,
+      "step": 622
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019978003663829333,
+      "loss": 1.6963,
+      "step": 623
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997793067386503,
+      "loss": 1.6828,
+      "step": 624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000199778575631345,
+      "loss": 1.9357,
+      "step": 625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997778433163864,
+      "loss": 1.6607,
+      "step": 626
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019977710979378325,
+      "loss": 1.7801,
+      "step": 627
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019977637506354446,
+      "loss": 1.62,
+      "step": 628
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019977563912567897,
+      "loss": 1.6274,
+      "step": 629
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019977490198019564,
+      "loss": 1.4263,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019977416362710338,
+      "loss": 1.532,
+      "step": 631
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019977342406641118,
+      "loss": 1.4962,
+      "step": 632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019977268329812793,
+      "loss": 1.5777,
+      "step": 633
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019977194132226263,
+      "loss": 1.7926,
+      "step": 634
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019977119813882426,
+      "loss": 1.6464,
+      "step": 635
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997704537478218,
+      "loss": 1.7884,
+      "step": 636
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997697081492643,
+      "loss": 1.6688,
+      "step": 637
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019976896134316072,
+      "loss": 1.6837,
+      "step": 638
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019976821332952013,
+      "loss": 1.7799,
+      "step": 639
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997674641083516,
+      "loss": 1.6806,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997667136796642,
+      "loss": 1.7332,
+      "step": 641
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000199765962043467,
+      "loss": 1.7577,
+      "step": 642
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019976520919976908,
+      "loss": 1.7549,
+      "step": 643
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019976445514857956,
+      "loss": 1.6703,
+      "step": 644
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997636998899076,
+      "loss": 1.6002,
+      "step": 645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019976294342376226,
+      "loss": 1.6214,
+      "step": 646
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997621857501528,
+      "loss": 1.5355,
+      "step": 647
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019976142686908832,
+      "loss": 1.5834,
+      "step": 648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019976066678057804,
+      "loss": 1.6887,
+      "step": 649
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019975990548463113,
+      "loss": 1.7532,
+      "step": 650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997591429812568,
+      "loss": 1.5203,
+      "step": 651
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019975837927046434,
+      "loss": 1.6618,
+      "step": 652
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019975761435226292,
+      "loss": 1.6847,
+      "step": 653
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019975684822666186,
+      "loss": 1.6859,
+      "step": 654
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019975608089367036,
+      "loss": 1.6125,
+      "step": 655
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019975531235329777,
+      "loss": 1.5429,
+      "step": 656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019975454260555338,
+      "loss": 1.6258,
+      "step": 657
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997537716504465,
+      "loss": 1.7018,
+      "step": 658
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019975299948798644,
+      "loss": 1.6068,
+      "step": 659
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019975222611818256,
+      "loss": 1.5509,
+      "step": 660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019975145154104427,
+      "loss": 1.6134,
+      "step": 661
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019975067575658085,
+      "loss": 1.7913,
+      "step": 662
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019974989876480175,
+      "loss": 1.622,
+      "step": 663
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997491205657164,
+      "loss": 1.7323,
+      "step": 664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019974834115933415,
+      "loss": 1.6377,
+      "step": 665
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019974756054566445,
+      "loss": 1.6395,
+      "step": 666
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019974677872471678,
+      "loss": 1.5828,
+      "step": 667
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997459956965006,
+      "loss": 1.7056,
+      "step": 668
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019974521146102537,
+      "loss": 1.5637,
+      "step": 669
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019974442601830057,
+      "loss": 1.5592,
+      "step": 670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019974363936833573,
+      "loss": 1.6444,
+      "step": 671
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019974285151114037,
+      "loss": 1.7274,
+      "step": 672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019974206244672402,
+      "loss": 1.5493,
+      "step": 673
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997412721750962,
+      "loss": 1.7595,
+      "step": 674
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019974048069626655,
+      "loss": 1.7434,
+      "step": 675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019973968801024458,
+      "loss": 1.5868,
+      "step": 676
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997388941170399,
+      "loss": 1.589,
+      "step": 677
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019973809901666214,
+      "loss": 1.6291,
+      "step": 678
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997373027091209,
+      "loss": 1.7207,
+      "step": 679
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019973650519442584,
+      "loss": 1.6464,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997357064725866,
+      "loss": 1.5749,
+      "step": 681
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019973490654361285,
+      "loss": 1.6073,
+      "step": 682
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019973410540751426,
+      "loss": 1.7443,
+      "step": 683
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019973330306430052,
+      "loss": 1.5342,
+      "step": 684
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019973249951398136,
+      "loss": 1.5949,
+      "step": 685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019973169475656652,
+      "loss": 1.5546,
+      "step": 686
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019973088879206573,
+      "loss": 1.5824,
+      "step": 687
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019973008162048873,
+      "loss": 1.5585,
+      "step": 688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997292732418453,
+      "loss": 1.6072,
+      "step": 689
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019972846365614523,
+      "loss": 1.5591,
+      "step": 690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019972765286339826,
+      "loss": 1.7423,
+      "step": 691
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997268408636143,
+      "loss": 1.5292,
+      "step": 692
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019972602765680312,
+      "loss": 1.4959,
+      "step": 693
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019972521324297456,
+      "loss": 1.6579,
+      "step": 694
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019972439762213853,
+      "loss": 1.5467,
+      "step": 695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019972358079430483,
+      "loss": 1.8022,
+      "step": 696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997227627594834,
+      "loss": 1.6216,
+      "step": 697
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019972194351768407,
+      "loss": 1.7475,
+      "step": 698
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019972112306891685,
+      "loss": 1.4679,
+      "step": 699
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019972030141319163,
+      "loss": 1.6484,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019971947855051833,
+      "loss": 1.6816,
+      "step": 701
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019971865448090693,
+      "loss": 1.717,
+      "step": 702
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001997178292043674,
+      "loss": 1.5418,
+      "step": 703
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019971700272090976,
+      "loss": 1.563,
+      "step": 704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019971617503054397,
+      "loss": 1.7972,
+      "step": 705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019971534613328005,
+      "loss": 1.5376,
+      "step": 706
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019971451602912804,
+      "loss": 1.6266,
+      "step": 707
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019971368471809805,
+      "loss": 1.6289,
+      "step": 708
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019971285220020003,
+      "loss": 1.695,
+      "step": 709
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019971201847544413,
+      "loss": 1.7424,
+      "step": 710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019971118354384043,
+      "loss": 1.5404,
+      "step": 711
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000199710347405399,
+      "loss": 1.5671,
+      "step": 712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019970951006013,
+      "loss": 1.6633,
+      "step": 713
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019970867150804353,
+      "loss": 1.5156,
+      "step": 714
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019970783174914978,
+      "loss": 1.6924,
+      "step": 715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019970699078345893,
+      "loss": 1.6776,
+      "step": 716
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019970614861098107,
+      "loss": 1.7459,
+      "step": 717
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019970530523172645,
+      "loss": 1.6395,
+      "step": 718
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001997044606457053,
+      "loss": 1.5646,
+      "step": 719
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001997036148529278,
+      "loss": 1.5496,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001997027678534042,
+      "loss": 1.6403,
+      "step": 721
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019970191964714478,
+      "loss": 1.682,
+      "step": 722
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019970107023415976,
+      "loss": 1.7485,
+      "step": 723
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019970021961445946,
+      "loss": 1.7541,
+      "step": 724
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019969936778805415,
+      "loss": 1.467,
+      "step": 725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996985147549541,
+      "loss": 1.5398,
+      "step": 726
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019969766051516975,
+      "loss": 1.7044,
+      "step": 727
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019969680506871137,
+      "loss": 1.6739,
+      "step": 728
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996959484155893,
+      "loss": 1.5621,
+      "step": 729
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019969509055581388,
+      "loss": 1.4948,
+      "step": 730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019969423148939557,
+      "loss": 1.5404,
+      "step": 731
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019969337121634472,
+      "loss": 1.6805,
+      "step": 732
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019969250973667175,
+      "loss": 1.5032,
+      "step": 733
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019969164705038712,
+      "loss": 1.6408,
+      "step": 734
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019969078315750123,
+      "loss": 1.6475,
+      "step": 735
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019968991805802456,
+      "loss": 1.5631,
+      "step": 736
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019968905175196756,
+      "loss": 1.782,
+      "step": 737
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019968818423934072,
+      "loss": 1.6239,
+      "step": 738
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019968731552015453,
+      "loss": 1.4891,
+      "step": 739
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019968644559441954,
+      "loss": 1.5504,
+      "step": 740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019968557446214627,
+      "loss": 1.5642,
+      "step": 741
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019968470212334522,
+      "loss": 1.5692,
+      "step": 742
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019968382857802702,
+      "loss": 1.5969,
+      "step": 743
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019968295382620217,
+      "loss": 1.657,
+      "step": 744
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019968207786788132,
+      "loss": 1.66,
+      "step": 745
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199681200703075,
+      "loss": 1.6166,
+      "step": 746
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019968032233179392,
+      "loss": 1.6924,
+      "step": 747
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019967944275404864,
+      "loss": 1.55,
+      "step": 748
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019967856196984981,
+      "loss": 1.7634,
+      "step": 749
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019967767997920813,
+      "loss": 1.652,
+      "step": 750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019967679678213425,
+      "loss": 1.5127,
+      "step": 751
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019967591237863883,
+      "loss": 1.586,
+      "step": 752
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019967502676873265,
+      "loss": 1.6856,
+      "step": 753
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019967413995242636,
+      "loss": 1.6432,
+      "step": 754
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019967325192973072,
+      "loss": 1.6175,
+      "step": 755
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019967236270065648,
+      "loss": 1.5103,
+      "step": 756
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019967147226521438,
+      "loss": 1.6074,
+      "step": 757
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019967058062341525,
+      "loss": 1.6773,
+      "step": 758
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019966968777526982,
+      "loss": 1.7853,
+      "step": 759
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019966879372078892,
+      "loss": 1.6245,
+      "step": 760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019966789845998338,
+      "loss": 1.63,
+      "step": 761
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019966700199286407,
+      "loss": 1.5437,
+      "step": 762
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019966610431944174,
+      "loss": 1.6225,
+      "step": 763
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019966520543972734,
+      "loss": 1.5221,
+      "step": 764
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019966430535373175,
+      "loss": 1.7629,
+      "step": 765
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996634040614658,
+      "loss": 1.5211,
+      "step": 766
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019966250156294049,
+      "loss": 1.5929,
+      "step": 767
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019966159785816663,
+      "loss": 1.723,
+      "step": 768
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019966069294715525,
+      "loss": 1.564,
+      "step": 769
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019965978682991725,
+      "loss": 1.6451,
+      "step": 770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019965887950646363,
+      "loss": 1.4646,
+      "step": 771
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996579709768054,
+      "loss": 1.5952,
+      "step": 772
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996570612409535,
+      "loss": 1.6318,
+      "step": 773
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019965615029891894,
+      "loss": 1.6843,
+      "step": 774
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019965523815071274,
+      "loss": 1.654,
+      "step": 775
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199654324796346,
+      "loss": 1.6399,
+      "step": 776
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996534102358297,
+      "loss": 1.5391,
+      "step": 777
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019965249446917497,
+      "loss": 1.6739,
+      "step": 778
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019965157749639287,
+      "loss": 1.6331,
+      "step": 779
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996506593174945,
+      "loss": 1.6259,
+      "step": 780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199649739932491,
+      "loss": 1.4002,
+      "step": 781
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019964881934139342,
+      "loss": 1.425,
+      "step": 782
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019964789754421295,
+      "loss": 1.7722,
+      "step": 783
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996469745409608,
+      "loss": 1.5814,
+      "step": 784
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019964605033164805,
+      "loss": 1.7721,
+      "step": 785
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019964512491628592,
+      "loss": 1.571,
+      "step": 786
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019964419829488565,
+      "loss": 1.5564,
+      "step": 787
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996432704674584,
+      "loss": 1.5317,
+      "step": 788
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019964234143401543,
+      "loss": 1.3829,
+      "step": 789
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019964141119456797,
+      "loss": 1.6002,
+      "step": 790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019964047974912729,
+      "loss": 1.5955,
+      "step": 791
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019963954709770464,
+      "loss": 1.6641,
+      "step": 792
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019963861324031134,
+      "loss": 1.3779,
+      "step": 793
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996376781769587,
+      "loss": 1.4325,
+      "step": 794
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199636741907658,
+      "loss": 1.6911,
+      "step": 795
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019963580443242056,
+      "loss": 1.6491,
+      "step": 796
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996348657512578,
+      "loss": 1.6682,
+      "step": 797
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199633925864181,
+      "loss": 1.4742,
+      "step": 798
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996329847712016,
+      "loss": 1.6766,
+      "step": 799
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019963204247233095,
+      "loss": 1.5756,
+      "step": 800
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 1.5861952304840088,
+      "eval_runtime": 843.9394,
+      "eval_samples_per_second": 28.509,
+      "eval_steps_per_second": 14.255,
+      "step": 800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019963109896758047,
+      "loss": 1.6516,
+      "step": 801
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996301542569616,
+      "loss": 1.59,
+      "step": 802
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019962920834048573,
+      "loss": 1.6031,
+      "step": 803
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019962826121816432,
+      "loss": 1.5271,
+      "step": 804
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019962731289000884,
+      "loss": 1.732,
+      "step": 805
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019962636335603079,
+      "loss": 1.5412,
+      "step": 806
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019962541261624164,
+      "loss": 1.6285,
+      "step": 807
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996244606706529,
+      "loss": 1.5039,
+      "step": 808
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996235075192761,
+      "loss": 1.6525,
+      "step": 809
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019962255316212274,
+      "loss": 1.4973,
+      "step": 810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019962159759920445,
+      "loss": 1.5777,
+      "step": 811
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996206408305327,
+      "loss": 1.5993,
+      "step": 812
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019961968285611912,
+      "loss": 1.6446,
+      "step": 813
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996187236759753,
+      "loss": 1.6184,
+      "step": 814
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996177632901129,
+      "loss": 1.4846,
+      "step": 815
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019961680169854344,
+      "loss": 1.4641,
+      "step": 816
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019961583890127865,
+      "loss": 1.6112,
+      "step": 817
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996148748983301,
+      "loss": 1.5912,
+      "step": 818
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019961390968970953,
+      "loss": 1.6099,
+      "step": 819
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019961294327542858,
+      "loss": 1.5493,
+      "step": 820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019961197565549897,
+      "loss": 1.6337,
+      "step": 821
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019961100682993238,
+      "loss": 1.4759,
+      "step": 822
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019961003679874062,
+      "loss": 1.5404,
+      "step": 823
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019960906556193534,
+      "loss": 1.6279,
+      "step": 824
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019960809311952827,
+      "loss": 1.514,
+      "step": 825
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996071194715313,
+      "loss": 1.4677,
+      "step": 826
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019960614461795616,
+      "loss": 1.6278,
+      "step": 827
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019960516855881458,
+      "loss": 1.4631,
+      "step": 828
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019960419129411847,
+      "loss": 1.5898,
+      "step": 829
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001996032128238796,
+      "loss": 1.5253,
+      "step": 830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019960223314810987,
+      "loss": 1.5117,
+      "step": 831
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019960125226682105,
+      "loss": 1.7111,
+      "step": 832
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019960027018002508,
+      "loss": 1.5224,
+      "step": 833
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019959928688773385,
+      "loss": 1.4446,
+      "step": 834
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995983023899592,
+      "loss": 1.6854,
+      "step": 835
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995973166867131,
+      "loss": 1.6591,
+      "step": 836
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019959632977800752,
+      "loss": 1.4695,
+      "step": 837
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019959534166385427,
+      "loss": 1.6047,
+      "step": 838
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019959435234426543,
+      "loss": 1.633,
+      "step": 839
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019959336181925292,
+      "loss": 1.7428,
+      "step": 840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019959237008882873,
+      "loss": 1.6543,
+      "step": 841
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019959137715300492,
+      "loss": 1.6832,
+      "step": 842
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019959038301179343,
+      "loss": 1.4911,
+      "step": 843
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019958938766520635,
+      "loss": 1.6603,
+      "step": 844
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019958839111325566,
+      "loss": 1.6137,
+      "step": 845
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995873933559535,
+      "loss": 1.5178,
+      "step": 846
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019958639439331192,
+      "loss": 1.4348,
+      "step": 847
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199585394225343,
+      "loss": 1.6197,
+      "step": 848
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019958439285205882,
+      "loss": 1.5359,
+      "step": 849
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019958339027347153,
+      "loss": 1.6809,
+      "step": 850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019958238648959327,
+      "loss": 1.6029,
+      "step": 851
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995813815004362,
+      "loss": 1.4948,
+      "step": 852
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995803753060124,
+      "loss": 1.4977,
+      "step": 853
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995793679063342,
+      "loss": 1.5217,
+      "step": 854
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019957835930141363,
+      "loss": 1.5443,
+      "step": 855
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019957734949126304,
+      "loss": 1.6297,
+      "step": 856
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019957633847589453,
+      "loss": 1.6591,
+      "step": 857
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019957532625532043,
+      "loss": 1.6353,
+      "step": 858
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019957431282955292,
+      "loss": 1.6561,
+      "step": 859
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995732981986043,
+      "loss": 1.5058,
+      "step": 860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019957228236248687,
+      "loss": 1.6526,
+      "step": 861
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019957126532121292,
+      "loss": 1.6292,
+      "step": 862
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995702470747947,
+      "loss": 1.5383,
+      "step": 863
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019956922762324458,
+      "loss": 1.6926,
+      "step": 864
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019956820696657494,
+      "loss": 1.5685,
+      "step": 865
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019956718510479808,
+      "loss": 1.5272,
+      "step": 866
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019956616203792635,
+      "loss": 1.589,
+      "step": 867
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019956513776597218,
+      "loss": 1.66,
+      "step": 868
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019956411228894794,
+      "loss": 1.4801,
+      "step": 869
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019956308560686606,
+      "loss": 1.4821,
+      "step": 870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019956205771973893,
+      "loss": 1.6331,
+      "step": 871
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019956102862757905,
+      "loss": 1.5814,
+      "step": 872
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995599983303988,
+      "loss": 1.4973,
+      "step": 873
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019955896682821074,
+      "loss": 1.6737,
+      "step": 874
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019955793412102727,
+      "loss": 1.6168,
+      "step": 875
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019955690020886094,
+      "loss": 1.6751,
+      "step": 876
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019955586509172427,
+      "loss": 1.3353,
+      "step": 877
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019955482876962975,
+      "loss": 1.7037,
+      "step": 878
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019955379124258994,
+      "loss": 1.5656,
+      "step": 879
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019955275251061738,
+      "loss": 1.5876,
+      "step": 880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995517125737247,
+      "loss": 1.5621,
+      "step": 881
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995506714319244,
+      "loss": 1.4952,
+      "step": 882
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019954962908522915,
+      "loss": 1.3719,
+      "step": 883
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019954858553365153,
+      "loss": 1.5169,
+      "step": 884
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019954754077720425,
+      "loss": 1.4917,
+      "step": 885
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019954649481589983,
+      "loss": 1.552,
+      "step": 886
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019954544764975098,
+      "loss": 1.5217,
+      "step": 887
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995443992787704,
+      "loss": 1.4747,
+      "step": 888
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019954334970297076,
+      "loss": 1.5358,
+      "step": 889
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995422989223648,
+      "loss": 1.5091,
+      "step": 890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019954124693696517,
+      "loss": 1.6436,
+      "step": 891
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019954019374678462,
+      "loss": 1.6061,
+      "step": 892
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019953913935183594,
+      "loss": 1.5423,
+      "step": 893
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019953808375213188,
+      "loss": 1.5426,
+      "step": 894
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995370269476852,
+      "loss": 1.4836,
+      "step": 895
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019953596893850864,
+      "loss": 1.533,
+      "step": 896
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995349097246151,
+      "loss": 1.5278,
+      "step": 897
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019953384930601736,
+      "loss": 1.7081,
+      "step": 898
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019953278768272825,
+      "loss": 1.5378,
+      "step": 899
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995317248547606,
+      "loss": 1.5865,
+      "step": 900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019953066082212733,
+      "loss": 1.4203,
+      "step": 901
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019952959558484126,
+      "loss": 1.6331,
+      "step": 902
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019952852914291528,
+      "loss": 1.6955,
+      "step": 903
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019952746149636238,
+      "loss": 1.4203,
+      "step": 904
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019952639264519538,
+      "loss": 1.5797,
+      "step": 905
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019952532258942732,
+      "loss": 1.5074,
+      "step": 906
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019952425132907102,
+      "loss": 1.7455,
+      "step": 907
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019952317886413957,
+      "loss": 1.4792,
+      "step": 908
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995221051946459,
+      "loss": 1.4953,
+      "step": 909
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199521030320603,
+      "loss": 1.4619,
+      "step": 910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019951995424202386,
+      "loss": 1.6254,
+      "step": 911
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019951887695892153,
+      "loss": 1.5158,
+      "step": 912
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019951779847130906,
+      "loss": 1.4869,
+      "step": 913
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995167187791995,
+      "loss": 1.5295,
+      "step": 914
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995156378826059,
+      "loss": 1.6046,
+      "step": 915
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019951455578154137,
+      "loss": 1.5768,
+      "step": 916
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019951347247601896,
+      "loss": 1.5046,
+      "step": 917
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019951238796605183,
+      "loss": 1.5135,
+      "step": 918
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019951130225165308,
+      "loss": 1.4209,
+      "step": 919
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019951021533283584,
+      "loss": 1.5117,
+      "step": 920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019950912720961332,
+      "loss": 1.4965,
+      "step": 921
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019950803788199863,
+      "loss": 1.4784,
+      "step": 922
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019950694735000496,
+      "loss": 1.4269,
+      "step": 923
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019950585561364555,
+      "loss": 1.4997,
+      "step": 924
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019950476267293358,
+      "loss": 1.5666,
+      "step": 925
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995036685278823,
+      "loss": 1.6009,
+      "step": 926
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001995025731785049,
+      "loss": 1.6578,
+      "step": 927
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019950147662481472,
+      "loss": 1.5251,
+      "step": 928
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019950037886682498,
+      "loss": 1.5513,
+      "step": 929
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019949927990454898,
+      "loss": 1.5817,
+      "step": 930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199498179738,
+      "loss": 1.6309,
+      "step": 931
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019949707836719138,
+      "loss": 1.5056,
+      "step": 932
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019949597579213646,
+      "loss": 1.6148,
+      "step": 933
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019949487201284852,
+      "loss": 1.4986,
+      "step": 934
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019949376702934103,
+      "loss": 1.3856,
+      "step": 935
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019949266084162725,
+      "loss": 1.5133,
+      "step": 936
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019949155344972064,
+      "loss": 1.5612,
+      "step": 937
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001994904448536346,
+      "loss": 1.5998,
+      "step": 938
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019948933505338249,
+      "loss": 1.4764,
+      "step": 939
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019948822404897785,
+      "loss": 1.6608,
+      "step": 940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199487111840434,
+      "loss": 1.4174,
+      "step": 941
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001994859984277645,
+      "loss": 1.5292,
+      "step": 942
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019948488381098275,
+      "loss": 1.5941,
+      "step": 943
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001994837679901023,
+      "loss": 1.4507,
+      "step": 944
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019948265096513662,
+      "loss": 1.545,
+      "step": 945
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019948153273609926,
+      "loss": 1.46,
+      "step": 946
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019948041330300373,
+      "loss": 1.6386,
+      "step": 947
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001994792926658636,
+      "loss": 1.4438,
+      "step": 948
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001994781708246924,
+      "loss": 1.459,
+      "step": 949
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019947704777950373,
+      "loss": 1.5635,
+      "step": 950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019947592353031118,
+      "loss": 1.458,
+      "step": 951
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019947479807712836,
+      "loss": 1.4895,
+      "step": 952
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001994736714199689,
+      "loss": 1.6264,
+      "step": 953
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001994725435588464,
+      "loss": 1.4972,
+      "step": 954
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019947141449377456,
+      "loss": 1.326,
+      "step": 955
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199470284224767,
+      "loss": 1.383,
+      "step": 956
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019946915275183743,
+      "loss": 1.5017,
+      "step": 957
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019946802007499956,
+      "loss": 1.6048,
+      "step": 958
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019946688619426706,
+      "loss": 1.4516,
+      "step": 959
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019946575110965363,
+      "loss": 1.5359,
+      "step": 960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019946461482117308,
+      "loss": 1.4337,
+      "step": 961
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019946347732883917,
+      "loss": 1.6624,
+      "step": 962
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019946233863266558,
+      "loss": 1.4409,
+      "step": 963
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019946119873266613,
+      "loss": 1.5586,
+      "step": 964
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019946005762885464,
+      "loss": 1.4583,
+      "step": 965
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019945891532124495,
+      "loss": 1.4828,
+      "step": 966
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019945777180985078,
+      "loss": 1.57,
+      "step": 967
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019945662709468608,
+      "loss": 1.4665,
+      "step": 968
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019945548117576465,
+      "loss": 1.4811,
+      "step": 969
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019945433405310034,
+      "loss": 1.3989,
+      "step": 970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001994531857267071,
+      "loss": 1.6084,
+      "step": 971
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019945203619659875,
+      "loss": 1.3859,
+      "step": 972
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019945088546278927,
+      "loss": 1.5504,
+      "step": 973
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019944973352529254,
+      "loss": 1.5543,
+      "step": 974
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019944858038412256,
+      "loss": 1.5585,
+      "step": 975
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001994474260392932,
+      "loss": 1.4154,
+      "step": 976
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019944627049081853,
+      "loss": 1.5861,
+      "step": 977
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019944511373871246,
+      "loss": 1.5819,
+      "step": 978
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019944395578298902,
+      "loss": 1.6049,
+      "step": 979
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019944279662366222,
+      "loss": 1.4866,
+      "step": 980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019944163626074612,
+      "loss": 1.5212,
+      "step": 981
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019944047469425474,
+      "loss": 1.6071,
+      "step": 982
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001994393119242021,
+      "loss": 1.527,
+      "step": 983
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019943814795060234,
+      "loss": 1.4864,
+      "step": 984
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001994369827734695,
+      "loss": 1.454,
+      "step": 985
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001994358163928177,
+      "loss": 1.539,
+      "step": 986
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019943464880866108,
+      "loss": 1.6056,
+      "step": 987
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019943348002101371,
+      "loss": 1.5323,
+      "step": 988
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019943231002988982,
+      "loss": 1.4323,
+      "step": 989
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001994311388353035,
+      "loss": 1.4786,
+      "step": 990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019942996643726893,
+      "loss": 1.4925,
+      "step": 991
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019942879283580035,
+      "loss": 1.5982,
+      "step": 992
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019942761803091194,
+      "loss": 1.5688,
+      "step": 993
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019942644202261791,
+      "loss": 1.6566,
+      "step": 994
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001994252648109325,
+      "loss": 1.571,
+      "step": 995
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019942408639586994,
+      "loss": 1.5055,
+      "step": 996
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019942290677744452,
+      "loss": 1.4047,
+      "step": 997
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001994217259556705,
+      "loss": 1.4665,
+      "step": 998
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019942054393056222,
+      "loss": 1.4866,
+      "step": 999
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001994193607021339,
+      "loss": 1.4847,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 1.5133675336837769,
+      "eval_runtime": 846.352,
+      "eval_samples_per_second": 28.428,
+      "eval_steps_per_second": 14.214,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019941817627039992,
+      "loss": 1.623,
+      "step": 1001
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019941699063537457,
+      "loss": 1.4983,
+      "step": 1002
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019941580379707226,
+      "loss": 1.5352,
+      "step": 1003
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019941461575550735,
+      "loss": 1.4337,
+      "step": 1004
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019941342651069418,
+      "loss": 1.4327,
+      "step": 1005
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019941223606264715,
+      "loss": 1.4835,
+      "step": 1006
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019941104441138067,
+      "loss": 1.4571,
+      "step": 1007
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001994098515569092,
+      "loss": 1.5587,
+      "step": 1008
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019940865749924714,
+      "loss": 1.6487,
+      "step": 1009
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019940746223840892,
+      "loss": 1.4038,
+      "step": 1010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019940626577440905,
+      "loss": 1.5395,
+      "step": 1011
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199405068107262,
+      "loss": 1.5219,
+      "step": 1012
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019940386923698227,
+      "loss": 1.6292,
+      "step": 1013
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019940266916358438,
+      "loss": 1.4952,
+      "step": 1014
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001994014678870828,
+      "loss": 1.5659,
+      "step": 1015
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019940026540749215,
+      "loss": 1.593,
+      "step": 1016
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019939906172482692,
+      "loss": 1.5823,
+      "step": 1017
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019939785683910173,
+      "loss": 1.4469,
+      "step": 1018
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993966507503311,
+      "loss": 1.7027,
+      "step": 1019
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019939544345852966,
+      "loss": 1.5539,
+      "step": 1020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019939423496371204,
+      "loss": 1.5161,
+      "step": 1021
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019939302526589286,
+      "loss": 1.3582,
+      "step": 1022
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019939181436508672,
+      "loss": 1.4165,
+      "step": 1023
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019939060226130834,
+      "loss": 1.361,
+      "step": 1024
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019938938895457236,
+      "loss": 1.5114,
+      "step": 1025
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019938817444489345,
+      "loss": 1.5797,
+      "step": 1026
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019938695873228631,
+      "loss": 1.6674,
+      "step": 1027
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019938574181676568,
+      "loss": 1.4633,
+      "step": 1028
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993845236983463,
+      "loss": 1.5057,
+      "step": 1029
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019938330437704286,
+      "loss": 1.5081,
+      "step": 1030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019938208385287016,
+      "loss": 1.4946,
+      "step": 1031
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019938086212584297,
+      "loss": 1.4018,
+      "step": 1032
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019937963919597603,
+      "loss": 1.6042,
+      "step": 1033
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019937841506328423,
+      "loss": 1.473,
+      "step": 1034
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993771897277823,
+      "loss": 1.5224,
+      "step": 1035
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993759631894851,
+      "loss": 1.6868,
+      "step": 1036
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993747354484075,
+      "loss": 1.628,
+      "step": 1037
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019937350650456434,
+      "loss": 1.6157,
+      "step": 1038
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019937227635797049,
+      "loss": 1.397,
+      "step": 1039
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019937104500864086,
+      "loss": 1.6419,
+      "step": 1040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019936981245659033,
+      "loss": 1.5574,
+      "step": 1041
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019936857870183377,
+      "loss": 1.3661,
+      "step": 1042
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019936734374438624,
+      "loss": 1.5328,
+      "step": 1043
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993661075842626,
+      "loss": 1.4169,
+      "step": 1044
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993648702214778,
+      "loss": 1.4808,
+      "step": 1045
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019936363165604687,
+      "loss": 1.4846,
+      "step": 1046
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019936239188798476,
+      "loss": 1.3835,
+      "step": 1047
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019936115091730646,
+      "loss": 1.3343,
+      "step": 1048
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019935990874402708,
+      "loss": 1.6892,
+      "step": 1049
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019935866536816157,
+      "loss": 1.6302,
+      "step": 1050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199357420789725,
+      "loss": 1.498,
+      "step": 1051
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993561750087324,
+      "loss": 1.5167,
+      "step": 1052
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993549280251989,
+      "loss": 1.4331,
+      "step": 1053
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993536798391396,
+      "loss": 1.5392,
+      "step": 1054
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019935243045056958,
+      "loss": 1.3856,
+      "step": 1055
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019935117985950393,
+      "loss": 1.4696,
+      "step": 1056
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019934992806595786,
+      "loss": 1.4681,
+      "step": 1057
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019934867506994647,
+      "loss": 1.5876,
+      "step": 1058
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019934742087148493,
+      "loss": 1.6119,
+      "step": 1059
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019934616547058847,
+      "loss": 1.3547,
+      "step": 1060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019934490886727217,
+      "loss": 1.5636,
+      "step": 1061
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019934365106155134,
+      "loss": 1.4305,
+      "step": 1062
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993423920534412,
+      "loss": 1.4853,
+      "step": 1063
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019934113184295695,
+      "loss": 1.6535,
+      "step": 1064
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019933987043011387,
+      "loss": 1.5058,
+      "step": 1065
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019933860781492718,
+      "loss": 1.4007,
+      "step": 1066
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993373439974122,
+      "loss": 1.5405,
+      "step": 1067
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019933607897758426,
+      "loss": 1.6341,
+      "step": 1068
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993348127554586,
+      "loss": 1.5,
+      "step": 1069
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019933354533105058,
+      "loss": 1.4697,
+      "step": 1070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019933227670437557,
+      "loss": 1.4753,
+      "step": 1071
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019933100687544885,
+      "loss": 1.5009,
+      "step": 1072
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019932973584428583,
+      "loss": 1.5291,
+      "step": 1073
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019932846361090195,
+      "loss": 1.4607,
+      "step": 1074
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993271901753125,
+      "loss": 1.3716,
+      "step": 1075
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019932591553753295,
+      "loss": 1.4985,
+      "step": 1076
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019932463969757875,
+      "loss": 1.4419,
+      "step": 1077
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019932336265546528,
+      "loss": 1.5199,
+      "step": 1078
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019932208441120807,
+      "loss": 1.5546,
+      "step": 1079
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019932080496482254,
+      "loss": 1.4043,
+      "step": 1080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019931952431632418,
+      "loss": 1.5538,
+      "step": 1081
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993182424657285,
+      "loss": 1.4595,
+      "step": 1082
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199316959413051,
+      "loss": 1.4617,
+      "step": 1083
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019931567515830726,
+      "loss": 1.4758,
+      "step": 1084
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019931438970151278,
+      "loss": 1.5543,
+      "step": 1085
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019931310304268314,
+      "loss": 1.6431,
+      "step": 1086
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019931181518183385,
+      "loss": 1.5661,
+      "step": 1087
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993105261189806,
+      "loss": 1.5188,
+      "step": 1088
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993092358541389,
+      "loss": 1.5116,
+      "step": 1089
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019930794438732439,
+      "loss": 1.5656,
+      "step": 1090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019930665171855275,
+      "loss": 1.5307,
+      "step": 1091
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019930535784783957,
+      "loss": 1.4509,
+      "step": 1092
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019930406277520054,
+      "loss": 1.6256,
+      "step": 1093
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993027665006513,
+      "loss": 1.6138,
+      "step": 1094
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993014690242076,
+      "loss": 1.4891,
+      "step": 1095
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001993001703458851,
+      "loss": 1.5439,
+      "step": 1096
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992988704656995,
+      "loss": 1.4691,
+      "step": 1097
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992975693836666,
+      "loss": 1.5182,
+      "step": 1098
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019929626709980206,
+      "loss": 1.6048,
+      "step": 1099
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019929496361412174,
+      "loss": 1.4678,
+      "step": 1100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019929365892664132,
+      "loss": 1.5983,
+      "step": 1101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019929235303737668,
+      "loss": 1.4703,
+      "step": 1102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019929104594634356,
+      "loss": 1.6134,
+      "step": 1103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992897376535578,
+      "loss": 1.6239,
+      "step": 1104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019928842815903524,
+      "loss": 1.3665,
+      "step": 1105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019928711746279172,
+      "loss": 1.4323,
+      "step": 1106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019928580556484313,
+      "loss": 1.3978,
+      "step": 1107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019928449246520532,
+      "loss": 1.5138,
+      "step": 1108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019928317816389417,
+      "loss": 1.6035,
+      "step": 1109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019928186266092562,
+      "loss": 1.5351,
+      "step": 1110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992805459563156,
+      "loss": 1.5351,
+      "step": 1111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019927922805008,
+      "loss": 1.3946,
+      "step": 1112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992779089422348,
+      "loss": 1.5221,
+      "step": 1113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019927658863279598,
+      "loss": 1.5685,
+      "step": 1114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992752671217795,
+      "loss": 1.468,
+      "step": 1115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019927394440920137,
+      "loss": 1.4842,
+      "step": 1116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019927262049507759,
+      "loss": 1.5845,
+      "step": 1117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019927129537942415,
+      "loss": 1.444,
+      "step": 1118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019926996906225713,
+      "loss": 1.5147,
+      "step": 1119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019926864154359256,
+      "loss": 1.541,
+      "step": 1120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019926731282344658,
+      "loss": 1.5207,
+      "step": 1121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019926598290183516,
+      "loss": 1.5316,
+      "step": 1122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019926465177877445,
+      "loss": 1.6307,
+      "step": 1123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019926331945428055,
+      "loss": 1.3729,
+      "step": 1124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019926198592836965,
+      "loss": 1.3465,
+      "step": 1125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992606512010578,
+      "loss": 1.5754,
+      "step": 1126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992593152723612,
+      "loss": 1.4224,
+      "step": 1127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199257978142296,
+      "loss": 1.5289,
+      "step": 1128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019925663981087838,
+      "loss": 1.5222,
+      "step": 1129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019925530027812457,
+      "loss": 1.4756,
+      "step": 1130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019925395954405075,
+      "loss": 1.4618,
+      "step": 1131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019925261760867318,
+      "loss": 1.3161,
+      "step": 1132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019925127447200806,
+      "loss": 1.3612,
+      "step": 1133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019924993013407168,
+      "loss": 1.5567,
+      "step": 1134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992485845948803,
+      "loss": 1.5223,
+      "step": 1135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019924723785445022,
+      "loss": 1.5227,
+      "step": 1136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019924588991279772,
+      "loss": 1.3888,
+      "step": 1137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019924454076993911,
+      "loss": 1.5172,
+      "step": 1138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019924319042589076,
+      "loss": 1.4495,
+      "step": 1139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019924183888066894,
+      "loss": 1.5854,
+      "step": 1140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019924048613429007,
+      "loss": 1.618,
+      "step": 1141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019923913218677052,
+      "loss": 1.4444,
+      "step": 1142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019923777703812664,
+      "loss": 1.5503,
+      "step": 1143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019923642068837486,
+      "loss": 1.3723,
+      "step": 1144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992350631375316,
+      "loss": 1.4895,
+      "step": 1145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019923370438561325,
+      "loss": 1.3311,
+      "step": 1146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992323444326363,
+      "loss": 1.5726,
+      "step": 1147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019923098327861725,
+      "loss": 1.5913,
+      "step": 1148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019922962092357247,
+      "loss": 1.5769,
+      "step": 1149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019922825736751852,
+      "loss": 1.3966,
+      "step": 1150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019922689261047186,
+      "loss": 1.4448,
+      "step": 1151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019922552665244903,
+      "loss": 1.3372,
+      "step": 1152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019922415949346657,
+      "loss": 1.4139,
+      "step": 1153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019922279113354104,
+      "loss": 1.3598,
+      "step": 1154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019922142157268897,
+      "loss": 1.4993,
+      "step": 1155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019922005081092695,
+      "loss": 1.427,
+      "step": 1156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019921867884827157,
+      "loss": 1.5393,
+      "step": 1157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019921730568473944,
+      "loss": 1.4199,
+      "step": 1158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019921593132034716,
+      "loss": 1.4398,
+      "step": 1159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992145557551114,
+      "loss": 1.4301,
+      "step": 1160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992131789890488,
+      "loss": 1.311,
+      "step": 1161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019921180102217597,
+      "loss": 1.4192,
+      "step": 1162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019921042185450967,
+      "loss": 1.5799,
+      "step": 1163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992090414860665,
+      "loss": 1.5596,
+      "step": 1164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019920765991686325,
+      "loss": 1.5006,
+      "step": 1165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992062771469166,
+      "loss": 1.4329,
+      "step": 1166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019920489317624336,
+      "loss": 1.4568,
+      "step": 1167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019920350800486017,
+      "loss": 1.5455,
+      "step": 1168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001992021216327838,
+      "loss": 1.6157,
+      "step": 1169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019920073406003115,
+      "loss": 1.5078,
+      "step": 1170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991993452866189,
+      "loss": 1.5561,
+      "step": 1171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991979553125639,
+      "loss": 1.5546,
+      "step": 1172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019919656413788296,
+      "loss": 1.4341,
+      "step": 1173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019919517176259294,
+      "loss": 1.396,
+      "step": 1174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019919377818671067,
+      "loss": 1.4208,
+      "step": 1175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019919238341025306,
+      "loss": 1.532,
+      "step": 1176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991909874332369,
+      "loss": 1.3355,
+      "step": 1177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991895902556792,
+      "loss": 1.5847,
+      "step": 1178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019918819187759676,
+      "loss": 1.4564,
+      "step": 1179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019918679229900658,
+      "loss": 1.5137,
+      "step": 1180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991853915199256,
+      "loss": 1.4251,
+      "step": 1181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019918398954037075,
+      "loss": 1.4187,
+      "step": 1182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199182586360359,
+      "loss": 1.491,
+      "step": 1183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019918118197990732,
+      "loss": 1.3919,
+      "step": 1184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019917977639903272,
+      "loss": 1.3672,
+      "step": 1185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019917836961775223,
+      "loss": 1.2956,
+      "step": 1186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019917696163608285,
+      "loss": 1.6111,
+      "step": 1187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019917555245404164,
+      "loss": 1.51,
+      "step": 1188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019917414207164564,
+      "loss": 1.4599,
+      "step": 1189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019917273048891194,
+      "loss": 1.5088,
+      "step": 1190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019917131770585763,
+      "loss": 1.3965,
+      "step": 1191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019916990372249974,
+      "loss": 1.4773,
+      "step": 1192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019916848853885548,
+      "loss": 1.4333,
+      "step": 1193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019916707215494192,
+      "loss": 1.3584,
+      "step": 1194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019916565457077623,
+      "loss": 1.3891,
+      "step": 1195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019916423578637555,
+      "loss": 1.4673,
+      "step": 1196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019916281580175706,
+      "loss": 1.3479,
+      "step": 1197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019916139461693795,
+      "loss": 1.3266,
+      "step": 1198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019915997223193542,
+      "loss": 1.455,
+      "step": 1199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019915854864676664,
+      "loss": 1.5603,
+      "step": 1200
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 1.447998046875,
+      "eval_runtime": 854.3623,
+      "eval_samples_per_second": 28.161,
+      "eval_steps_per_second": 14.081,
+      "step": 1200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019915712386144892,
+      "loss": 1.3234,
+      "step": 1201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019915569787599946,
+      "loss": 1.5676,
+      "step": 1202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991542706904355,
+      "loss": 1.419,
+      "step": 1203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019915284230477437,
+      "loss": 1.5285,
+      "step": 1204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991514127190333,
+      "loss": 1.2717,
+      "step": 1205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019914998193322963,
+      "loss": 1.4457,
+      "step": 1206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019914854994738067,
+      "loss": 1.4077,
+      "step": 1207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019914711676150378,
+      "loss": 1.3675,
+      "step": 1208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019914568237561624,
+      "loss": 1.5219,
+      "step": 1209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019914424678973543,
+      "loss": 1.6765,
+      "step": 1210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991428100038788,
+      "loss": 1.3485,
+      "step": 1211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019914137201806364,
+      "loss": 1.5957,
+      "step": 1212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019913993283230742,
+      "loss": 1.4228,
+      "step": 1213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991384924466275,
+      "loss": 1.4146,
+      "step": 1214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019913705086104137,
+      "loss": 1.4604,
+      "step": 1215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019913560807556646,
+      "loss": 1.3201,
+      "step": 1216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991341640902202,
+      "loss": 1.4253,
+      "step": 1217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019913271890502013,
+      "loss": 1.5708,
+      "step": 1218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991312725199837,
+      "loss": 1.2653,
+      "step": 1219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991298249351284,
+      "loss": 1.3261,
+      "step": 1220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991283761504718,
+      "loss": 1.4881,
+      "step": 1221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991269261660314,
+      "loss": 1.2535,
+      "step": 1222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019912547498182476,
+      "loss": 1.5078,
+      "step": 1223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019912402259786944,
+      "loss": 1.3638,
+      "step": 1224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199122569014183,
+      "loss": 1.4127,
+      "step": 1225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019912111423078308,
+      "loss": 1.3971,
+      "step": 1226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019911965824768728,
+      "loss": 1.3482,
+      "step": 1227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019911820106491317,
+      "loss": 1.5959,
+      "step": 1228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019911674268247845,
+      "loss": 1.5172,
+      "step": 1229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019911528310040074,
+      "loss": 1.4755,
+      "step": 1230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019911382231869768,
+      "loss": 1.4695,
+      "step": 1231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199112360337387,
+      "loss": 1.4731,
+      "step": 1232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991108971564864,
+      "loss": 1.283,
+      "step": 1233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991094327760135,
+      "loss": 1.4885,
+      "step": 1234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019910796719598613,
+      "loss": 1.3012,
+      "step": 1235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019910650041642195,
+      "loss": 1.4847,
+      "step": 1236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991050324373388,
+      "loss": 1.3364,
+      "step": 1237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019910356325875437,
+      "loss": 1.4219,
+      "step": 1238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019910209288068645,
+      "loss": 1.3965,
+      "step": 1239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001991006213031529,
+      "loss": 1.5299,
+      "step": 1240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019909914852617148,
+      "loss": 1.3727,
+      "step": 1241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019909767454975996,
+      "loss": 1.4257,
+      "step": 1242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019909619937393628,
+      "loss": 1.4041,
+      "step": 1243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019909472299871824,
+      "loss": 1.5627,
+      "step": 1244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019909324542412375,
+      "loss": 1.4931,
+      "step": 1245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019909176665017068,
+      "loss": 1.4168,
+      "step": 1246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001990902866768769,
+      "loss": 1.408,
+      "step": 1247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019908880550426031,
+      "loss": 1.3479,
+      "step": 1248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019908732313233892,
+      "loss": 1.5398,
+      "step": 1249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019908583956113057,
+      "loss": 1.3611,
+      "step": 1250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019908435479065327,
+      "loss": 1.3875,
+      "step": 1251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019908286882092498,
+      "loss": 1.5308,
+      "step": 1252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001990813816519637,
+      "loss": 1.4377,
+      "step": 1253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019907989328378744,
+      "loss": 1.4371,
+      "step": 1254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001990784037164142,
+      "loss": 1.3985,
+      "step": 1255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019907691294986194,
+      "loss": 1.461,
+      "step": 1256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019907542098414883,
+      "loss": 1.4294,
+      "step": 1257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019907392781929283,
+      "loss": 1.4281,
+      "step": 1258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019907243345531207,
+      "loss": 1.354,
+      "step": 1259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019907093789222458,
+      "loss": 1.4299,
+      "step": 1260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001990694411300485,
+      "loss": 1.5635,
+      "step": 1261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019906794316880194,
+      "loss": 1.5043,
+      "step": 1262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019906644400850305,
+      "loss": 1.336,
+      "step": 1263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019906494364916992,
+      "loss": 1.4528,
+      "step": 1264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019906344209082077,
+      "loss": 1.4246,
+      "step": 1265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019906193933347374,
+      "loss": 1.501,
+      "step": 1266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019906043537714702,
+      "loss": 1.4423,
+      "step": 1267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019905893022185883,
+      "loss": 1.5352,
+      "step": 1268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001990574238676274,
+      "loss": 1.5591,
+      "step": 1269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019905591631447093,
+      "loss": 1.4131,
+      "step": 1270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019905440756240762,
+      "loss": 1.4832,
+      "step": 1271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019905289761145583,
+      "loss": 1.3052,
+      "step": 1272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019905138646163377,
+      "loss": 1.4088,
+      "step": 1273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001990498741129598,
+      "loss": 1.4112,
+      "step": 1274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001990483605654521,
+      "loss": 1.4868,
+      "step": 1275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019904684581912912,
+      "loss": 1.439,
+      "step": 1276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019904532987400912,
+      "loss": 1.2671,
+      "step": 1277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019904381273011046,
+      "loss": 1.5947,
+      "step": 1278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019904229438745153,
+      "loss": 1.5204,
+      "step": 1279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019904077484605068,
+      "loss": 1.3991,
+      "step": 1280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019903925410592627,
+      "loss": 1.5684,
+      "step": 1281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019903773216709678,
+      "loss": 1.467,
+      "step": 1282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019903620902958057,
+      "loss": 1.3646,
+      "step": 1283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019903468469339614,
+      "loss": 1.3031,
+      "step": 1284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019903315915856186,
+      "loss": 1.3599,
+      "step": 1285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019903163242509627,
+      "loss": 1.3378,
+      "step": 1286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019903010449301775,
+      "loss": 1.4344,
+      "step": 1287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019902857536234492,
+      "loss": 1.4245,
+      "step": 1288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019902704503309618,
+      "loss": 1.5226,
+      "step": 1289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019902551350529007,
+      "loss": 1.3244,
+      "step": 1290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019902398077894522,
+      "loss": 1.4261,
+      "step": 1291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019902244685408006,
+      "loss": 1.388,
+      "step": 1292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019902091173071323,
+      "loss": 1.4425,
+      "step": 1293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019901937540886327,
+      "loss": 1.5033,
+      "step": 1294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001990178378885488,
+      "loss": 1.4415,
+      "step": 1295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001990162991697884,
+      "loss": 1.279,
+      "step": 1296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019901475925260075,
+      "loss": 1.395,
+      "step": 1297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019901321813700445,
+      "loss": 1.4116,
+      "step": 1298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019901167582301815,
+      "loss": 1.5902,
+      "step": 1299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019901013231066048,
+      "loss": 1.2382,
+      "step": 1300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001990085875999502,
+      "loss": 1.5384,
+      "step": 1301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019900704169090595,
+      "loss": 1.4335,
+      "step": 1302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001990054945835465,
+      "loss": 1.3088,
+      "step": 1303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001990039462778905,
+      "loss": 1.3953,
+      "step": 1304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019900239677395675,
+      "loss": 1.3835,
+      "step": 1305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019900084607176396,
+      "loss": 1.2827,
+      "step": 1306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019899929417133093,
+      "loss": 1.3629,
+      "step": 1307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019899774107267644,
+      "loss": 1.5221,
+      "step": 1308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019899618677581926,
+      "loss": 1.5193,
+      "step": 1309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019899463128077822,
+      "loss": 1.3744,
+      "step": 1310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019899307458757216,
+      "loss": 1.3253,
+      "step": 1311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001989915166962199,
+      "loss": 1.4326,
+      "step": 1312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001989899576067403,
+      "loss": 1.5074,
+      "step": 1313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019898839731915226,
+      "loss": 1.3198,
+      "step": 1314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019898683583347463,
+      "loss": 1.3394,
+      "step": 1315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019898527314972627,
+      "loss": 1.4402,
+      "step": 1316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019898370926792623,
+      "loss": 1.3311,
+      "step": 1317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001989821441880933,
+      "loss": 1.2963,
+      "step": 1318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019898057791024645,
+      "loss": 1.4622,
+      "step": 1319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001989790104344047,
+      "loss": 1.5302,
+      "step": 1320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019897744176058696,
+      "loss": 1.4686,
+      "step": 1321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019897587188881226,
+      "loss": 1.6166,
+      "step": 1322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019897430081909957,
+      "loss": 1.34,
+      "step": 1323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001989727285514679,
+      "loss": 1.5206,
+      "step": 1324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019897115508593632,
+      "loss": 1.3539,
+      "step": 1325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019896958042252383,
+      "loss": 1.2299,
+      "step": 1326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019896800456124951,
+      "loss": 1.4265,
+      "step": 1327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019896642750213243,
+      "loss": 1.2916,
+      "step": 1328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001989648492451917,
+      "loss": 1.4109,
+      "step": 1329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019896326979044636,
+      "loss": 1.3597,
+      "step": 1330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019896168913791556,
+      "loss": 1.5052,
+      "step": 1331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001989601072876185,
+      "loss": 1.4664,
+      "step": 1332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019895852423957422,
+      "loss": 1.2843,
+      "step": 1333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019895693999380192,
+      "loss": 1.5043,
+      "step": 1334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001989553545503208,
+      "loss": 1.4511,
+      "step": 1335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019895376790914998,
+      "loss": 1.3678,
+      "step": 1336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019895218007030878,
+      "loss": 1.3846,
+      "step": 1337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001989505910338163,
+      "loss": 1.3052,
+      "step": 1338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019894900079969182,
+      "loss": 1.4477,
+      "step": 1339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019894740936795462,
+      "loss": 1.403,
+      "step": 1340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001989458167386239,
+      "loss": 1.4769,
+      "step": 1341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000198944222911719,
+      "loss": 1.4691,
+      "step": 1342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019894262788725916,
+      "loss": 1.4735,
+      "step": 1343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001989410316652637,
+      "loss": 1.4378,
+      "step": 1344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019893943424575193,
+      "loss": 1.4044,
+      "step": 1345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019893783562874323,
+      "loss": 1.2626,
+      "step": 1346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019893623581425688,
+      "loss": 1.3395,
+      "step": 1347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019893463480231228,
+      "loss": 1.469,
+      "step": 1348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019893303259292883,
+      "loss": 1.4379,
+      "step": 1349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019893142918612588,
+      "loss": 1.5179,
+      "step": 1350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019892982458192288,
+      "loss": 1.4062,
+      "step": 1351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019892821878033919,
+      "loss": 1.6174,
+      "step": 1352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001989266117813943,
+      "loss": 1.4138,
+      "step": 1353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001989250035851076,
+      "loss": 1.4034,
+      "step": 1354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019892339419149866,
+      "loss": 1.3674,
+      "step": 1355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019892178360058686,
+      "loss": 1.4972,
+      "step": 1356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019892017181239173,
+      "loss": 1.3039,
+      "step": 1357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019891855882693275,
+      "loss": 1.4787,
+      "step": 1358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019891694464422953,
+      "loss": 1.3597,
+      "step": 1359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001989153292643015,
+      "loss": 1.5048,
+      "step": 1360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019891371268716826,
+      "loss": 1.4035,
+      "step": 1361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001989120949128494,
+      "loss": 1.2618,
+      "step": 1362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019891047594136442,
+      "loss": 1.5,
+      "step": 1363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000198908855772733,
+      "loss": 1.4598,
+      "step": 1364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019890723440697472,
+      "loss": 1.4299,
+      "step": 1365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001989056118441092,
+      "loss": 1.3765,
+      "step": 1366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019890398808415606,
+      "loss": 1.3671,
+      "step": 1367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000198902363127135,
+      "loss": 1.4121,
+      "step": 1368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019890073697306567,
+      "loss": 1.3612,
+      "step": 1369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019889910962196772,
+      "loss": 1.4198,
+      "step": 1370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019889748107386086,
+      "loss": 1.361,
+      "step": 1371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019889585132876482,
+      "loss": 1.5283,
+      "step": 1372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019889422038669934,
+      "loss": 1.4386,
+      "step": 1373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988925882476841,
+      "loss": 1.4992,
+      "step": 1374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988909549117389,
+      "loss": 1.2294,
+      "step": 1375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019888932037888347,
+      "loss": 1.6264,
+      "step": 1376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019888768464913764,
+      "loss": 1.3712,
+      "step": 1377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019888604772252123,
+      "loss": 1.2984,
+      "step": 1378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019888440959905395,
+      "loss": 1.4363,
+      "step": 1379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019888277027875573,
+      "loss": 1.4183,
+      "step": 1380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019888112976164634,
+      "loss": 1.4699,
+      "step": 1381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019887948804774567,
+      "loss": 1.3608,
+      "step": 1382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988778451370736,
+      "loss": 1.3284,
+      "step": 1383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019887620102965002,
+      "loss": 1.3777,
+      "step": 1384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019887455572549476,
+      "loss": 1.3867,
+      "step": 1385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988729092246278,
+      "loss": 1.2883,
+      "step": 1386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019887126152706906,
+      "loss": 1.2855,
+      "step": 1387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019886961263283846,
+      "loss": 1.4898,
+      "step": 1388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019886796254195598,
+      "loss": 1.595,
+      "step": 1389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019886631125444156,
+      "loss": 1.4736,
+      "step": 1390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019886465877031525,
+      "loss": 1.5136,
+      "step": 1391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019886300508959695,
+      "loss": 1.502,
+      "step": 1392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988613502123068,
+      "loss": 1.3556,
+      "step": 1393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988596941384647,
+      "loss": 1.5341,
+      "step": 1394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988580368680908,
+      "loss": 1.3104,
+      "step": 1395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988563784012051,
+      "loss": 1.4196,
+      "step": 1396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988547187378277,
+      "loss": 1.4005,
+      "step": 1397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019885305787797867,
+      "loss": 1.4323,
+      "step": 1398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019885139582167814,
+      "loss": 1.4628,
+      "step": 1399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019884973256894614,
+      "loss": 1.347,
+      "step": 1400
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 1.3938785791397095,
+      "eval_runtime": 845.2439,
+      "eval_samples_per_second": 28.465,
+      "eval_steps_per_second": 14.233,
+      "step": 1400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019884806811980293,
+      "loss": 1.353,
+      "step": 1401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019884640247426857,
+      "loss": 1.4482,
+      "step": 1402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019884473563236323,
+      "loss": 1.4621,
+      "step": 1403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988430675941071,
+      "loss": 1.4886,
+      "step": 1404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019884139835952037,
+      "loss": 1.5339,
+      "step": 1405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019883972792862322,
+      "loss": 1.2392,
+      "step": 1406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988380563014359,
+      "loss": 1.3823,
+      "step": 1407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019883638347797862,
+      "loss": 1.3743,
+      "step": 1408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019883470945827165,
+      "loss": 1.4193,
+      "step": 1409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019883303424233522,
+      "loss": 1.426,
+      "step": 1410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988313578301896,
+      "loss": 1.2971,
+      "step": 1411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019882968022185514,
+      "loss": 1.4454,
+      "step": 1412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988280014173521,
+      "loss": 1.397,
+      "step": 1413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988263214167008,
+      "loss": 1.5336,
+      "step": 1414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019882464021992154,
+      "loss": 1.416,
+      "step": 1415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019882295782703478,
+      "loss": 1.4484,
+      "step": 1416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019882127423806077,
+      "loss": 1.5188,
+      "step": 1417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988195894530199,
+      "loss": 1.4351,
+      "step": 1418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019881790347193263,
+      "loss": 1.3905,
+      "step": 1419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019881621629481928,
+      "loss": 1.4176,
+      "step": 1420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019881452792170037,
+      "loss": 1.3606,
+      "step": 1421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019881283835259623,
+      "loss": 1.4434,
+      "step": 1422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988111475875274,
+      "loss": 1.3518,
+      "step": 1423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019880945562651426,
+      "loss": 1.2715,
+      "step": 1424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019880776246957734,
+      "loss": 1.4658,
+      "step": 1425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019880606811673715,
+      "loss": 1.501,
+      "step": 1426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019880437256801414,
+      "loss": 1.4129,
+      "step": 1427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001988026758234289,
+      "loss": 1.3697,
+      "step": 1428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019880097788300188,
+      "loss": 1.4737,
+      "step": 1429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001987992787467537,
+      "loss": 1.2478,
+      "step": 1430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019879757841470493,
+      "loss": 1.5273,
+      "step": 1431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019879587688687608,
+      "loss": 1.4718,
+      "step": 1432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019879417416328782,
+      "loss": 1.3476,
+      "step": 1433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001987924702439607,
+      "loss": 1.3189,
+      "step": 1434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019879076512891542,
+      "loss": 1.327,
+      "step": 1435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019878905881817252,
+      "loss": 1.2492,
+      "step": 1436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001987873513117527,
+      "loss": 1.5667,
+      "step": 1437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019878564260967667,
+      "loss": 1.447,
+      "step": 1438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019878393271196502,
+      "loss": 1.3489,
+      "step": 1439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019878222161863853,
+      "loss": 1.5189,
+      "step": 1440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019878050932971785,
+      "loss": 1.414,
+      "step": 1441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019877879584522372,
+      "loss": 1.3022,
+      "step": 1442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019877708116517693,
+      "loss": 1.5749,
+      "step": 1443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019877536528959815,
+      "loss": 1.4076,
+      "step": 1444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019877364821850818,
+      "loss": 1.1879,
+      "step": 1445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019877192995192784,
+      "loss": 1.252,
+      "step": 1446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019877021048987788,
+      "loss": 1.3749,
+      "step": 1447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019876848983237913,
+      "loss": 1.2986,
+      "step": 1448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001987667679794524,
+      "loss": 1.2868,
+      "step": 1449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019876504493111856,
+      "loss": 1.3575,
+      "step": 1450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019876332068739845,
+      "loss": 1.3431,
+      "step": 1451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019876159524831292,
+      "loss": 1.4676,
+      "step": 1452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001987598686138829,
+      "loss": 1.4699,
+      "step": 1453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019875814078412925,
+      "loss": 1.3882,
+      "step": 1454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019875641175907288,
+      "loss": 1.421,
+      "step": 1455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001987546815387347,
+      "loss": 1.5307,
+      "step": 1456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019875295012313572,
+      "loss": 1.3228,
+      "step": 1457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019875121751229685,
+      "loss": 1.4236,
+      "step": 1458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019874948370623906,
+      "loss": 1.236,
+      "step": 1459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019874774870498336,
+      "loss": 1.2366,
+      "step": 1460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001987460125085507,
+      "loss": 1.3441,
+      "step": 1461
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019874427511696214,
+      "loss": 1.4607,
+      "step": 1462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019874253653023866,
+      "loss": 1.4244,
+      "step": 1463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019874079674840137,
+      "loss": 1.3803,
+      "step": 1464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019873905577147129,
+      "loss": 1.3064,
+      "step": 1465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001987373135994695,
+      "loss": 1.4738,
+      "step": 1466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019873557023241707,
+      "loss": 1.4615,
+      "step": 1467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001987338256703351,
+      "loss": 1.324,
+      "step": 1468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019873207991324472,
+      "loss": 1.463,
+      "step": 1469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019873033296116706,
+      "loss": 1.3059,
+      "step": 1470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019872858481412325,
+      "loss": 1.437,
+      "step": 1471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019872683547213446,
+      "loss": 1.2529,
+      "step": 1472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019872508493522188,
+      "loss": 1.3566,
+      "step": 1473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019872333320340664,
+      "loss": 1.5463,
+      "step": 1474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019872158027671,
+      "loss": 1.3307,
+      "step": 1475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019871982615515317,
+      "loss": 1.3298,
+      "step": 1476
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019871807083875735,
+      "loss": 1.1388,
+      "step": 1477
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001987163143275438,
+      "loss": 1.3766,
+      "step": 1478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019871455662153379,
+      "loss": 1.3308,
+      "step": 1479
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019871279772074857,
+      "loss": 1.3836,
+      "step": 1480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019871103762520944,
+      "loss": 1.3639,
+      "step": 1481
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019870927633493772,
+      "loss": 1.3479,
+      "step": 1482
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001987075138499547,
+      "loss": 1.4622,
+      "step": 1483
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019870575017028172,
+      "loss": 1.4179,
+      "step": 1484
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019870398529594016,
+      "loss": 1.3406,
+      "step": 1485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019870221922695133,
+      "loss": 1.5578,
+      "step": 1486
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019870045196333663,
+      "loss": 1.2313,
+      "step": 1487
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986986835051175,
+      "loss": 1.3853,
+      "step": 1488
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019869691385231524,
+      "loss": 1.4064,
+      "step": 1489
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019869514300495134,
+      "loss": 1.1968,
+      "step": 1490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019869337096304718,
+      "loss": 1.5992,
+      "step": 1491
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986915977266243,
+      "loss": 1.3329,
+      "step": 1492
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019868982329570404,
+      "loss": 1.3748,
+      "step": 1493
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000198688047670308,
+      "loss": 1.4621,
+      "step": 1494
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019868627085045757,
+      "loss": 1.4262,
+      "step": 1495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019868449283617432,
+      "loss": 1.4358,
+      "step": 1496
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019868271362747972,
+      "loss": 1.3738,
+      "step": 1497
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019868093322439537,
+      "loss": 1.3732,
+      "step": 1498
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019867915162694274,
+      "loss": 1.4088,
+      "step": 1499
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019867736883514344,
+      "loss": 1.4106,
+      "step": 1500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019867558484901906,
+      "loss": 1.407,
+      "step": 1501
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019867379966859118,
+      "loss": 1.4913,
+      "step": 1502
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019867201329388137,
+      "loss": 1.3291,
+      "step": 1503
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986702257249113,
+      "loss": 1.2714,
+      "step": 1504
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986684369617026,
+      "loss": 1.2537,
+      "step": 1505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019866664700427688,
+      "loss": 1.3331,
+      "step": 1506
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019866485585265587,
+      "loss": 1.1974,
+      "step": 1507
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019866306350686114,
+      "loss": 1.2832,
+      "step": 1508
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986612699669145,
+      "loss": 1.3194,
+      "step": 1509
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986594752328376,
+      "loss": 1.2946,
+      "step": 1510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986576793046522,
+      "loss": 1.4549,
+      "step": 1511
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019865588218238,
+      "loss": 1.3482,
+      "step": 1512
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986540838660427,
+      "loss": 1.3945,
+      "step": 1513
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986522843556622,
+      "loss": 1.3282,
+      "step": 1514
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986504836512602,
+      "loss": 1.3199,
+      "step": 1515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019864868175285846,
+      "loss": 1.3254,
+      "step": 1516
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986468786604789,
+      "loss": 1.3513,
+      "step": 1517
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019864507437414322,
+      "loss": 1.3826,
+      "step": 1518
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019864326889387332,
+      "loss": 1.4813,
+      "step": 1519
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019864146221969101,
+      "loss": 1.4176,
+      "step": 1520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019863965435161824,
+      "loss": 1.3027,
+      "step": 1521
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019863784528967684,
+      "loss": 1.4584,
+      "step": 1522
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019863603503388869,
+      "loss": 1.421,
+      "step": 1523
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986342235842757,
+      "loss": 1.3096,
+      "step": 1524
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019863241094085984,
+      "loss": 1.4589,
+      "step": 1525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019863059710366302,
+      "loss": 1.3011,
+      "step": 1526
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986287820727072,
+      "loss": 1.4787,
+      "step": 1527
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019862696584801434,
+      "loss": 1.4041,
+      "step": 1528
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986251484296064,
+      "loss": 1.4185,
+      "step": 1529
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019862332981750543,
+      "loss": 1.2851,
+      "step": 1530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986215100117334,
+      "loss": 1.2378,
+      "step": 1531
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019861968901231235,
+      "loss": 1.4699,
+      "step": 1532
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019861786681926434,
+      "loss": 1.4193,
+      "step": 1533
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019861604343261139,
+      "loss": 1.5145,
+      "step": 1534
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019861421885237557,
+      "loss": 1.4066,
+      "step": 1535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000198612393078579,
+      "loss": 1.2268,
+      "step": 1536
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986105661112437,
+      "loss": 1.3956,
+      "step": 1537
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986087379503919,
+      "loss": 1.3366,
+      "step": 1538
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019860690859604564,
+      "loss": 1.2978,
+      "step": 1539
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001986050780482271,
+      "loss": 1.5215,
+      "step": 1540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019860324630695839,
+      "loss": 1.4012,
+      "step": 1541
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019860141337226173,
+      "loss": 1.5428,
+      "step": 1542
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019859957924415927,
+      "loss": 1.3863,
+      "step": 1543
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019859774392267326,
+      "loss": 1.3872,
+      "step": 1544
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019859590740782583,
+      "loss": 1.4383,
+      "step": 1545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001985940696996393,
+      "loss": 1.3502,
+      "step": 1546
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019859223079813582,
+      "loss": 1.2706,
+      "step": 1547
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019859039070333775,
+      "loss": 1.4311,
+      "step": 1548
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019858854941526727,
+      "loss": 1.487,
+      "step": 1549
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001985867069339467,
+      "loss": 1.2886,
+      "step": 1550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019858486325939837,
+      "loss": 1.1533,
+      "step": 1551
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019858301839164452,
+      "loss": 1.3024,
+      "step": 1552
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019858117233070757,
+      "loss": 1.2767,
+      "step": 1553
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001985793250766098,
+      "loss": 1.3045,
+      "step": 1554
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001985774766293736,
+      "loss": 1.4337,
+      "step": 1555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019857562698902132,
+      "loss": 1.2606,
+      "step": 1556
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019857377615557534,
+      "loss": 1.3975,
+      "step": 1557
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019857192412905812,
+      "loss": 1.4364,
+      "step": 1558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019857007090949203,
+      "loss": 1.3271,
+      "step": 1559
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019856821649689947,
+      "loss": 1.4012,
+      "step": 1560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001985663608913029,
+      "loss": 1.3442,
+      "step": 1561
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019856450409272487,
+      "loss": 1.3345,
+      "step": 1562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019856264610118772,
+      "loss": 1.4034,
+      "step": 1563
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019856078691671402,
+      "loss": 1.4501,
+      "step": 1564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019855892653932627,
+      "loss": 1.3364,
+      "step": 1565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019855706496904694,
+      "loss": 1.425,
+      "step": 1566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001985552022058986,
+      "loss": 1.3047,
+      "step": 1567
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019855333824990378,
+      "loss": 1.3974,
+      "step": 1568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019855147310108504,
+      "loss": 1.4178,
+      "step": 1569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019854960675946495,
+      "loss": 1.3879,
+      "step": 1570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019854773922506612,
+      "loss": 1.3818,
+      "step": 1571
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019854587049791114,
+      "loss": 1.3216,
+      "step": 1572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019854400057802264,
+      "loss": 1.3382,
+      "step": 1573
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019854212946542323,
+      "loss": 1.4309,
+      "step": 1574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019854025716013559,
+      "loss": 1.3897,
+      "step": 1575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019853838366218235,
+      "loss": 1.388,
+      "step": 1576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019853650897158618,
+      "loss": 1.201,
+      "step": 1577
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019853463308836978,
+      "loss": 1.4013,
+      "step": 1578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019853275601255588,
+      "loss": 1.3145,
+      "step": 1579
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019853087774416716,
+      "loss": 1.4255,
+      "step": 1580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001985289982832264,
+      "loss": 1.2957,
+      "step": 1581
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001985271176297563,
+      "loss": 1.3998,
+      "step": 1582
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019852523578377964,
+      "loss": 1.339,
+      "step": 1583
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019852335274531918,
+      "loss": 1.3493,
+      "step": 1584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019852146851439778,
+      "loss": 1.2854,
+      "step": 1585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019851958309103813,
+      "loss": 1.318,
+      "step": 1586
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019851769647526315,
+      "loss": 1.1384,
+      "step": 1587
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019851580866709565,
+      "loss": 1.4932,
+      "step": 1588
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019851391966655844,
+      "loss": 1.4705,
+      "step": 1589
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019851202947367442,
+      "loss": 1.316,
+      "step": 1590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019851013808846646,
+      "loss": 1.444,
+      "step": 1591
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019850824551095747,
+      "loss": 1.332,
+      "step": 1592
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019850635174117033,
+      "loss": 1.2078,
+      "step": 1593
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019850445677912796,
+      "loss": 1.2532,
+      "step": 1594
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001985025606248533,
+      "loss": 1.3327,
+      "step": 1595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001985006632783693,
+      "loss": 1.4101,
+      "step": 1596
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019849876473969892,
+      "loss": 1.2839,
+      "step": 1597
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019849686500886517,
+      "loss": 1.2811,
+      "step": 1598
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000198494964085891,
+      "loss": 1.4443,
+      "step": 1599
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019849306197079945,
+      "loss": 1.2312,
+      "step": 1600
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 1.344958782196045,
+      "eval_runtime": 853.5281,
+      "eval_samples_per_second": 28.189,
+      "eval_steps_per_second": 14.094,
+      "step": 1600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019849115866361352,
+      "loss": 1.2967,
+      "step": 1601
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019848925416435624,
+      "loss": 1.314,
+      "step": 1602
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001984873484730507,
+      "loss": 1.4782,
+      "step": 1603
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019848544158971993,
+      "loss": 1.4346,
+      "step": 1604
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019848353351438705,
+      "loss": 1.4878,
+      "step": 1605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001984816242470751,
+      "loss": 1.3461,
+      "step": 1606
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001984797137878072,
+      "loss": 1.2518,
+      "step": 1607
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001984778021366065,
+      "loss": 1.3791,
+      "step": 1608
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019847588929349611,
+      "loss": 1.2735,
+      "step": 1609
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001984739752584992,
+      "loss": 1.2355,
+      "step": 1610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019847206003163894,
+      "loss": 1.3502,
+      "step": 1611
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001984701436129385,
+      "loss": 1.2874,
+      "step": 1612
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001984682260024211,
+      "loss": 1.485,
+      "step": 1613
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019846630720010987,
+      "loss": 1.2652,
+      "step": 1614
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019846438720602815,
+      "loss": 1.3025,
+      "step": 1615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001984624660201991,
+      "loss": 1.1707,
+      "step": 1616
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019846054364264598,
+      "loss": 1.3486,
+      "step": 1617
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019845862007339206,
+      "loss": 1.1837,
+      "step": 1618
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019845669531246064,
+      "loss": 1.4184,
+      "step": 1619
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019845476935987502,
+      "loss": 1.4533,
+      "step": 1620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019845284221565846,
+      "loss": 1.2627,
+      "step": 1621
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019845091387983436,
+      "loss": 1.2662,
+      "step": 1622
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019844898435242603,
+      "loss": 1.4094,
+      "step": 1623
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019844705363345676,
+      "loss": 1.3836,
+      "step": 1624
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019844512172295003,
+      "loss": 1.3788,
+      "step": 1625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019844318862092912,
+      "loss": 1.426,
+      "step": 1626
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001984412543274175,
+      "loss": 1.4527,
+      "step": 1627
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019843931884243855,
+      "loss": 1.5341,
+      "step": 1628
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001984373821660157,
+      "loss": 1.449,
+      "step": 1629
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019843544429817237,
+      "loss": 1.3345,
+      "step": 1630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019843350523893207,
+      "loss": 1.4473,
+      "step": 1631
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001984315649883182,
+      "loss": 1.4306,
+      "step": 1632
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001984296235463543,
+      "loss": 1.389,
+      "step": 1633
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019842768091306382,
+      "loss": 1.2924,
+      "step": 1634
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019842573708847032,
+      "loss": 1.3763,
+      "step": 1635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001984237920725973,
+      "loss": 1.33,
+      "step": 1636
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001984218458654683,
+      "loss": 1.4085,
+      "step": 1637
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001984198984671069,
+      "loss": 1.1533,
+      "step": 1638
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019841794987753662,
+      "loss": 1.274,
+      "step": 1639
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001984160000967811,
+      "loss": 1.4952,
+      "step": 1640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019841404912486387,
+      "loss": 1.3872,
+      "step": 1641
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019841209696180861,
+      "loss": 1.4679,
+      "step": 1642
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019841014360763896,
+      "loss": 1.2578,
+      "step": 1643
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001984081890623785,
+      "loss": 1.3467,
+      "step": 1644
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019840623332605093,
+      "loss": 1.1851,
+      "step": 1645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019840427639867986,
+      "loss": 1.3761,
+      "step": 1646
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019840231828028906,
+      "loss": 1.3163,
+      "step": 1647
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019840035897090215,
+      "loss": 1.2519,
+      "step": 1648
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019839839847054296,
+      "loss": 1.2279,
+      "step": 1649
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019839643677923508,
+      "loss": 1.148,
+      "step": 1650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019839447389700233,
+      "loss": 1.2212,
+      "step": 1651
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019839250982386846,
+      "loss": 1.3851,
+      "step": 1652
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019839054455985723,
+      "loss": 1.466,
+      "step": 1653
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019838857810499242,
+      "loss": 1.2524,
+      "step": 1654
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019838661045929785,
+      "loss": 1.3338,
+      "step": 1655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019838464162279734,
+      "loss": 1.5305,
+      "step": 1656
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001983826715955147,
+      "loss": 1.3382,
+      "step": 1657
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019838070037747377,
+      "loss": 1.3637,
+      "step": 1658
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019837872796869837,
+      "loss": 1.2982,
+      "step": 1659
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001983767543692125,
+      "loss": 1.2906,
+      "step": 1660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019837477957903996,
+      "loss": 1.3454,
+      "step": 1661
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019837280359820463,
+      "loss": 1.2607,
+      "step": 1662
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019837082642673045,
+      "loss": 1.38,
+      "step": 1663
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001983688480646414,
+      "loss": 1.397,
+      "step": 1664
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019836686851196133,
+      "loss": 1.2822,
+      "step": 1665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019836488776871428,
+      "loss": 1.2634,
+      "step": 1666
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019836290583492418,
+      "loss": 1.3493,
+      "step": 1667
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019836092271061505,
+      "loss": 1.4424,
+      "step": 1668
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019835893839581086,
+      "loss": 1.2654,
+      "step": 1669
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001983569528905356,
+      "loss": 1.2782,
+      "step": 1670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019835496619481343,
+      "loss": 1.2446,
+      "step": 1671
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019835297830866826,
+      "loss": 1.269,
+      "step": 1672
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019835098923212422,
+      "loss": 1.3569,
+      "step": 1673
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001983489989652053,
+      "loss": 1.1567,
+      "step": 1674
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001983470075079357,
+      "loss": 1.2301,
+      "step": 1675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019834501486033947,
+      "loss": 1.3305,
+      "step": 1676
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019834302102244073,
+      "loss": 1.3073,
+      "step": 1677
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019834102599426362,
+      "loss": 1.3473,
+      "step": 1678
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019833902977583228,
+      "loss": 1.4603,
+      "step": 1679
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019833703236717086,
+      "loss": 1.4078,
+      "step": 1680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019833503376830356,
+      "loss": 1.3619,
+      "step": 1681
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019833303397925452,
+      "loss": 1.2757,
+      "step": 1682
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019833103300004803,
+      "loss": 1.2762,
+      "step": 1683
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019832903083070823,
+      "loss": 1.2862,
+      "step": 1684
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019832702747125938,
+      "loss": 1.3452,
+      "step": 1685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019832502292172578,
+      "loss": 1.3718,
+      "step": 1686
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019832301718213156,
+      "loss": 1.2736,
+      "step": 1687
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019832101025250113,
+      "loss": 1.303,
+      "step": 1688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019831900213285872,
+      "loss": 1.4179,
+      "step": 1689
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019831699282322863,
+      "loss": 1.3307,
+      "step": 1690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019831498232363522,
+      "loss": 1.2888,
+      "step": 1691
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019831297063410276,
+      "loss": 1.1974,
+      "step": 1692
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019831095775465566,
+      "loss": 1.3069,
+      "step": 1693
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019830894368531824,
+      "loss": 1.3416,
+      "step": 1694
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019830692842611489,
+      "loss": 1.3408,
+      "step": 1695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019830491197707004,
+      "loss": 1.3446,
+      "step": 1696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000198302894338208,
+      "loss": 1.4906,
+      "step": 1697
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001983008755095533,
+      "loss": 1.3928,
+      "step": 1698
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019829885549113035,
+      "loss": 1.4524,
+      "step": 1699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001982968342829635,
+      "loss": 1.3768,
+      "step": 1700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019829481188507732,
+      "loss": 1.2599,
+      "step": 1701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019829278829749627,
+      "loss": 1.2543,
+      "step": 1702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019829076352024483,
+      "loss": 1.3408,
+      "step": 1703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001982887375533475,
+      "loss": 1.3602,
+      "step": 1704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019828671039682878,
+      "loss": 1.167,
+      "step": 1705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019828468205071323,
+      "loss": 1.0686,
+      "step": 1706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019828265251502542,
+      "loss": 1.2242,
+      "step": 1707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001982806217897899,
+      "loss": 1.3517,
+      "step": 1708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019827858987503123,
+      "loss": 1.3633,
+      "step": 1709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000198276556770774,
+      "loss": 1.374,
+      "step": 1710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019827452247704288,
+      "loss": 1.2583,
+      "step": 1711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001982724869938624,
+      "loss": 1.3118,
+      "step": 1712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019827045032125726,
+      "loss": 1.3169,
+      "step": 1713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019826841245925212,
+      "loss": 1.3556,
+      "step": 1714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019826637340787157,
+      "loss": 1.2536,
+      "step": 1715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019826433316714036,
+      "loss": 1.2905,
+      "step": 1716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019826229173708316,
+      "loss": 1.3702,
+      "step": 1717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019826024911772465,
+      "loss": 1.2956,
+      "step": 1718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019825820530908962,
+      "loss": 1.3635,
+      "step": 1719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019825616031120272,
+      "loss": 1.323,
+      "step": 1720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001982541141240888,
+      "loss": 1.2738,
+      "step": 1721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019825206674777254,
+      "loss": 1.2895,
+      "step": 1722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019825001818227876,
+      "loss": 1.3171,
+      "step": 1723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019824796842763225,
+      "loss": 1.3549,
+      "step": 1724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019824591748385782,
+      "loss": 1.3079,
+      "step": 1725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019824386535098026,
+      "loss": 1.2317,
+      "step": 1726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001982418120290245,
+      "loss": 1.5024,
+      "step": 1727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019823975751801526,
+      "loss": 1.2031,
+      "step": 1728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019823770181797752,
+      "loss": 1.3806,
+      "step": 1729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001982356449289361,
+      "loss": 1.3226,
+      "step": 1730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019823358685091595,
+      "loss": 1.2307,
+      "step": 1731
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001982315275839419,
+      "loss": 1.3032,
+      "step": 1732
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019822946712803896,
+      "loss": 1.3095,
+      "step": 1733
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000198227405483232,
+      "loss": 1.3554,
+      "step": 1734
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019822534264954598,
+      "loss": 1.2743,
+      "step": 1735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001982232786270059,
+      "loss": 1.3747,
+      "step": 1736
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001982212134156367,
+      "loss": 1.2931,
+      "step": 1737
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019821914701546346,
+      "loss": 1.2746,
+      "step": 1738
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001982170794265111,
+      "loss": 1.2063,
+      "step": 1739
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019821501064880469,
+      "loss": 1.3583,
+      "step": 1740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019821294068236923,
+      "loss": 1.2974,
+      "step": 1741
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001982108695272298,
+      "loss": 1.2475,
+      "step": 1742
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001982087971834115,
+      "loss": 1.3775,
+      "step": 1743
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019820672365093936,
+      "loss": 1.3974,
+      "step": 1744
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001982046489298385,
+      "loss": 1.4875,
+      "step": 1745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000198202573020134,
+      "loss": 1.2887,
+      "step": 1746
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019820049592185105,
+      "loss": 1.3625,
+      "step": 1747
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019819841763501475,
+      "loss": 1.2296,
+      "step": 1748
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019819633815965025,
+      "loss": 1.406,
+      "step": 1749
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019819425749578273,
+      "loss": 1.3048,
+      "step": 1750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019819217564343735,
+      "loss": 1.3028,
+      "step": 1751
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019819009260263934,
+      "loss": 1.2664,
+      "step": 1752
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001981880083734139,
+      "loss": 1.4576,
+      "step": 1753
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019818592295578626,
+      "loss": 1.2141,
+      "step": 1754
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019818383634978164,
+      "loss": 1.2897,
+      "step": 1755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019818174855542532,
+      "loss": 1.4906,
+      "step": 1756
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019817965957274255,
+      "loss": 1.3827,
+      "step": 1757
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001981775694017586,
+      "loss": 1.1883,
+      "step": 1758
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019817547804249882,
+      "loss": 1.2726,
+      "step": 1759
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001981733854949885,
+      "loss": 1.215,
+      "step": 1760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019817129175925296,
+      "loss": 1.2849,
+      "step": 1761
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019816919683531751,
+      "loss": 1.3224,
+      "step": 1762
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019816710072320757,
+      "loss": 1.4219,
+      "step": 1763
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019816500342294843,
+      "loss": 1.3084,
+      "step": 1764
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001981629049345656,
+      "loss": 1.4805,
+      "step": 1765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019816080525808435,
+      "loss": 1.3061,
+      "step": 1766
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001981587043935301,
+      "loss": 1.4379,
+      "step": 1767
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001981566023409284,
+      "loss": 1.1277,
+      "step": 1768
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019815449910030454,
+      "loss": 1.3186,
+      "step": 1769
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019815239467168408,
+      "loss": 1.3413,
+      "step": 1770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019815028905509246,
+      "loss": 1.2468,
+      "step": 1771
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019814818225055518,
+      "loss": 1.2485,
+      "step": 1772
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001981460742580977,
+      "loss": 1.3112,
+      "step": 1773
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019814396507774555,
+      "loss": 1.2775,
+      "step": 1774
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019814185470952428,
+      "loss": 1.3419,
+      "step": 1775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019813974315345943,
+      "loss": 1.3742,
+      "step": 1776
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019813763040957651,
+      "loss": 1.1866,
+      "step": 1777
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019813551647790116,
+      "loss": 1.3392,
+      "step": 1778
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019813340135845891,
+      "loss": 1.4103,
+      "step": 1779
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019813128505127538,
+      "loss": 1.186,
+      "step": 1780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001981291675563762,
+      "loss": 1.3438,
+      "step": 1781
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019812704887378697,
+      "loss": 1.2177,
+      "step": 1782
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019812492900353335,
+      "loss": 1.1884,
+      "step": 1783
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019812280794564098,
+      "loss": 1.3201,
+      "step": 1784
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019812068570013557,
+      "loss": 1.3303,
+      "step": 1785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019811856226704277,
+      "loss": 1.2491,
+      "step": 1786
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019811643764638828,
+      "loss": 1.3772,
+      "step": 1787
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001981143118381979,
+      "loss": 1.3398,
+      "step": 1788
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001981121848424972,
+      "loss": 1.1842,
+      "step": 1789
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019811005665931205,
+      "loss": 1.4456,
+      "step": 1790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019810792728866815,
+      "loss": 1.3168,
+      "step": 1791
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019810579673059132,
+      "loss": 1.193,
+      "step": 1792
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001981036649851073,
+      "loss": 1.2172,
+      "step": 1793
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019810153205224192,
+      "loss": 1.3829,
+      "step": 1794
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019809939793202097,
+      "loss": 1.2553,
+      "step": 1795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019809726262447032,
+      "loss": 1.3837,
+      "step": 1796
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019809512612961578,
+      "loss": 1.2359,
+      "step": 1797
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001980929884474832,
+      "loss": 1.2188,
+      "step": 1798
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001980908495780985,
+      "loss": 1.2024,
+      "step": 1799
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019808870952148753,
+      "loss": 1.3749,
+      "step": 1800
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 1.3037443161010742,
+      "eval_runtime": 854.5045,
+      "eval_samples_per_second": 28.157,
+      "eval_steps_per_second": 14.078,
+      "step": 1800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001980865682776762,
+      "loss": 1.329,
+      "step": 1801
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001980844258466904,
+      "loss": 1.2151,
+      "step": 1802
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019808228222855615,
+      "loss": 1.3412,
+      "step": 1803
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019808013742329928,
+      "loss": 1.1813,
+      "step": 1804
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001980779914309458,
+      "loss": 1.2177,
+      "step": 1805
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001980758442515217,
+      "loss": 1.2986,
+      "step": 1806
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019807369588505295,
+      "loss": 1.2791,
+      "step": 1807
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019807154633156557,
+      "loss": 1.2908,
+      "step": 1808
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019806939559108554,
+      "loss": 1.2385,
+      "step": 1809
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019806724366363895,
+      "loss": 1.2658,
+      "step": 1810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019806509054925178,
+      "loss": 1.3092,
+      "step": 1811
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019806293624795011,
+      "loss": 1.2426,
+      "step": 1812
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019806078075976005,
+      "loss": 1.3071,
+      "step": 1813
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019805862408470767,
+      "loss": 1.2982,
+      "step": 1814
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019805646622281903,
+      "loss": 1.2045,
+      "step": 1815
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001980543071741203,
+      "loss": 1.3549,
+      "step": 1816
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001980521469386376,
+      "loss": 1.2216,
+      "step": 1817
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019804998551639706,
+      "loss": 1.2178,
+      "step": 1818
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019804782290742487,
+      "loss": 1.3313,
+      "step": 1819
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019804565911174716,
+      "loss": 1.3279,
+      "step": 1820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019804349412939014,
+      "loss": 1.3423,
+      "step": 1821
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019804132796038003,
+      "loss": 1.4188,
+      "step": 1822
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019803916060474305,
+      "loss": 1.4791,
+      "step": 1823
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001980369920625054,
+      "loss": 1.2343,
+      "step": 1824
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019803482233369336,
+      "loss": 1.3213,
+      "step": 1825
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019803265141833318,
+      "loss": 1.2879,
+      "step": 1826
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019803047931645109,
+      "loss": 1.2323,
+      "step": 1827
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019802830602807348,
+      "loss": 1.425,
+      "step": 1828
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019802613155322653,
+      "loss": 1.2525,
+      "step": 1829
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019802395589193667,
+      "loss": 1.3682,
+      "step": 1830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019802177904423016,
+      "loss": 1.395,
+      "step": 1831
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001980196010101334,
+      "loss": 1.1672,
+      "step": 1832
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001980174217896727,
+      "loss": 1.2897,
+      "step": 1833
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019801524138287447,
+      "loss": 1.1673,
+      "step": 1834
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001980130597897651,
+      "loss": 1.2307,
+      "step": 1835
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019801087701037097,
+      "loss": 1.2062,
+      "step": 1836
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019800869304471852,
+      "loss": 1.1848,
+      "step": 1837
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019800650789283418,
+      "loss": 1.1778,
+      "step": 1838
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001980043215547444,
+      "loss": 1.2083,
+      "step": 1839
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001980021340304756,
+      "loss": 1.3953,
+      "step": 1840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019799994532005433,
+      "loss": 1.1791,
+      "step": 1841
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019799775542350703,
+      "loss": 1.2334,
+      "step": 1842
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019799556434086022,
+      "loss": 1.4113,
+      "step": 1843
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019799337207214042,
+      "loss": 1.4872,
+      "step": 1844
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019799117861737418,
+      "loss": 1.3198,
+      "step": 1845
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000197988983976588,
+      "loss": 1.3471,
+      "step": 1846
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019798678814980848,
+      "loss": 1.3882,
+      "step": 1847
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019798459113706219,
+      "loss": 1.2833,
+      "step": 1848
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019798239293837571,
+      "loss": 1.2498,
+      "step": 1849
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019798019355377568,
+      "loss": 1.358,
+      "step": 1850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001979779929832887,
+      "loss": 1.323,
+      "step": 1851
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019797579122694136,
+      "loss": 1.3035,
+      "step": 1852
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001979735882847604,
+      "loss": 1.2181,
+      "step": 1853
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019797138415677238,
+      "loss": 1.1979,
+      "step": 1854
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001979691788430041,
+      "loss": 1.3548,
+      "step": 1855
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019796697234348212,
+      "loss": 1.2362,
+      "step": 1856
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001979647646582332,
+      "loss": 1.4053,
+      "step": 1857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001979625557872841,
+      "loss": 1.3142,
+      "step": 1858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001979603457306615,
+      "loss": 1.2153,
+      "step": 1859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001979581344883922,
+      "loss": 1.157,
+      "step": 1860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001979559220605029,
+      "loss": 1.2994,
+      "step": 1861
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019795370844702043,
+      "loss": 1.3992,
+      "step": 1862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019795149364797157,
+      "loss": 1.3094,
+      "step": 1863
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019794927766338312,
+      "loss": 1.3274,
+      "step": 1864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001979470604932819,
+      "loss": 1.2617,
+      "step": 1865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019794484213769473,
+      "loss": 1.2471,
+      "step": 1866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001979426225966485,
+      "loss": 1.3801,
+      "step": 1867
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019794040187017005,
+      "loss": 1.2978,
+      "step": 1868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019793817995828622,
+      "loss": 1.3507,
+      "step": 1869
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019793595686102398,
+      "loss": 1.2832,
+      "step": 1870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019793373257841017,
+      "loss": 1.2416,
+      "step": 1871
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019793150711047174,
+      "loss": 1.2886,
+      "step": 1872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019792928045723565,
+      "loss": 1.2907,
+      "step": 1873
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019792705261872882,
+      "loss": 1.2196,
+      "step": 1874
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001979248235949782,
+      "loss": 1.1376,
+      "step": 1875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001979225933860108,
+      "loss": 1.3269,
+      "step": 1876
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019792036199185357,
+      "loss": 1.2553,
+      "step": 1877
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019791812941253358,
+      "loss": 1.0336,
+      "step": 1878
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019791589564807782,
+      "loss": 1.218,
+      "step": 1879
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019791366069851332,
+      "loss": 1.284,
+      "step": 1880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019791142456386713,
+      "loss": 1.2764,
+      "step": 1881
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019790918724416632,
+      "loss": 1.2328,
+      "step": 1882
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019790694873943795,
+      "loss": 1.2716,
+      "step": 1883
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019790470904970916,
+      "loss": 1.3621,
+      "step": 1884
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019790246817500703,
+      "loss": 1.1856,
+      "step": 1885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019790022611535868,
+      "loss": 1.3029,
+      "step": 1886
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019789798287079123,
+      "loss": 1.1386,
+      "step": 1887
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019789573844133186,
+      "loss": 1.2591,
+      "step": 1888
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019789349282700773,
+      "loss": 1.1566,
+      "step": 1889
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019789124602784602,
+      "loss": 1.2456,
+      "step": 1890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019788899804387388,
+      "loss": 1.4217,
+      "step": 1891
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001978867488751186,
+      "loss": 1.3218,
+      "step": 1892
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019788449852160733,
+      "loss": 1.3006,
+      "step": 1893
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019788224698336735,
+      "loss": 1.1134,
+      "step": 1894
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019787999426042592,
+      "loss": 1.3839,
+      "step": 1895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019787774035281023,
+      "loss": 1.3762,
+      "step": 1896
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019787548526054764,
+      "loss": 1.206,
+      "step": 1897
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019787322898366544,
+      "loss": 1.3181,
+      "step": 1898
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019787097152219085,
+      "loss": 1.4395,
+      "step": 1899
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001978687128761513,
+      "loss": 1.2829,
+      "step": 1900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019786645304557408,
+      "loss": 1.2893,
+      "step": 1901
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019786419203048655,
+      "loss": 1.2962,
+      "step": 1902
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019786192983091603,
+      "loss": 1.3886,
+      "step": 1903
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019785966644689,
+      "loss": 1.2419,
+      "step": 1904
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019785740187843573,
+      "loss": 1.1724,
+      "step": 1905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019785513612558076,
+      "loss": 1.1498,
+      "step": 1906
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001978528691883524,
+      "loss": 1.2432,
+      "step": 1907
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019785060106677818,
+      "loss": 1.1885,
+      "step": 1908
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019784833176088545,
+      "loss": 1.28,
+      "step": 1909
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019784606127070179,
+      "loss": 1.2826,
+      "step": 1910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001978437895962546,
+      "loss": 1.39,
+      "step": 1911
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019784151673757138,
+      "loss": 1.3445,
+      "step": 1912
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019783924269467968,
+      "loss": 1.1705,
+      "step": 1913
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000197836967467607,
+      "loss": 1.2457,
+      "step": 1914
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019783469105638085,
+      "loss": 1.3033,
+      "step": 1915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019783241346102884,
+      "loss": 1.2731,
+      "step": 1916
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019783013468157852,
+      "loss": 1.2676,
+      "step": 1917
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019782785471805743,
+      "loss": 1.4201,
+      "step": 1918
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001978255735704932,
+      "loss": 1.2846,
+      "step": 1919
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019782329123891346,
+      "loss": 1.3131,
+      "step": 1920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019782100772334577,
+      "loss": 1.2892,
+      "step": 1921
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019781872302381784,
+      "loss": 1.2596,
+      "step": 1922
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019781643714035724,
+      "loss": 1.2509,
+      "step": 1923
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019781415007299173,
+      "loss": 1.2379,
+      "step": 1924
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019781186182174894,
+      "loss": 1.3854,
+      "step": 1925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019780957238665656,
+      "loss": 1.1662,
+      "step": 1926
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019780728176774232,
+      "loss": 1.2627,
+      "step": 1927
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001978049899650339,
+      "loss": 1.2816,
+      "step": 1928
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001978026969785591,
+      "loss": 1.2857,
+      "step": 1929
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019780040280834564,
+      "loss": 1.2328,
+      "step": 1930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019779810745442132,
+      "loss": 1.1348,
+      "step": 1931
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019779581091681388,
+      "loss": 1.1863,
+      "step": 1932
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001977935131955511,
+      "loss": 1.3457,
+      "step": 1933
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019779121429066085,
+      "loss": 1.334,
+      "step": 1934
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019778891420217093,
+      "loss": 1.2454,
+      "step": 1935
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019778661293010913,
+      "loss": 1.3117,
+      "step": 1936
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019778431047450342,
+      "loss": 1.3627,
+      "step": 1937
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019778200683538154,
+      "loss": 1.192,
+      "step": 1938
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019777970201277143,
+      "loss": 1.1627,
+      "step": 1939
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000197777396006701,
+      "loss": 1.507,
+      "step": 1940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019777508881719814,
+      "loss": 1.1724,
+      "step": 1941
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019777278044429077,
+      "loss": 1.2008,
+      "step": 1942
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001977704708880068,
+      "loss": 1.164,
+      "step": 1943
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001977681601483743,
+      "loss": 1.2942,
+      "step": 1944
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019776584822542108,
+      "loss": 1.158,
+      "step": 1945
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019776353511917526,
+      "loss": 1.3325,
+      "step": 1946
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019776122082966474,
+      "loss": 1.2775,
+      "step": 1947
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019775890535691755,
+      "loss": 1.2982,
+      "step": 1948
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019775658870096175,
+      "loss": 1.3579,
+      "step": 1949
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001977542708618254,
+      "loss": 1.3304,
+      "step": 1950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019775195183953644,
+      "loss": 1.2551,
+      "step": 1951
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019774963163412306,
+      "loss": 1.3079,
+      "step": 1952
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001977473102456133,
+      "loss": 1.3124,
+      "step": 1953
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019774498767403522,
+      "loss": 1.1526,
+      "step": 1954
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019774266391941697,
+      "loss": 1.2683,
+      "step": 1955
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019774033898178667,
+      "loss": 1.3403,
+      "step": 1956
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019773801286117245,
+      "loss": 1.3242,
+      "step": 1957
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019773568555760248,
+      "loss": 1.3161,
+      "step": 1958
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019773335707110493,
+      "loss": 1.1387,
+      "step": 1959
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019773102740170795,
+      "loss": 1.3246,
+      "step": 1960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019772869654943975,
+      "loss": 1.1735,
+      "step": 1961
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019772636451432856,
+      "loss": 1.2273,
+      "step": 1962
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001977240312964026,
+      "loss": 1.1524,
+      "step": 1963
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001977216968956901,
+      "loss": 1.3522,
+      "step": 1964
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019771936131221927,
+      "loss": 1.3521,
+      "step": 1965
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001977170245460185,
+      "loss": 1.2165,
+      "step": 1966
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019771468659711595,
+      "loss": 1.275,
+      "step": 1967
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019771234746554,
+      "loss": 1.089,
+      "step": 1968
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001977100071513189,
+      "loss": 1.223,
+      "step": 1969
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000197707665654481,
+      "loss": 1.3271,
+      "step": 1970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019770532297505467,
+      "loss": 1.1669,
+      "step": 1971
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019770297911306822,
+      "loss": 1.3045,
+      "step": 1972
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019770063406855,
+      "loss": 1.2376,
+      "step": 1973
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019769828784152848,
+      "loss": 1.2806,
+      "step": 1974
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019769594043203198,
+      "loss": 1.3596,
+      "step": 1975
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019769359184008893,
+      "loss": 1.1551,
+      "step": 1976
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001976912420657278,
+      "loss": 1.374,
+      "step": 1977
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019768889110897695,
+      "loss": 1.2285,
+      "step": 1978
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019768653896986492,
+      "loss": 1.3297,
+      "step": 1979
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019768418564842009,
+      "loss": 1.3128,
+      "step": 1980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000197681831144671,
+      "loss": 1.3029,
+      "step": 1981
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019767947545864614,
+      "loss": 1.154,
+      "step": 1982
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000197677118590374,
+      "loss": 1.2105,
+      "step": 1983
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019767476053988317,
+      "loss": 1.3421,
+      "step": 1984
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001976724013072021,
+      "loss": 1.2478,
+      "step": 1985
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001976700408923594,
+      "loss": 1.1307,
+      "step": 1986
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019766767929538365,
+      "loss": 1.3441,
+      "step": 1987
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019766531651630337,
+      "loss": 1.3121,
+      "step": 1988
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001976629525551472,
+      "loss": 1.2634,
+      "step": 1989
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001976605874119438,
+      "loss": 1.2733,
+      "step": 1990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001976582210867217,
+      "loss": 1.4118,
+      "step": 1991
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001976558535795096,
+      "loss": 1.2292,
+      "step": 1992
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019765348489033616,
+      "loss": 1.3097,
+      "step": 1993
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019765111501923,
+      "loss": 1.3654,
+      "step": 1994
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019764874396621985,
+      "loss": 1.3863,
+      "step": 1995
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019764637173133438,
+      "loss": 1.3233,
+      "step": 1996
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001976439983146023,
+      "loss": 1.2364,
+      "step": 1997
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019764162371605236,
+      "loss": 1.2273,
+      "step": 1998
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019763924793571332,
+      "loss": 1.3451,
+      "step": 1999
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019763687097361387,
+      "loss": 1.2194,
+      "step": 2000
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 1.265727162361145,
+      "eval_runtime": 847.8628,
+      "eval_samples_per_second": 28.377,
+      "eval_steps_per_second": 14.189,
+      "step": 2000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019763449282978281,
+      "loss": 1.1173,
+      "step": 2001
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019763211350424896,
+      "loss": 1.3483,
+      "step": 2002
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019762973299704105,
+      "loss": 1.1972,
+      "step": 2003
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019762735130818794,
+      "loss": 1.2235,
+      "step": 2004
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019762496843771845,
+      "loss": 1.2873,
+      "step": 2005
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019762258438566142,
+      "loss": 1.3267,
+      "step": 2006
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019762019915204568,
+      "loss": 1.3506,
+      "step": 2007
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019761781273690013,
+      "loss": 1.2776,
+      "step": 2008
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019761542514025365,
+      "loss": 1.1562,
+      "step": 2009
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001976130363621351,
+      "loss": 1.2807,
+      "step": 2010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019761064640257345,
+      "loss": 1.2277,
+      "step": 2011
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019760825526159758,
+      "loss": 1.1403,
+      "step": 2012
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019760586293923646,
+      "loss": 1.1847,
+      "step": 2013
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019760346943551902,
+      "loss": 1.2236,
+      "step": 2014
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019760107475047428,
+      "loss": 1.288,
+      "step": 2015
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019759867888413115,
+      "loss": 1.2327,
+      "step": 2016
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001975962818365187,
+      "loss": 1.1732,
+      "step": 2017
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019759388360766585,
+      "loss": 1.2608,
+      "step": 2018
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019759148419760174,
+      "loss": 1.2981,
+      "step": 2019
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019758908360635534,
+      "loss": 1.2918,
+      "step": 2020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001975866818339557,
+      "loss": 1.2011,
+      "step": 2021
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019758427888043193,
+      "loss": 1.2025,
+      "step": 2022
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001975818747458131,
+      "loss": 1.2526,
+      "step": 2023
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019757946943012826,
+      "loss": 1.2389,
+      "step": 2024
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019757706293340658,
+      "loss": 1.3148,
+      "step": 2025
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001975746552556772,
+      "loss": 1.2478,
+      "step": 2026
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001975722463969692,
+      "loss": 1.3301,
+      "step": 2027
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019756983635731178,
+      "loss": 1.244,
+      "step": 2028
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019756742513673409,
+      "loss": 1.2853,
+      "step": 2029
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019756501273526534,
+      "loss": 1.3357,
+      "step": 2030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019756259915293468,
+      "loss": 1.3493,
+      "step": 2031
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019756018438977138,
+      "loss": 1.2814,
+      "step": 2032
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019755776844580463,
+      "loss": 1.2937,
+      "step": 2033
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019755535132106363,
+      "loss": 1.313,
+      "step": 2034
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019755293301557773,
+      "loss": 1.2769,
+      "step": 2035
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019755051352937617,
+      "loss": 1.2318,
+      "step": 2036
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019754809286248815,
+      "loss": 1.4377,
+      "step": 2037
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001975456710149431,
+      "loss": 1.3342,
+      "step": 2038
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019754324798677025,
+      "loss": 1.3234,
+      "step": 2039
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019754082377799895,
+      "loss": 1.1599,
+      "step": 2040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019753839838865852,
+      "loss": 1.1641,
+      "step": 2041
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019753597181877832,
+      "loss": 1.2866,
+      "step": 2042
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019753354406838773,
+      "loss": 1.2025,
+      "step": 2043
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019753111513751614,
+      "loss": 1.3079,
+      "step": 2044
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019752868502619295,
+      "loss": 1.2474,
+      "step": 2045
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019752625373444755,
+      "loss": 1.2814,
+      "step": 2046
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001975238212623094,
+      "loss": 1.2401,
+      "step": 2047
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019752138760980792,
+      "loss": 1.1564,
+      "step": 2048
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019751895277697254,
+      "loss": 1.1456,
+      "step": 2049
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019751651676383277,
+      "loss": 1.2969,
+      "step": 2050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001975140795704181,
+      "loss": 1.2753,
+      "step": 2051
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000197511641196758,
+      "loss": 1.431,
+      "step": 2052
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019750920164288198,
+      "loss": 1.142,
+      "step": 2053
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019750676090881958,
+      "loss": 1.3086,
+      "step": 2054
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019750431899460036,
+      "loss": 1.303,
+      "step": 2055
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019750187590025385,
+      "loss": 1.2767,
+      "step": 2056
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001974994316258096,
+      "loss": 1.3056,
+      "step": 2057
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019749698617129723,
+      "loss": 1.0539,
+      "step": 2058
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019749453953674632,
+      "loss": 1.3441,
+      "step": 2059
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001974920917221865,
+      "loss": 1.3575,
+      "step": 2060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019748964272764738,
+      "loss": 1.3084,
+      "step": 2061
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001974871925531586,
+      "loss": 1.1228,
+      "step": 2062
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019748474119874983,
+      "loss": 1.3448,
+      "step": 2063
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019748228866445073,
+      "loss": 1.1227,
+      "step": 2064
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019747983495029098,
+      "loss": 1.1579,
+      "step": 2065
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001974773800563003,
+      "loss": 1.3284,
+      "step": 2066
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019747492398250837,
+      "loss": 1.1585,
+      "step": 2067
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019747246672894493,
+      "loss": 1.237,
+      "step": 2068
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019747000829563972,
+      "loss": 1.3229,
+      "step": 2069
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019746754868262253,
+      "loss": 1.2458,
+      "step": 2070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019746508788992306,
+      "loss": 1.3669,
+      "step": 2071
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019746262591757114,
+      "loss": 1.2738,
+      "step": 2072
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019746016276559657,
+      "loss": 1.2091,
+      "step": 2073
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019745769843402915,
+      "loss": 1.2716,
+      "step": 2074
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001974552329228987,
+      "loss": 1.2952,
+      "step": 2075
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019745276623223508,
+      "loss": 1.2305,
+      "step": 2076
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019745029836206813,
+      "loss": 1.2743,
+      "step": 2077
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019744782931242774,
+      "loss": 1.3914,
+      "step": 2078
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019744535908334374,
+      "loss": 1.2424,
+      "step": 2079
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001974428876748461,
+      "loss": 1.2274,
+      "step": 2080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019744041508696467,
+      "loss": 1.4854,
+      "step": 2081
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019743794131972943,
+      "loss": 1.2406,
+      "step": 2082
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001974354663731703,
+      "loss": 1.2322,
+      "step": 2083
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019743299024731719,
+      "loss": 1.2697,
+      "step": 2084
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019743051294220015,
+      "loss": 1.16,
+      "step": 2085
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001974280344578491,
+      "loss": 1.252,
+      "step": 2086
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019742555479429407,
+      "loss": 1.2884,
+      "step": 2087
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019742307395156507,
+      "loss": 1.1933,
+      "step": 2088
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001974205919296921,
+      "loss": 1.2152,
+      "step": 2089
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019741810872870525,
+      "loss": 1.2304,
+      "step": 2090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019741562434863453,
+      "loss": 1.2834,
+      "step": 2091
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019741313878951004,
+      "loss": 1.35,
+      "step": 2092
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019741065205136185,
+      "loss": 1.2547,
+      "step": 2093
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019740816413422004,
+      "loss": 1.2837,
+      "step": 2094
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001974056750381147,
+      "loss": 1.2672,
+      "step": 2095
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019740318476307603,
+      "loss": 1.1754,
+      "step": 2096
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019740069330913417,
+      "loss": 1.1071,
+      "step": 2097
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019739820067631923,
+      "loss": 1.1341,
+      "step": 2098
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019739570686466134,
+      "loss": 1.4716,
+      "step": 2099
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001973932118741908,
+      "loss": 1.3136,
+      "step": 2100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019739071570493768,
+      "loss": 1.2773,
+      "step": 2101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001973882183569323,
+      "loss": 1.319,
+      "step": 2102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001973857198302048,
+      "loss": 1.1439,
+      "step": 2103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019738322012478548,
+      "loss": 1.2609,
+      "step": 2104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019738071924070458,
+      "loss": 1.22,
+      "step": 2105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019737821717799237,
+      "loss": 1.3524,
+      "step": 2106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001973757139366791,
+      "loss": 1.1513,
+      "step": 2107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001973732095167951,
+      "loss": 1.3142,
+      "step": 2108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019737070391837068,
+      "loss": 1.2797,
+      "step": 2109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001973681971414362,
+      "loss": 1.3009,
+      "step": 2110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001973656891860219,
+      "loss": 1.2256,
+      "step": 2111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019736318005215823,
+      "loss": 1.1413,
+      "step": 2112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019736066973987553,
+      "loss": 1.2128,
+      "step": 2113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019735815824920415,
+      "loss": 1.3217,
+      "step": 2114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019735564558017457,
+      "loss": 1.1531,
+      "step": 2115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001973531317328171,
+      "loss": 1.2615,
+      "step": 2116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019735061670716225,
+      "loss": 1.1102,
+      "step": 2117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001973481005032404,
+      "loss": 1.2023,
+      "step": 2118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019734558312108202,
+      "loss": 1.3026,
+      "step": 2119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001973430645607176,
+      "loss": 1.2111,
+      "step": 2120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019734054482217764,
+      "loss": 1.1482,
+      "step": 2121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019733802390549255,
+      "loss": 1.2417,
+      "step": 2122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019733550181069294,
+      "loss": 1.255,
+      "step": 2123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019733297853780928,
+      "loss": 1.2938,
+      "step": 2124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019733045408687212,
+      "loss": 1.2774,
+      "step": 2125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019732792845791202,
+      "loss": 1.3917,
+      "step": 2126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019732540165095957,
+      "loss": 1.1729,
+      "step": 2127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019732287366604534,
+      "loss": 1.17,
+      "step": 2128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019732034450319988,
+      "loss": 1.2268,
+      "step": 2129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019731781416245384,
+      "loss": 1.2448,
+      "step": 2130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019731528264383788,
+      "loss": 1.197,
+      "step": 2131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001973127499473826,
+      "loss": 1.2937,
+      "step": 2132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019731021607311862,
+      "loss": 1.3495,
+      "step": 2133
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019730768102107666,
+      "loss": 1.0909,
+      "step": 2134
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019730514479128738,
+      "loss": 1.1926,
+      "step": 2135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001973026073837815,
+      "loss": 1.2975,
+      "step": 2136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019730006879858976,
+      "loss": 1.2416,
+      "step": 2137
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001972975290357428,
+      "loss": 1.1544,
+      "step": 2138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001972949880952714,
+      "loss": 1.1879,
+      "step": 2139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019729244597720632,
+      "loss": 1.282,
+      "step": 2140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019728990268157832,
+      "loss": 1.2319,
+      "step": 2141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001972873582084182,
+      "loss": 1.2613,
+      "step": 2142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019728481255775672,
+      "loss": 1.1501,
+      "step": 2143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019728226572962473,
+      "loss": 1.2203,
+      "step": 2144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019727971772405306,
+      "loss": 1.3124,
+      "step": 2145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001972771685410725,
+      "loss": 1.2588,
+      "step": 2146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019727461818071394,
+      "loss": 1.2206,
+      "step": 2147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019727206664300824,
+      "loss": 1.2671,
+      "step": 2148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019726951392798627,
+      "loss": 1.2401,
+      "step": 2149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019726696003567894,
+      "loss": 1.1124,
+      "step": 2150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001972644049661172,
+      "loss": 1.3418,
+      "step": 2151
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019726184871933188,
+      "loss": 1.2498,
+      "step": 2152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000197259291295354,
+      "loss": 1.387,
+      "step": 2153
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019725673269421447,
+      "loss": 1.3037,
+      "step": 2154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001972541729159443,
+      "loss": 1.1846,
+      "step": 2155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019725161196057443,
+      "loss": 1.3776,
+      "step": 2156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019724904982813587,
+      "loss": 1.1022,
+      "step": 2157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019724648651865966,
+      "loss": 1.2601,
+      "step": 2158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019724392203217676,
+      "loss": 1.2307,
+      "step": 2159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019724135636871824,
+      "loss": 1.2356,
+      "step": 2160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001972387895283152,
+      "loss": 1.3646,
+      "step": 2161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019723622151099864,
+      "loss": 1.3147,
+      "step": 2162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019723365231679967,
+      "loss": 1.3939,
+      "step": 2163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019723108194574936,
+      "loss": 1.1676,
+      "step": 2164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019722851039787888,
+      "loss": 1.3384,
+      "step": 2165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019722593767321933,
+      "loss": 1.2986,
+      "step": 2166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001972233637718018,
+      "loss": 1.3701,
+      "step": 2167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001972207886936575,
+      "loss": 1.1944,
+      "step": 2168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019721821243881755,
+      "loss": 1.137,
+      "step": 2169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019721563500731319,
+      "loss": 1.1889,
+      "step": 2170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019721305639917554,
+      "loss": 1.1773,
+      "step": 2171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001972104766144359,
+      "loss": 1.3254,
+      "step": 2172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019720789565312542,
+      "loss": 1.2107,
+      "step": 2173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001972053135152754,
+      "loss": 1.3676,
+      "step": 2174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019720273020091698,
+      "loss": 1.3698,
+      "step": 2175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019720014571008158,
+      "loss": 1.2649,
+      "step": 2176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019719756004280039,
+      "loss": 1.314,
+      "step": 2177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001971949731991047,
+      "loss": 1.2545,
+      "step": 2178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019719238517902586,
+      "loss": 1.2247,
+      "step": 2179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019718979598259514,
+      "loss": 1.192,
+      "step": 2180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019718720560984394,
+      "loss": 1.2137,
+      "step": 2181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019718461406080358,
+      "loss": 1.2563,
+      "step": 2182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019718202133550542,
+      "loss": 1.3271,
+      "step": 2183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001971794274339809,
+      "loss": 1.2985,
+      "step": 2184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019717683235626128,
+      "loss": 1.2002,
+      "step": 2185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019717423610237812,
+      "loss": 1.2837,
+      "step": 2186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019717163867236272,
+      "loss": 1.1078,
+      "step": 2187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019716904006624662,
+      "loss": 1.2959,
+      "step": 2188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019716644028406122,
+      "loss": 1.3006,
+      "step": 2189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000197163839325838,
+      "loss": 1.2104,
+      "step": 2190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019716123719160843,
+      "loss": 1.2826,
+      "step": 2191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019715863388140398,
+      "loss": 1.1502,
+      "step": 2192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001971560293952562,
+      "loss": 1.131,
+      "step": 2193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019715342373319663,
+      "loss": 1.2412,
+      "step": 2194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019715081689525676,
+      "loss": 1.3527,
+      "step": 2195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019714820888146816,
+      "loss": 1.3254,
+      "step": 2196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019714559969186238,
+      "loss": 1.2042,
+      "step": 2197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019714298932647098,
+      "loss": 1.2103,
+      "step": 2198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019714037778532565,
+      "loss": 1.1553,
+      "step": 2199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019713776506845792,
+      "loss": 1.2434,
+      "step": 2200
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 1.2296520471572876,
+      "eval_runtime": 847.4415,
+      "eval_samples_per_second": 28.391,
+      "eval_steps_per_second": 14.196,
+      "step": 2200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001971351511758994,
+      "loss": 1.0597,
+      "step": 2201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001971325361076818,
+      "loss": 1.2946,
+      "step": 2202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001971299198638367,
+      "loss": 1.0715,
+      "step": 2203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019712730244439582,
+      "loss": 1.2457,
+      "step": 2204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019712468384939076,
+      "loss": 1.2244,
+      "step": 2205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019712206407885332,
+      "loss": 1.1929,
+      "step": 2206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019711944313281513,
+      "loss": 1.2064,
+      "step": 2207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001971168210113079,
+      "loss": 1.3525,
+      "step": 2208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019711419771436343,
+      "loss": 1.2328,
+      "step": 2209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019711157324201346,
+      "loss": 1.2284,
+      "step": 2210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019710894759428972,
+      "loss": 1.2729,
+      "step": 2211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019710632077122402,
+      "loss": 1.2964,
+      "step": 2212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019710369277284813,
+      "loss": 1.3251,
+      "step": 2213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019710106359919385,
+      "loss": 1.399,
+      "step": 2214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019709843325029305,
+      "loss": 1.1675,
+      "step": 2215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019709580172617752,
+      "loss": 1.1382,
+      "step": 2216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019709316902687913,
+      "loss": 1.3137,
+      "step": 2217
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019709053515242972,
+      "loss": 1.2957,
+      "step": 2218
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001970879001028612,
+      "loss": 1.0388,
+      "step": 2219
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019708526387820547,
+      "loss": 1.1814,
+      "step": 2220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019708262647849441,
+      "loss": 1.2956,
+      "step": 2221
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019707998790375992,
+      "loss": 1.2293,
+      "step": 2222
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019707734815403401,
+      "loss": 1.1661,
+      "step": 2223
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019707470722934853,
+      "loss": 1.3241,
+      "step": 2224
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019707206512973552,
+      "loss": 1.0774,
+      "step": 2225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019706942185522696,
+      "loss": 1.1491,
+      "step": 2226
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001970667774058548,
+      "loss": 1.2843,
+      "step": 2227
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001970641317816511,
+      "loss": 1.1331,
+      "step": 2228
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001970614849826478,
+      "loss": 1.3136,
+      "step": 2229
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000197058837008877,
+      "loss": 1.1934,
+      "step": 2230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019705618786037074,
+      "loss": 1.1964,
+      "step": 2231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019705353753716106,
+      "loss": 1.3267,
+      "step": 2232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019705088603928008,
+      "loss": 1.2088,
+      "step": 2233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019704823336675985,
+      "loss": 1.2002,
+      "step": 2234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019704557951963247,
+      "loss": 1.1677,
+      "step": 2235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019704292449793012,
+      "loss": 1.3227,
+      "step": 2236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019704026830168485,
+      "loss": 1.2827,
+      "step": 2237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001970376109309289,
+      "loss": 1.207,
+      "step": 2238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019703495238569435,
+      "loss": 1.3523,
+      "step": 2239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019703229266601345,
+      "loss": 1.262,
+      "step": 2240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019702963177191832,
+      "loss": 1.2286,
+      "step": 2241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001970269697034412,
+      "loss": 1.1431,
+      "step": 2242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019702430646061436,
+      "loss": 1.3312,
+      "step": 2243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019702164204346995,
+      "loss": 1.3341,
+      "step": 2244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019701897645204027,
+      "loss": 1.2541,
+      "step": 2245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019701630968635756,
+      "loss": 1.1945,
+      "step": 2246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019701364174645408,
+      "loss": 1.2118,
+      "step": 2247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001970109726323622,
+      "loss": 1.1869,
+      "step": 2248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019700830234411414,
+      "loss": 1.164,
+      "step": 2249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019700563088174223,
+      "loss": 1.248,
+      "step": 2250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019700295824527886,
+      "loss": 1.3045,
+      "step": 2251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001970002844347563,
+      "loss": 1.2368,
+      "step": 2252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000196997609450207,
+      "loss": 1.0304,
+      "step": 2253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019699493329166322,
+      "loss": 1.1888,
+      "step": 2254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001969922559591575,
+      "loss": 1.2429,
+      "step": 2255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019698957745272212,
+      "loss": 1.191,
+      "step": 2256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019698689777238954,
+      "loss": 1.3002,
+      "step": 2257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001969842169181922,
+      "loss": 1.0969,
+      "step": 2258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019698153489016256,
+      "loss": 1.3715,
+      "step": 2259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019697885168833306,
+      "loss": 1.2773,
+      "step": 2260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019697616731273617,
+      "loss": 1.0582,
+      "step": 2261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001969734817634044,
+      "loss": 1.3486,
+      "step": 2262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019697079504037021,
+      "loss": 1.2355,
+      "step": 2263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001969681071436662,
+      "loss": 1.2366,
+      "step": 2264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019696541807332483,
+      "loss": 1.1761,
+      "step": 2265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019696272782937868,
+      "loss": 1.2072,
+      "step": 2266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019696003641186033,
+      "loss": 1.2826,
+      "step": 2267
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019695734382080228,
+      "loss": 1.1745,
+      "step": 2268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001969546500562372,
+      "loss": 1.2046,
+      "step": 2269
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019695195511819762,
+      "loss": 1.2275,
+      "step": 2270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019694925900671625,
+      "loss": 1.1627,
+      "step": 2271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001969465617218256,
+      "loss": 1.2246,
+      "step": 2272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019694386326355847,
+      "loss": 1.2532,
+      "step": 2273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019694116363194739,
+      "loss": 1.1642,
+      "step": 2274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001969384628270251,
+      "loss": 1.1621,
+      "step": 2275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019693576084882426,
+      "loss": 1.4628,
+      "step": 2276
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019693305769737758,
+      "loss": 1.3343,
+      "step": 2277
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019693035337271778,
+      "loss": 1.072,
+      "step": 2278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019692764787487761,
+      "loss": 1.2495,
+      "step": 2279
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019692494120388977,
+      "loss": 1.2905,
+      "step": 2280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019692223335978704,
+      "loss": 1.2081,
+      "step": 2281
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019691952434260226,
+      "loss": 1.1841,
+      "step": 2282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001969168141523681,
+      "loss": 1.1196,
+      "step": 2283
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019691410278911743,
+      "loss": 1.2231,
+      "step": 2284
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001969113902528831,
+      "loss": 1.1277,
+      "step": 2285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019690867654369786,
+      "loss": 1.2394,
+      "step": 2286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001969059616615946,
+      "loss": 1.1817,
+      "step": 2287
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001969032456066062,
+      "loss": 1.1776,
+      "step": 2288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019690052837876548,
+      "loss": 1.1534,
+      "step": 2289
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019689780997810533,
+      "loss": 1.2329,
+      "step": 2290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001968950904046587,
+      "loss": 1.1466,
+      "step": 2291
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001968923696584585,
+      "loss": 1.2082,
+      "step": 2292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019688964773953763,
+      "loss": 1.2526,
+      "step": 2293
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000196886924647929,
+      "loss": 1.1928,
+      "step": 2294
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001968842003836657,
+      "loss": 1.224,
+      "step": 2295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019688147494678054,
+      "loss": 1.1131,
+      "step": 2296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001968787483373066,
+      "loss": 1.2336,
+      "step": 2297
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001968760205552769,
+      "loss": 1.1038,
+      "step": 2298
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001968732916007244,
+      "loss": 1.2677,
+      "step": 2299
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019687056147368214,
+      "loss": 1.2575,
+      "step": 2300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019686783017418318,
+      "loss": 1.3084,
+      "step": 2301
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019686509770226057,
+      "loss": 1.0924,
+      "step": 2302
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001968623640579474,
+      "loss": 1.2799,
+      "step": 2303
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019685962924127673,
+      "loss": 1.2651,
+      "step": 2304
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019685689325228163,
+      "loss": 1.1498,
+      "step": 2305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001968541560909953,
+      "loss": 1.2193,
+      "step": 2306
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001968514177574508,
+      "loss": 1.179,
+      "step": 2307
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001968486782516813,
+      "loss": 1.0958,
+      "step": 2308
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019684593757371996,
+      "loss": 1.1915,
+      "step": 2309
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019684319572359994,
+      "loss": 1.2011,
+      "step": 2310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019684045270135445,
+      "loss": 1.3506,
+      "step": 2311
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019683770850701665,
+      "loss": 1.0382,
+      "step": 2312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019683496314061976,
+      "loss": 1.2276,
+      "step": 2313
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019683221660219704,
+      "loss": 1.1833,
+      "step": 2314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019682946889178172,
+      "loss": 1.2752,
+      "step": 2315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019682672000940704,
+      "loss": 1.2472,
+      "step": 2316
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019682396995510627,
+      "loss": 1.1584,
+      "step": 2317
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001968212187289127,
+      "loss": 1.3133,
+      "step": 2318
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019681846633085967,
+      "loss": 1.2607,
+      "step": 2319
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019681571276098042,
+      "loss": 1.3651,
+      "step": 2320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019681295801930837,
+      "loss": 1.2148,
+      "step": 2321
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019681020210587677,
+      "loss": 1.1649,
+      "step": 2322
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000196807445020719,
+      "loss": 1.3608,
+      "step": 2323
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019680468676386848,
+      "loss": 1.1512,
+      "step": 2324
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019680192733535852,
+      "loss": 1.1897,
+      "step": 2325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019679916673522255,
+      "loss": 1.1955,
+      "step": 2326
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000196796404963494,
+      "loss": 1.2406,
+      "step": 2327
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001967936420202063,
+      "loss": 1.4019,
+      "step": 2328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019679087790539285,
+      "loss": 1.197,
+      "step": 2329
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019678811261908717,
+      "loss": 1.1434,
+      "step": 2330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019678534616132266,
+      "loss": 1.1644,
+      "step": 2331
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019678257853213285,
+      "loss": 1.298,
+      "step": 2332
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019677980973155119,
+      "loss": 1.331,
+      "step": 2333
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019677703975961124,
+      "loss": 1.2838,
+      "step": 2334
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001967742686163465,
+      "loss": 1.0399,
+      "step": 2335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001967714963017905,
+      "loss": 1.1809,
+      "step": 2336
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019676872281597686,
+      "loss": 1.1013,
+      "step": 2337
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019676594815893906,
+      "loss": 1.2169,
+      "step": 2338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019676317233071076,
+      "loss": 1.1086,
+      "step": 2339
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019676039533132546,
+      "loss": 1.1507,
+      "step": 2340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019675761716081688,
+      "loss": 1.2544,
+      "step": 2341
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001967548378192186,
+      "loss": 1.3044,
+      "step": 2342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019675205730656421,
+      "loss": 1.2222,
+      "step": 2343
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019674927562288747,
+      "loss": 1.3638,
+      "step": 2344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019674649276822193,
+      "loss": 1.2912,
+      "step": 2345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019674370874260135,
+      "loss": 1.138,
+      "step": 2346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019674092354605942,
+      "loss": 1.2616,
+      "step": 2347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001967381371786298,
+      "loss": 1.3453,
+      "step": 2348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019673534964034626,
+      "loss": 1.3206,
+      "step": 2349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019673256093124254,
+      "loss": 1.1878,
+      "step": 2350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019672977105135238,
+      "loss": 1.1783,
+      "step": 2351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019672698000070953,
+      "loss": 1.2277,
+      "step": 2352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001967241877793478,
+      "loss": 1.2596,
+      "step": 2353
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019672139438730096,
+      "loss": 1.2057,
+      "step": 2354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019671859982460285,
+      "loss": 1.1845,
+      "step": 2355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019671580409128726,
+      "loss": 1.0987,
+      "step": 2356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019671300718738805,
+      "loss": 1.189,
+      "step": 2357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019671020911293907,
+      "loss": 1.2864,
+      "step": 2358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019670740986797418,
+      "loss": 1.1087,
+      "step": 2359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019670460945252726,
+      "loss": 1.351,
+      "step": 2360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001967018078666322,
+      "loss": 1.1311,
+      "step": 2361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019669900511032294,
+      "loss": 1.0734,
+      "step": 2362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019669620118363334,
+      "loss": 1.2477,
+      "step": 2363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001966933960865974,
+      "loss": 1.2039,
+      "step": 2364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000196690589819249,
+      "loss": 1.3293,
+      "step": 2365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019668778238162219,
+      "loss": 1.3318,
+      "step": 2366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001966849737737509,
+      "loss": 1.3243,
+      "step": 2367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019668216399566915,
+      "loss": 1.1944,
+      "step": 2368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019667935304741092,
+      "loss": 1.1097,
+      "step": 2369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019667654092901023,
+      "loss": 1.2099,
+      "step": 2370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019667372764050112,
+      "loss": 1.0661,
+      "step": 2371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019667091318191767,
+      "loss": 1.2857,
+      "step": 2372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019666809755329392,
+      "loss": 1.0982,
+      "step": 2373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019666528075466396,
+      "loss": 1.155,
+      "step": 2374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019666246278606187,
+      "loss": 1.2419,
+      "step": 2375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019665964364752174,
+      "loss": 1.2874,
+      "step": 2376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001966568233390777,
+      "loss": 1.2783,
+      "step": 2377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019665400186076392,
+      "loss": 1.2374,
+      "step": 2378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019665117921261453,
+      "loss": 1.3279,
+      "step": 2379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001966483553946637,
+      "loss": 1.3098,
+      "step": 2380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001966455304069456,
+      "loss": 1.3417,
+      "step": 2381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001966427042494944,
+      "loss": 1.2148,
+      "step": 2382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001966398769223443,
+      "loss": 1.2762,
+      "step": 2383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019663704842552959,
+      "loss": 1.2259,
+      "step": 2384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019663421875908444,
+      "loss": 1.2648,
+      "step": 2385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019663138792304312,
+      "loss": 1.2081,
+      "step": 2386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001966285559174399,
+      "loss": 1.1219,
+      "step": 2387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019662572274230903,
+      "loss": 1.2634,
+      "step": 2388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019662288839768485,
+      "loss": 1.1795,
+      "step": 2389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001966200528836016,
+      "loss": 1.1957,
+      "step": 2390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019661721620009364,
+      "loss": 1.0982,
+      "step": 2391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001966143783471953,
+      "loss": 1.1814,
+      "step": 2392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019661153932494093,
+      "loss": 1.2612,
+      "step": 2393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019660869913336486,
+      "loss": 1.2559,
+      "step": 2394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001966058577725015,
+      "loss": 1.2456,
+      "step": 2395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019660301524238524,
+      "loss": 1.1779,
+      "step": 2396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019660017154305047,
+      "loss": 1.2368,
+      "step": 2397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019659732667453162,
+      "loss": 1.2174,
+      "step": 2398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001965944806368631,
+      "loss": 1.1374,
+      "step": 2399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001965916334300794,
+      "loss": 1.2037,
+      "step": 2400
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 1.198292851448059,
+      "eval_runtime": 854.574,
+      "eval_samples_per_second": 28.154,
+      "eval_steps_per_second": 14.077,
+      "step": 2400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001965887850542149,
+      "loss": 1.2365,
+      "step": 2401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019658593550930415,
+      "loss": 1.4205,
+      "step": 2402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001965830847953816,
+      "loss": 1.3176,
+      "step": 2403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019658023291248183,
+      "loss": 1.2039,
+      "step": 2404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019657737986063925,
+      "loss": 1.2114,
+      "step": 2405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019657452563988845,
+      "loss": 1.2833,
+      "step": 2406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019657167025026394,
+      "loss": 1.2736,
+      "step": 2407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019656881369180035,
+      "loss": 1.2372,
+      "step": 2408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019656595596453214,
+      "loss": 1.131,
+      "step": 2409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000196563097068494,
+      "loss": 1.0472,
+      "step": 2410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019656023700372052,
+      "loss": 1.2018,
+      "step": 2411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019655737577024625,
+      "loss": 1.17,
+      "step": 2412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019655451336810594,
+      "loss": 1.2186,
+      "step": 2413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001965516497973341,
+      "loss": 1.1777,
+      "step": 2414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019654878505796544,
+      "loss": 1.1654,
+      "step": 2415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019654591915003463,
+      "loss": 1.3502,
+      "step": 2416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001965430520735764,
+      "loss": 1.1731,
+      "step": 2417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019654018382862543,
+      "loss": 1.3358,
+      "step": 2418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001965373144152164,
+      "loss": 1.2187,
+      "step": 2419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019653444383338408,
+      "loss": 1.1249,
+      "step": 2420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019653157208316318,
+      "loss": 1.0219,
+      "step": 2421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019652869916458846,
+      "loss": 1.2397,
+      "step": 2422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019652582507769472,
+      "loss": 1.1276,
+      "step": 2423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019652294982251675,
+      "loss": 1.1415,
+      "step": 2424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019652007339908932,
+      "loss": 1.2827,
+      "step": 2425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019651719580744724,
+      "loss": 1.3574,
+      "step": 2426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019651431704762539,
+      "loss": 1.1901,
+      "step": 2427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019651143711965857,
+      "loss": 1.3041,
+      "step": 2428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001965085560235816,
+      "loss": 1.1854,
+      "step": 2429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019650567375942946,
+      "loss": 1.1111,
+      "step": 2430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019650279032723696,
+      "loss": 1.1601,
+      "step": 2431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000196499905727039,
+      "loss": 1.1335,
+      "step": 2432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019649701995887048,
+      "loss": 1.182,
+      "step": 2433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019649413302276635,
+      "loss": 1.1892,
+      "step": 2434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001964912449187616,
+      "loss": 1.1021,
+      "step": 2435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001964883556468911,
+      "loss": 1.274,
+      "step": 2436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019648546520718987,
+      "loss": 1.1693,
+      "step": 2437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019648257359969286,
+      "loss": 1.1931,
+      "step": 2438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001964796808244351,
+      "loss": 1.2704,
+      "step": 2439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001964767868814516,
+      "loss": 1.1382,
+      "step": 2440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001964738917707774,
+      "loss": 1.1799,
+      "step": 2441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019647099549244748,
+      "loss": 1.1406,
+      "step": 2442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019646809804649694,
+      "loss": 1.1291,
+      "step": 2443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019646519943296084,
+      "loss": 1.2337,
+      "step": 2444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019646229965187428,
+      "loss": 1.1559,
+      "step": 2445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019645939870327234,
+      "loss": 1.2524,
+      "step": 2446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019645649658719014,
+      "loss": 1.2174,
+      "step": 2447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001964535933036628,
+      "loss": 1.3174,
+      "step": 2448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001964506888527254,
+      "loss": 1.1879,
+      "step": 2449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001964477832344132,
+      "loss": 1.267,
+      "step": 2450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019644487644876132,
+      "loss": 1.0925,
+      "step": 2451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019644196849580493,
+      "loss": 1.3004,
+      "step": 2452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019643905937557924,
+      "loss": 1.2432,
+      "step": 2453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019643614908811948,
+      "loss": 0.9927,
+      "step": 2454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019643323763346082,
+      "loss": 1.1964,
+      "step": 2455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019643032501163855,
+      "loss": 1.0973,
+      "step": 2456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001964274112226879,
+      "loss": 1.3987,
+      "step": 2457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019642449626664413,
+      "loss": 1.248,
+      "step": 2458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019642158014354254,
+      "loss": 1.086,
+      "step": 2459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019641866285341838,
+      "loss": 1.2454,
+      "step": 2460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019641574439630701,
+      "loss": 1.1209,
+      "step": 2461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019641282477224375,
+      "loss": 1.2686,
+      "step": 2462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001964099039812639,
+      "loss": 1.1091,
+      "step": 2463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019640698202340285,
+      "loss": 1.3838,
+      "step": 2464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019640405889869596,
+      "loss": 1.1008,
+      "step": 2465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019640113460717859,
+      "loss": 1.1089,
+      "step": 2466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019639820914888614,
+      "loss": 1.1597,
+      "step": 2467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000196395282523854,
+      "loss": 1.1729,
+      "step": 2468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019639235473211766,
+      "loss": 1.1201,
+      "step": 2469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019638942577371248,
+      "loss": 1.1408,
+      "step": 2470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019638649564867393,
+      "loss": 1.2314,
+      "step": 2471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019638356435703748,
+      "loss": 1.1092,
+      "step": 2472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019638063189883863,
+      "loss": 1.1079,
+      "step": 2473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019637769827411287,
+      "loss": 1.1918,
+      "step": 2474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019637476348289567,
+      "loss": 1.1158,
+      "step": 2475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019637182752522256,
+      "loss": 1.2513,
+      "step": 2476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019636889040112912,
+      "loss": 1.2902,
+      "step": 2477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019636595211065084,
+      "loss": 1.3088,
+      "step": 2478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019636301265382332,
+      "loss": 1.1491,
+      "step": 2479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001963600720306821,
+      "loss": 1.2222,
+      "step": 2480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019635713024126282,
+      "loss": 1.1555,
+      "step": 2481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019635418728560105,
+      "loss": 1.1646,
+      "step": 2482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019635124316373244,
+      "loss": 1.1567,
+      "step": 2483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001963482978756926,
+      "loss": 0.9081,
+      "step": 2484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019634535142151716,
+      "loss": 1.0699,
+      "step": 2485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001963424038012418,
+      "loss": 1.1287,
+      "step": 2486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019633945501490224,
+      "loss": 1.1557,
+      "step": 2487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019633650506253414,
+      "loss": 1.2627,
+      "step": 2488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019633355394417317,
+      "loss": 1.1267,
+      "step": 2489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019633060165985506,
+      "loss": 1.2051,
+      "step": 2490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019632764820961556,
+      "loss": 1.1728,
+      "step": 2491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019632469359349045,
+      "loss": 1.1821,
+      "step": 2492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019632173781151542,
+      "loss": 1.2357,
+      "step": 2493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019631878086372629,
+      "loss": 1.2781,
+      "step": 2494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019631582275015882,
+      "loss": 1.2016,
+      "step": 2495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019631286347084889,
+      "loss": 1.3439,
+      "step": 2496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019630990302583218,
+      "loss": 1.181,
+      "step": 2497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019630694141514464,
+      "loss": 1.2914,
+      "step": 2498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019630397863882207,
+      "loss": 1.085,
+      "step": 2499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019630101469690034,
+      "loss": 1.2097,
+      "step": 2500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019629804958941528,
+      "loss": 1.3287,
+      "step": 2501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019629508331640288,
+      "loss": 1.1779,
+      "step": 2502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019629211587789892,
+      "loss": 1.1376,
+      "step": 2503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001962891472739394,
+      "loss": 1.1618,
+      "step": 2504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019628617750456023,
+      "loss": 1.228,
+      "step": 2505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019628320656979732,
+      "loss": 1.1526,
+      "step": 2506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019628023446968665,
+      "loss": 1.3166,
+      "step": 2507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019627726120426426,
+      "loss": 1.1086,
+      "step": 2508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019627428677356602,
+      "loss": 1.2035,
+      "step": 2509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019627131117762797,
+      "loss": 1.272,
+      "step": 2510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001962683344164862,
+      "loss": 1.0978,
+      "step": 2511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019626535649017664,
+      "loss": 1.258,
+      "step": 2512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019626237739873536,
+      "loss": 1.2739,
+      "step": 2513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019625939714219845,
+      "loss": 1.0127,
+      "step": 2514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019625641572060194,
+      "loss": 1.1646,
+      "step": 2515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019625343313398193,
+      "loss": 1.0507,
+      "step": 2516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019625044938237455,
+      "loss": 1.3563,
+      "step": 2517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019624746446581589,
+      "loss": 1.1513,
+      "step": 2518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019624447838434205,
+      "loss": 1.1287,
+      "step": 2519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019624149113798919,
+      "loss": 1.2412,
+      "step": 2520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001962385027267935,
+      "loss": 1.228,
+      "step": 2521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019623551315079113,
+      "loss": 1.1377,
+      "step": 2522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019623252241001823,
+      "loss": 1.1441,
+      "step": 2523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000196229530504511,
+      "loss": 1.197,
+      "step": 2524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001962265374343057,
+      "loss": 1.1436,
+      "step": 2525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019622354319943854,
+      "loss": 1.1618,
+      "step": 2526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019622054779994575,
+      "loss": 1.2323,
+      "step": 2527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019621755123586354,
+      "loss": 1.2602,
+      "step": 2528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019621455350722825,
+      "loss": 1.2505,
+      "step": 2529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019621155461407616,
+      "loss": 1.1598,
+      "step": 2530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019620855455644353,
+      "loss": 1.1077,
+      "step": 2531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019620555333436668,
+      "loss": 1.2858,
+      "step": 2532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019620255094788194,
+      "loss": 1.2394,
+      "step": 2533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019619954739702564,
+      "loss": 1.1708,
+      "step": 2534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019619654268183414,
+      "loss": 1.2243,
+      "step": 2535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019619353680234384,
+      "loss": 1.2201,
+      "step": 2536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019619052975859106,
+      "loss": 1.2436,
+      "step": 2537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001961875215506122,
+      "loss": 1.1985,
+      "step": 2538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019618451217844372,
+      "loss": 1.1987,
+      "step": 2539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019618150164212202,
+      "loss": 1.1976,
+      "step": 2540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019617848994168353,
+      "loss": 1.2567,
+      "step": 2541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001961754770771647,
+      "loss": 1.1949,
+      "step": 2542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019617246304860203,
+      "loss": 1.1191,
+      "step": 2543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001961694478560319,
+      "loss": 1.0912,
+      "step": 2544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019616643149949096,
+      "loss": 1.1684,
+      "step": 2545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019616341397901558,
+      "loss": 1.1437,
+      "step": 2546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019616039529464233,
+      "loss": 1.118,
+      "step": 2547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001961573754464078,
+      "loss": 1.2058,
+      "step": 2548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019615435443434844,
+      "loss": 1.1351,
+      "step": 2549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001961513322585009,
+      "loss": 1.1499,
+      "step": 2550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001961483089189017,
+      "loss": 1.046,
+      "step": 2551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019614528441558746,
+      "loss": 1.2372,
+      "step": 2552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001961422587485948,
+      "loss": 1.0565,
+      "step": 2553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001961392319179603,
+      "loss": 1.2379,
+      "step": 2554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019613620392372062,
+      "loss": 1.2162,
+      "step": 2555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019613317476591244,
+      "loss": 1.1289,
+      "step": 2556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019613014444457236,
+      "loss": 1.3371,
+      "step": 2557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019612711295973712,
+      "loss": 1.0432,
+      "step": 2558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019612408031144334,
+      "loss": 1.2791,
+      "step": 2559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001961210464997278,
+      "loss": 1.2145,
+      "step": 2560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019611801152462715,
+      "loss": 1.1593,
+      "step": 2561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019611497538617816,
+      "loss": 1.1385,
+      "step": 2562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019611193808441758,
+      "loss": 1.106,
+      "step": 2563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019610889961938218,
+      "loss": 1.2402,
+      "step": 2564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001961058599911087,
+      "loss": 1.1123,
+      "step": 2565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019610281919963396,
+      "loss": 1.2612,
+      "step": 2566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019609977724499475,
+      "loss": 1.1921,
+      "step": 2567
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019609673412722785,
+      "loss": 1.0816,
+      "step": 2568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001960936898463702,
+      "loss": 1.3265,
+      "step": 2569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019609064440245856,
+      "loss": 1.291,
+      "step": 2570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019608759779552977,
+      "loss": 1.1052,
+      "step": 2571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001960845500256208,
+      "loss": 1.0763,
+      "step": 2572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019608150109276848,
+      "loss": 1.1778,
+      "step": 2573
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019607845099700968,
+      "loss": 1.1707,
+      "step": 2574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019607539973838135,
+      "loss": 1.0765,
+      "step": 2575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019607234731692047,
+      "loss": 1.2224,
+      "step": 2576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019606929373266387,
+      "loss": 1.1688,
+      "step": 2577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001960662389856486,
+      "loss": 1.1616,
+      "step": 2578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019606318307591162,
+      "loss": 1.0694,
+      "step": 2579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019606012600348989,
+      "loss": 1.0636,
+      "step": 2580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001960570677684204,
+      "loss": 1.139,
+      "step": 2581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019605400837074025,
+      "loss": 1.0871,
+      "step": 2582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019605094781048634,
+      "loss": 1.2316,
+      "step": 2583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001960478860876958,
+      "loss": 1.0703,
+      "step": 2584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001960448232024057,
+      "loss": 1.2121,
+      "step": 2585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019604175915465306,
+      "loss": 1.0084,
+      "step": 2586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019603869394447499,
+      "loss": 1.2239,
+      "step": 2587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019603562757190856,
+      "loss": 1.2633,
+      "step": 2588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019603256003699094,
+      "loss": 1.0056,
+      "step": 2589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001960294913397592,
+      "loss": 1.1808,
+      "step": 2590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001960264214802505,
+      "loss": 1.2245,
+      "step": 2591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000196023350458502,
+      "loss": 1.2134,
+      "step": 2592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001960202782745509,
+      "loss": 1.1586,
+      "step": 2593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019601720492843435,
+      "loss": 1.1147,
+      "step": 2594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001960141304201895,
+      "loss": 1.1237,
+      "step": 2595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019601105474985366,
+      "loss": 1.2166,
+      "step": 2596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019600797791746402,
+      "loss": 1.2484,
+      "step": 2597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001960048999230578,
+      "loss": 1.149,
+      "step": 2598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019600182076667225,
+      "loss": 1.1796,
+      "step": 2599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019599874044834464,
+      "loss": 1.2709,
+      "step": 2600
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 1.1679049730300903,
+      "eval_runtime": 855.2871,
+      "eval_samples_per_second": 28.131,
+      "eval_steps_per_second": 14.065,
+      "step": 2600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019599565896811228,
+      "loss": 1.1034,
+      "step": 2601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019599257632601244,
+      "loss": 1.0346,
+      "step": 2602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019598949252208245,
+      "loss": 1.2283,
+      "step": 2603
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001959864075563596,
+      "loss": 1.1347,
+      "step": 2604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019598332142888128,
+      "loss": 1.038,
+      "step": 2605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019598023413968483,
+      "loss": 1.2011,
+      "step": 2606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019597714568880757,
+      "loss": 1.1494,
+      "step": 2607
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019597405607628694,
+      "loss": 1.2141,
+      "step": 2608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001959709653021603,
+      "loss": 1.2627,
+      "step": 2609
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019596787336646502,
+      "loss": 1.2525,
+      "step": 2610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019596478026923865,
+      "loss": 1.159,
+      "step": 2611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019596168601051847,
+      "loss": 1.0066,
+      "step": 2612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019595859059034205,
+      "loss": 1.1122,
+      "step": 2613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019595549400874677,
+      "loss": 1.2434,
+      "step": 2614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001959523962657702,
+      "loss": 1.2194,
+      "step": 2615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019594929736144976,
+      "loss": 1.1108,
+      "step": 2616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019594619729582297,
+      "loss": 1.0736,
+      "step": 2617
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019594309606892738,
+      "loss": 1.0909,
+      "step": 2618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001959399936808005,
+      "loss": 1.1931,
+      "step": 2619
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019593689013147985,
+      "loss": 0.9864,
+      "step": 2620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019593378542100308,
+      "loss": 1.277,
+      "step": 2621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019593067954940766,
+      "loss": 1.0465,
+      "step": 2622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019592757251673127,
+      "loss": 1.3427,
+      "step": 2623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001959244643230115,
+      "loss": 1.1112,
+      "step": 2624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019592135496828594,
+      "loss": 1.072,
+      "step": 2625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019591824445259223,
+      "loss": 1.1333,
+      "step": 2626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019591513277596802,
+      "loss": 1.3316,
+      "step": 2627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019591201993845097,
+      "loss": 1.146,
+      "step": 2628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001959089059400788,
+      "loss": 1.0821,
+      "step": 2629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019590579078088914,
+      "loss": 1.1585,
+      "step": 2630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001959026744609197,
+      "loss": 1.082,
+      "step": 2631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019589955698020824,
+      "loss": 1.1397,
+      "step": 2632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001958964383387924,
+      "loss": 1.2695,
+      "step": 2633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019589331853671005,
+      "loss": 1.1113,
+      "step": 2634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019589019757399888,
+      "loss": 1.2881,
+      "step": 2635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019588707545069667,
+      "loss": 1.2249,
+      "step": 2636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019588395216684122,
+      "loss": 1.2494,
+      "step": 2637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019588082772247032,
+      "loss": 1.0933,
+      "step": 2638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019587770211762178,
+      "loss": 1.1532,
+      "step": 2639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019587457535233345,
+      "loss": 1.0871,
+      "step": 2640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019587144742664318,
+      "loss": 1.1803,
+      "step": 2641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019586831834058876,
+      "loss": 1.2392,
+      "step": 2642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019586518809420816,
+      "loss": 1.1,
+      "step": 2643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001958620566875392,
+      "loss": 1.1825,
+      "step": 2644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019585892412061982,
+      "loss": 1.1703,
+      "step": 2645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019585579039348792,
+      "loss": 1.0768,
+      "step": 2646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001958526555061814,
+      "loss": 1.0962,
+      "step": 2647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019584951945873826,
+      "loss": 1.1828,
+      "step": 2648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019584638225119642,
+      "loss": 1.0241,
+      "step": 2649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019584324388359386,
+      "loss": 1.1085,
+      "step": 2650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019584010435596852,
+      "loss": 1.1093,
+      "step": 2651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019583696366835848,
+      "loss": 1.208,
+      "step": 2652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001958338218208017,
+      "loss": 1.0845,
+      "step": 2653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019583067881333622,
+      "loss": 1.1916,
+      "step": 2654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001958275346460001,
+      "loss": 1.0389,
+      "step": 2655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001958243893188313,
+      "loss": 0.9692,
+      "step": 2656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019582124283186802,
+      "loss": 1.1234,
+      "step": 2657
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001958180951851483,
+      "loss": 1.23,
+      "step": 2658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019581494637871017,
+      "loss": 1.1873,
+      "step": 2659
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019581179641259185,
+      "loss": 1.1779,
+      "step": 2660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019580864528683138,
+      "loss": 1.1627,
+      "step": 2661
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019580549300146695,
+      "loss": 1.0669,
+      "step": 2662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001958023395565367,
+      "loss": 1.2917,
+      "step": 2663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019579918495207874,
+      "loss": 1.1435,
+      "step": 2664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019579602918813133,
+      "loss": 1.2653,
+      "step": 2665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019579287226473264,
+      "loss": 1.1868,
+      "step": 2666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019578971418192088,
+      "loss": 1.1138,
+      "step": 2667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019578655493973426,
+      "loss": 1.2118,
+      "step": 2668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000195783394538211,
+      "loss": 1.0755,
+      "step": 2669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019578023297738941,
+      "loss": 1.1345,
+      "step": 2670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001957770702573077,
+      "loss": 1.2947,
+      "step": 2671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001957739063780042,
+      "loss": 1.2004,
+      "step": 2672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019577074133951715,
+      "loss": 1.0588,
+      "step": 2673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001957675751418849,
+      "loss": 1.0629,
+      "step": 2674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019576440778514576,
+      "loss": 1.3012,
+      "step": 2675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019576123926933804,
+      "loss": 1.0991,
+      "step": 2676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019575806959450012,
+      "loss": 1.099,
+      "step": 2677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019575489876067035,
+      "loss": 1.1617,
+      "step": 2678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019575172676788709,
+      "loss": 1.1936,
+      "step": 2679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001957485536161888,
+      "loss": 1.2037,
+      "step": 2680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001957453793056138,
+      "loss": 1.2395,
+      "step": 2681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019574220383620055,
+      "loss": 1.164,
+      "step": 2682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001957390272079875,
+      "loss": 1.2381,
+      "step": 2683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001957358494210131,
+      "loss": 0.9913,
+      "step": 2684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019573267047531573,
+      "loss": 1.1166,
+      "step": 2685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019572949037093395,
+      "loss": 1.1634,
+      "step": 2686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019572630910790623,
+      "loss": 1.1095,
+      "step": 2687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001957231266862711,
+      "loss": 1.1011,
+      "step": 2688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000195719943106067,
+      "loss": 1.2051,
+      "step": 2689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019571675836733255,
+      "loss": 1.2634,
+      "step": 2690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019571357247010623,
+      "loss": 1.1205,
+      "step": 2691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019571038541442665,
+      "loss": 1.0548,
+      "step": 2692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019570719720033236,
+      "loss": 1.2669,
+      "step": 2693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019570400782786194,
+      "loss": 1.2205,
+      "step": 2694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000195700817297054,
+      "loss": 1.1097,
+      "step": 2695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019569762560794717,
+      "loss": 1.2309,
+      "step": 2696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019569443276058003,
+      "loss": 1.1499,
+      "step": 2697
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001956912387549913,
+      "loss": 1.0591,
+      "step": 2698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019568804359121957,
+      "loss": 1.0794,
+      "step": 2699
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019568484726930355,
+      "loss": 1.078,
+      "step": 2700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019568164978928194,
+      "loss": 1.2537,
+      "step": 2701
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019567845115119338,
+      "loss": 1.1168,
+      "step": 2702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019567525135507664,
+      "loss": 1.1759,
+      "step": 2703
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019567205040097046,
+      "loss": 1.056,
+      "step": 2704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019566884828891353,
+      "loss": 1.2648,
+      "step": 2705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001956656450189446,
+      "loss": 1.0193,
+      "step": 2706
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001956624405911025,
+      "loss": 1.2575,
+      "step": 2707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019565923500542594,
+      "loss": 1.116,
+      "step": 2708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001956560282619538,
+      "loss": 1.1845,
+      "step": 2709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019565282036072487,
+      "loss": 1.1089,
+      "step": 2710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001956496113017779,
+      "loss": 1.1269,
+      "step": 2711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019564640108515186,
+      "loss": 1.3482,
+      "step": 2712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019564318971088552,
+      "loss": 1.2154,
+      "step": 2713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019563997717901774,
+      "loss": 1.0327,
+      "step": 2714
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019563676348958742,
+      "loss": 0.9974,
+      "step": 2715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019563354864263348,
+      "loss": 1.1569,
+      "step": 2716
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001956303326381948,
+      "loss": 1.2203,
+      "step": 2717
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019562711547631032,
+      "loss": 1.0525,
+      "step": 2718
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019562389715701897,
+      "loss": 1.038,
+      "step": 2719
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001956206776803597,
+      "loss": 1.3221,
+      "step": 2720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019561745704637151,
+      "loss": 1.0764,
+      "step": 2721
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019561423525509333,
+      "loss": 1.1498,
+      "step": 2722
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001956110123065642,
+      "loss": 1.1085,
+      "step": 2723
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001956077882008231,
+      "loss": 1.0614,
+      "step": 2724
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019560456293790904,
+      "loss": 1.0937,
+      "step": 2725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001956013365178611,
+      "loss": 1.2969,
+      "step": 2726
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001955981089407183,
+      "loss": 1.2244,
+      "step": 2727
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019559488020651968,
+      "loss": 1.1905,
+      "step": 2728
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019559165031530438,
+      "loss": 1.1896,
+      "step": 2729
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019558841926711144,
+      "loss": 1.2613,
+      "step": 2730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019558518706198002,
+      "loss": 1.078,
+      "step": 2731
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019558195369994918,
+      "loss": 1.1849,
+      "step": 2732
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019557871918105807,
+      "loss": 1.127,
+      "step": 2733
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001955754835053459,
+      "loss": 1.1551,
+      "step": 2734
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019557224667285174,
+      "loss": 1.2058,
+      "step": 2735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019556900868361482,
+      "loss": 1.1576,
+      "step": 2736
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019556576953767434,
+      "loss": 1.2586,
+      "step": 2737
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001955625292350695,
+      "loss": 0.9509,
+      "step": 2738
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019555928777583944,
+      "loss": 1.0978,
+      "step": 2739
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001955560451600235,
+      "loss": 1.2854,
+      "step": 2740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001955528013876609,
+      "loss": 1.2867,
+      "step": 2741
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019554955645879082,
+      "loss": 1.252,
+      "step": 2742
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019554631037345263,
+      "loss": 1.27,
+      "step": 2743
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001955430631316856,
+      "loss": 1.2318,
+      "step": 2744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019553981473352901,
+      "loss": 1.1831,
+      "step": 2745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001955365651790222,
+      "loss": 1.1849,
+      "step": 2746
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019553331446820445,
+      "loss": 1.1517,
+      "step": 2747
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019553006260111515,
+      "loss": 1.097,
+      "step": 2748
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019552680957779365,
+      "loss": 1.2604,
+      "step": 2749
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001955235553982793,
+      "loss": 1.2019,
+      "step": 2750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019552030006261158,
+      "loss": 1.1736,
+      "step": 2751
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019551704357082973,
+      "loss": 1.088,
+      "step": 2752
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001955137859229733,
+      "loss": 1.344,
+      "step": 2753
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019551052711908168,
+      "loss": 1.1415,
+      "step": 2754
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019550726715919425,
+      "loss": 1.0394,
+      "step": 2755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019550400604335057,
+      "loss": 1.1756,
+      "step": 2756
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019550074377159003,
+      "loss": 1.2487,
+      "step": 2757
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019549748034395216,
+      "loss": 1.1799,
+      "step": 2758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019549421576047643,
+      "loss": 1.2248,
+      "step": 2759
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019549095002120237,
+      "loss": 1.1379,
+      "step": 2760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001954876831261695,
+      "loss": 1.2114,
+      "step": 2761
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019548441507541737,
+      "loss": 1.1157,
+      "step": 2762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001954811458689855,
+      "loss": 1.2496,
+      "step": 2763
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001954778755069135,
+      "loss": 1.2387,
+      "step": 2764
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019547460398924092,
+      "loss": 1.1208,
+      "step": 2765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019547133131600738,
+      "loss": 1.1132,
+      "step": 2766
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001954680574872525,
+      "loss": 1.1434,
+      "step": 2767
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001954647825030159,
+      "loss": 1.1845,
+      "step": 2768
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019546150636333716,
+      "loss": 1.1431,
+      "step": 2769
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000195458229068256,
+      "loss": 1.1941,
+      "step": 2770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019545495061781205,
+      "loss": 1.2934,
+      "step": 2771
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000195451671012045,
+      "loss": 1.0902,
+      "step": 2772
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019544839025099458,
+      "loss": 1.2429,
+      "step": 2773
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019544510833470043,
+      "loss": 1.2569,
+      "step": 2774
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019544182526320234,
+      "loss": 1.0123,
+      "step": 2775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019543854103653999,
+      "loss": 1.199,
+      "step": 2776
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019543525565475316,
+      "loss": 1.1079,
+      "step": 2777
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001954319691178816,
+      "loss": 1.1013,
+      "step": 2778
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019542868142596512,
+      "loss": 1.1224,
+      "step": 2779
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019542539257904346,
+      "loss": 1.1149,
+      "step": 2780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019542210257715649,
+      "loss": 1.1466,
+      "step": 2781
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019541881142034394,
+      "loss": 1.1774,
+      "step": 2782
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019541551910864574,
+      "loss": 1.0757,
+      "step": 2783
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001954122256421017,
+      "loss": 1.1251,
+      "step": 2784
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019540893102075165,
+      "loss": 1.2928,
+      "step": 2785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001954056352446355,
+      "loss": 1.0343,
+      "step": 2786
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019540233831379314,
+      "loss": 1.3684,
+      "step": 2787
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019539904022826448,
+      "loss": 1.0766,
+      "step": 2788
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001953957409880894,
+      "loss": 1.0571,
+      "step": 2789
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019539244059330787,
+      "loss": 0.9106,
+      "step": 2790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001953891390439598,
+      "loss": 1.1841,
+      "step": 2791
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001953858363400852,
+      "loss": 1.2192,
+      "step": 2792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019538253248172402,
+      "loss": 1.12,
+      "step": 2793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019537922746891624,
+      "loss": 1.1658,
+      "step": 2794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019537592130170186,
+      "loss": 1.013,
+      "step": 2795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001953726139801209,
+      "loss": 1.1796,
+      "step": 2796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001953693055042134,
+      "loss": 1.0407,
+      "step": 2797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019536599587401937,
+      "loss": 1.2012,
+      "step": 2798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019536268508957893,
+      "loss": 1.145,
+      "step": 2799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001953593731509321,
+      "loss": 1.1537,
+      "step": 2800
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 1.1405117511749268,
+      "eval_runtime": 847.2068,
+      "eval_samples_per_second": 28.399,
+      "eval_steps_per_second": 14.2,
+      "step": 2800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019535606005811898,
+      "loss": 1.0657,
+      "step": 2801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019535274581117964,
+      "loss": 1.1344,
+      "step": 2802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019534943041015423,
+      "loss": 1.1518,
+      "step": 2803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001953461138550829,
+      "loss": 1.1976,
+      "step": 2804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019534279614600574,
+      "loss": 1.0645,
+      "step": 2805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019533947728296293,
+      "loss": 1.2103,
+      "step": 2806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019533615726599463,
+      "loss": 1.07,
+      "step": 2807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019533283609514106,
+      "loss": 1.1834,
+      "step": 2808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019532951377044234,
+      "loss": 1.0041,
+      "step": 2809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019532619029193876,
+      "loss": 1.1014,
+      "step": 2810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001953228656596705,
+      "loss": 1.1337,
+      "step": 2811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001953195398736778,
+      "loss": 1.1445,
+      "step": 2812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019531621293400093,
+      "loss": 1.1354,
+      "step": 2813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019531288484068017,
+      "loss": 1.1341,
+      "step": 2814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019530955559375577,
+      "loss": 0.9783,
+      "step": 2815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019530622519326802,
+      "loss": 1.085,
+      "step": 2816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019530289363925727,
+      "loss": 1.0498,
+      "step": 2817
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019529956093176384,
+      "loss": 1.1236,
+      "step": 2818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000195296227070828,
+      "loss": 1.2181,
+      "step": 2819
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019529289205649019,
+      "loss": 1.1146,
+      "step": 2820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001952895558887907,
+      "loss": 1.0315,
+      "step": 2821
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019528621856776995,
+      "loss": 1.1387,
+      "step": 2822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001952828800934683,
+      "loss": 1.1805,
+      "step": 2823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019527954046592622,
+      "loss": 1.0416,
+      "step": 2824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019527619968518408,
+      "loss": 1.1594,
+      "step": 2825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019527285775128232,
+      "loss": 1.2053,
+      "step": 2826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001952695146642614,
+      "loss": 1.129,
+      "step": 2827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019526617042416175,
+      "loss": 1.118,
+      "step": 2828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019526282503102392,
+      "loss": 1.1349,
+      "step": 2829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019525947848488833,
+      "loss": 1.1105,
+      "step": 2830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001952561307857955,
+      "loss": 1.1462,
+      "step": 2831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019525278193378593,
+      "loss": 1.182,
+      "step": 2832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001952494319289002,
+      "loss": 1.0039,
+      "step": 2833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019524608077117888,
+      "loss": 1.0299,
+      "step": 2834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019524272846066244,
+      "loss": 1.0995,
+      "step": 2835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001952393749973915,
+      "loss": 1.0782,
+      "step": 2836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019523602038140666,
+      "loss": 1.1647,
+      "step": 2837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001952326646127485,
+      "loss": 1.0543,
+      "step": 2838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019522930769145765,
+      "loss": 1.0636,
+      "step": 2839
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019522594961757475,
+      "loss": 1.1186,
+      "step": 2840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001952225903911404,
+      "loss": 1.1586,
+      "step": 2841
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019521923001219527,
+      "loss": 1.1195,
+      "step": 2842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001952158684807801,
+      "loss": 1.1253,
+      "step": 2843
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019521250579693548,
+      "loss": 1.0795,
+      "step": 2844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019520914196070218,
+      "loss": 1.3187,
+      "step": 2845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001952057769721209,
+      "loss": 1.0886,
+      "step": 2846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019520241083123235,
+      "loss": 1.125,
+      "step": 2847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019519904353807724,
+      "loss": 1.0891,
+      "step": 2848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019519567509269638,
+      "loss": 1.0868,
+      "step": 2849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019519230549513053,
+      "loss": 1.2038,
+      "step": 2850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019518893474542047,
+      "loss": 1.1999,
+      "step": 2851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019518556284360696,
+      "loss": 1.0513,
+      "step": 2852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019518218978973088,
+      "loss": 1.0994,
+      "step": 2853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000195178815583833,
+      "loss": 1.1785,
+      "step": 2854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019517544022595417,
+      "loss": 1.1219,
+      "step": 2855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019517206371613528,
+      "loss": 1.2072,
+      "step": 2856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019516868605441715,
+      "loss": 1.1133,
+      "step": 2857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019516530724084067,
+      "loss": 1.0365,
+      "step": 2858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019516192727544676,
+      "loss": 1.2082,
+      "step": 2859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019515854615827628,
+      "loss": 1.0691,
+      "step": 2860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019515516388937023,
+      "loss": 1.2835,
+      "step": 2861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019515178046876946,
+      "loss": 1.1234,
+      "step": 2862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000195148395896515,
+      "loss": 1.02,
+      "step": 2863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019514501017264777,
+      "loss": 1.033,
+      "step": 2864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019514162329720876,
+      "loss": 1.1583,
+      "step": 2865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019513823527023895,
+      "loss": 1.17,
+      "step": 2866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001951348460917793,
+      "loss": 1.0066,
+      "step": 2867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019513145576187095,
+      "loss": 1.0966,
+      "step": 2868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019512806428055484,
+      "loss": 1.303,
+      "step": 2869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019512467164787208,
+      "loss": 1.0975,
+      "step": 2870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001951212778638637,
+      "loss": 1.0501,
+      "step": 2871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019511788292857077,
+      "loss": 1.1732,
+      "step": 2872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019511448684203435,
+      "loss": 1.103,
+      "step": 2873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019511108960429562,
+      "loss": 1.1228,
+      "step": 2874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019510769121539566,
+      "loss": 0.9973,
+      "step": 2875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019510429167537556,
+      "loss": 1.1161,
+      "step": 2876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001951008909842765,
+      "loss": 1.1357,
+      "step": 2877
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001950974891421397,
+      "loss": 1.1819,
+      "step": 2878
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019509408614900624,
+      "loss": 1.1929,
+      "step": 2879
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019509068200491736,
+      "loss": 1.1687,
+      "step": 2880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019508727670991424,
+      "loss": 1.2717,
+      "step": 2881
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001950838702640381,
+      "loss": 1.1644,
+      "step": 2882
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001950804626673302,
+      "loss": 1.0149,
+      "step": 2883
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019507705391983172,
+      "loss": 0.9198,
+      "step": 2884
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019507364402158398,
+      "loss": 1.2865,
+      "step": 2885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001950702329726282,
+      "loss": 1.0235,
+      "step": 2886
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019506682077300572,
+      "loss": 1.1165,
+      "step": 2887
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001950634074227578,
+      "loss": 0.9573,
+      "step": 2888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019505999292192573,
+      "loss": 1.0884,
+      "step": 2889
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001950565772705509,
+      "loss": 1.0589,
+      "step": 2890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019505316046867458,
+      "loss": 1.0206,
+      "step": 2891
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019504974251633822,
+      "loss": 1.2226,
+      "step": 2892
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001950463234135831,
+      "loss": 1.0295,
+      "step": 2893
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019504290316045065,
+      "loss": 1.0879,
+      "step": 2894
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019503948175698222,
+      "loss": 1.2031,
+      "step": 2895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019503605920321925,
+      "loss": 1.0588,
+      "step": 2896
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001950326354992032,
+      "loss": 1.2432,
+      "step": 2897
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019502921064497547,
+      "loss": 1.0914,
+      "step": 2898
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001950257846405775,
+      "loss": 1.1192,
+      "step": 2899
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001950223574860508,
+      "loss": 1.0037,
+      "step": 2900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019501892918143677,
+      "loss": 1.0713,
+      "step": 2901
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019501549972677698,
+      "loss": 1.2523,
+      "step": 2902
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019501206912211295,
+      "loss": 1.0887,
+      "step": 2903
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019500863736748611,
+      "loss": 1.1564,
+      "step": 2904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019500520446293808,
+      "loss": 1.1078,
+      "step": 2905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019500177040851038,
+      "loss": 1.1156,
+      "step": 2906
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001949983352042446,
+      "loss": 0.9986,
+      "step": 2907
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019499489885018223,
+      "loss": 0.9519,
+      "step": 2908
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019499146134636496,
+      "loss": 1.1535,
+      "step": 2909
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019498802269283437,
+      "loss": 1.1705,
+      "step": 2910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019498458288963203,
+      "loss": 1.104,
+      "step": 2911
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019498114193679963,
+      "loss": 1.0203,
+      "step": 2912
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001949776998343788,
+      "loss": 1.0403,
+      "step": 2913
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001949742565824112,
+      "loss": 1.0917,
+      "step": 2914
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001949708121809385,
+      "loss": 1.0487,
+      "step": 2915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019496736663000242,
+      "loss": 1.0754,
+      "step": 2916
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001949639199296446,
+      "loss": 1.0492,
+      "step": 2917
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019496047207990679,
+      "loss": 1.1187,
+      "step": 2918
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019495702308083075,
+      "loss": 1.0917,
+      "step": 2919
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019495357293245816,
+      "loss": 1.1418,
+      "step": 2920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019495012163483086,
+      "loss": 1.1173,
+      "step": 2921
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019494666918799057,
+      "loss": 1.1141,
+      "step": 2922
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019494321559197902,
+      "loss": 1.0561,
+      "step": 2923
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019493976084683813,
+      "loss": 1.0461,
+      "step": 2924
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019493630495260965,
+      "loss": 1.083,
+      "step": 2925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019493284790933542,
+      "loss": 1.1817,
+      "step": 2926
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019492938971705725,
+      "loss": 1.0478,
+      "step": 2927
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019492593037581703,
+      "loss": 1.1497,
+      "step": 2928
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019492246988565666,
+      "loss": 1.0653,
+      "step": 2929
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019491900824661796,
+      "loss": 1.2606,
+      "step": 2930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019491554545874283,
+      "loss": 1.0681,
+      "step": 2931
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001949120815220732,
+      "loss": 1.013,
+      "step": 2932
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000194908616436651,
+      "loss": 1.2579,
+      "step": 2933
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019490515020251822,
+      "loss": 1.2315,
+      "step": 2934
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001949016828197167,
+      "loss": 1.2434,
+      "step": 2935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019489821428828852,
+      "loss": 1.3071,
+      "step": 2936
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019489474460827557,
+      "loss": 1.0933,
+      "step": 2937
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019489127377971988,
+      "loss": 1.071,
+      "step": 2938
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001948878018026635,
+      "loss": 1.2572,
+      "step": 2939
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019488432867714842,
+      "loss": 1.0963,
+      "step": 2940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001948808544032166,
+      "loss": 1.04,
+      "step": 2941
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019487737898091022,
+      "loss": 1.1407,
+      "step": 2942
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001948739024102713,
+      "loss": 1.1465,
+      "step": 2943
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019487042469134186,
+      "loss": 1.1006,
+      "step": 2944
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019486694582416405,
+      "loss": 1.0147,
+      "step": 2945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019486346580877996,
+      "loss": 0.9714,
+      "step": 2946
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019485998464523172,
+      "loss": 1.1468,
+      "step": 2947
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019485650233356142,
+      "loss": 1.1049,
+      "step": 2948
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001948530188738113,
+      "loss": 1.0709,
+      "step": 2949
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019484953426602343,
+      "loss": 0.9843,
+      "step": 2950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019484604851024003,
+      "loss": 1.0828,
+      "step": 2951
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019484256160650327,
+      "loss": 1.1092,
+      "step": 2952
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019483907355485536,
+      "loss": 1.1804,
+      "step": 2953
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019483558435533853,
+      "loss": 1.0253,
+      "step": 2954
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019483209400799497,
+      "loss": 0.9716,
+      "step": 2955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000194828602512867,
+      "loss": 1.0626,
+      "step": 2956
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001948251098699968,
+      "loss": 1.038,
+      "step": 2957
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019482161607942667,
+      "loss": 1.1883,
+      "step": 2958
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019481812114119894,
+      "loss": 0.9894,
+      "step": 2959
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019481462505535583,
+      "loss": 1.2908,
+      "step": 2960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019481112782193972,
+      "loss": 1.1408,
+      "step": 2961
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019480762944099292,
+      "loss": 1.2224,
+      "step": 2962
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019480412991255776,
+      "loss": 1.1218,
+      "step": 2963
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001948006292366766,
+      "loss": 1.0171,
+      "step": 2964
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019479712741339182,
+      "loss": 1.1416,
+      "step": 2965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001947936244427458,
+      "loss": 1.2906,
+      "step": 2966
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019479012032478092,
+      "loss": 1.0742,
+      "step": 2967
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001947866150595396,
+      "loss": 1.1965,
+      "step": 2968
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001947831086470643,
+      "loss": 1.1165,
+      "step": 2969
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001947796010873974,
+      "loss": 1.1465,
+      "step": 2970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001947760923805814,
+      "loss": 1.233,
+      "step": 2971
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019477258252665873,
+      "loss": 1.1907,
+      "step": 2972
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019476907152567193,
+      "loss": 1.2407,
+      "step": 2973
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019476555937766342,
+      "loss": 1.0566,
+      "step": 2974
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019476204608267578,
+      "loss": 1.1406,
+      "step": 2975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019475853164075148,
+      "loss": 1.1637,
+      "step": 2976
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019475501605193307,
+      "loss": 1.2581,
+      "step": 2977
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019475149931626312,
+      "loss": 1.1309,
+      "step": 2978
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019474798143378417,
+      "loss": 1.1565,
+      "step": 2979
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001947444624045388,
+      "loss": 1.1198,
+      "step": 2980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019474094222856963,
+      "loss": 1.0732,
+      "step": 2981
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019473742090591922,
+      "loss": 1.0812,
+      "step": 2982
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001947338984366302,
+      "loss": 1.106,
+      "step": 2983
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019473037482074524,
+      "loss": 1.0828,
+      "step": 2984
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000194726850058307,
+      "loss": 1.0865,
+      "step": 2985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019472332414935804,
+      "loss": 1.049,
+      "step": 2986
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019471979709394116,
+      "loss": 1.1347,
+      "step": 2987
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000194716268892099,
+      "loss": 1.1126,
+      "step": 2988
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019471273954387418,
+      "loss": 1.1884,
+      "step": 2989
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019470920904930954,
+      "loss": 1.0099,
+      "step": 2990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019470567740844776,
+      "loss": 1.1137,
+      "step": 2991
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019470214462133161,
+      "loss": 1.167,
+      "step": 2992
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019469861068800376,
+      "loss": 1.0914,
+      "step": 2993
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001946950756085071,
+      "loss": 1.1482,
+      "step": 2994
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019469153938288437,
+      "loss": 1.0318,
+      "step": 2995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019468800201117832,
+      "loss": 1.0478,
+      "step": 2996
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019468446349343185,
+      "loss": 1.1035,
+      "step": 2997
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019468092382968772,
+      "loss": 1.1485,
+      "step": 2998
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019467738301998878,
+      "loss": 0.993,
+      "step": 2999
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019467384106437794,
+      "loss": 1.1408,
+      "step": 3000
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 1.1153876781463623,
+      "eval_runtime": 847.449,
+      "eval_samples_per_second": 28.391,
+      "eval_steps_per_second": 14.196,
+      "step": 3000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019467029796289804,
+      "loss": 1.0502,
+      "step": 3001
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019466675371559194,
+      "loss": 1.0869,
+      "step": 3002
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019466320832250253,
+      "loss": 1.0212,
+      "step": 3003
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019465966178367278,
+      "loss": 1.1455,
+      "step": 3004
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019465611409914555,
+      "loss": 0.9807,
+      "step": 3005
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019465256526896382,
+      "loss": 1.1323,
+      "step": 3006
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019464901529317054,
+      "loss": 1.1375,
+      "step": 3007
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019464546417180867,
+      "loss": 1.1674,
+      "step": 3008
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019464191190492116,
+      "loss": 1.1554,
+      "step": 3009
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019463835849255106,
+      "loss": 1.1709,
+      "step": 3010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019463480393474135,
+      "loss": 1.1773,
+      "step": 3011
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019463124823153503,
+      "loss": 1.1215,
+      "step": 3012
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019462769138297515,
+      "loss": 1.0463,
+      "step": 3013
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001946241333891048,
+      "loss": 1.012,
+      "step": 3014
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000194620574249967,
+      "loss": 1.0563,
+      "step": 3015
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019461701396560486,
+      "loss": 1.23,
+      "step": 3016
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019461345253606141,
+      "loss": 0.958,
+      "step": 3017
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001946098899613798,
+      "loss": 1.097,
+      "step": 3018
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019460632624160318,
+      "loss": 1.1439,
+      "step": 3019
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001946027613767746,
+      "loss": 1.1094,
+      "step": 3020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019459919536693728,
+      "loss": 1.0614,
+      "step": 3021
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019459562821213435,
+      "loss": 1.0947,
+      "step": 3022
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000194592059912409,
+      "loss": 1.221,
+      "step": 3023
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019458849046780438,
+      "loss": 1.2204,
+      "step": 3024
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019458491987836375,
+      "loss": 1.0419,
+      "step": 3025
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019458134814413026,
+      "loss": 1.0817,
+      "step": 3026
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001945777752651472,
+      "loss": 1.0231,
+      "step": 3027
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001945742012414578,
+      "loss": 1.0179,
+      "step": 3028
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019457062607310528,
+      "loss": 1.0691,
+      "step": 3029
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019456704976013295,
+      "loss": 1.0847,
+      "step": 3030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001945634723025841,
+      "loss": 1.2167,
+      "step": 3031
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000194559893700502,
+      "loss": 1.0138,
+      "step": 3032
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019455631395392996,
+      "loss": 1.1564,
+      "step": 3033
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019455273306291133,
+      "loss": 1.0998,
+      "step": 3034
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019454915102748945,
+      "loss": 1.1906,
+      "step": 3035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019454556784770767,
+      "loss": 1.2193,
+      "step": 3036
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019454198352360935,
+      "loss": 1.1032,
+      "step": 3037
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019453839805523787,
+      "loss": 1.0429,
+      "step": 3038
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019453481144263667,
+      "loss": 1.1578,
+      "step": 3039
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001945312236858491,
+      "loss": 1.1458,
+      "step": 3040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001945276347849186,
+      "loss": 1.0836,
+      "step": 3041
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019452404473988862,
+      "loss": 1.0594,
+      "step": 3042
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001945204535508026,
+      "loss": 1.1519,
+      "step": 3043
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019451686121770402,
+      "loss": 1.0857,
+      "step": 3044
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019451326774063636,
+      "loss": 1.0978,
+      "step": 3045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001945096731196431,
+      "loss": 1.1115,
+      "step": 3046
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019450607735476774,
+      "loss": 1.0649,
+      "step": 3047
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019450248044605383,
+      "loss": 1.0519,
+      "step": 3048
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019449888239354488,
+      "loss": 1.0528,
+      "step": 3049
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019449528319728448,
+      "loss": 1.0689,
+      "step": 3050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019449168285731608,
+      "loss": 1.0878,
+      "step": 3051
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019448808137368338,
+      "loss": 1.1224,
+      "step": 3052
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019448447874642995,
+      "loss": 1.1025,
+      "step": 3053
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019448087497559935,
+      "loss": 1.089,
+      "step": 3054
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019447727006123523,
+      "loss": 1.1372,
+      "step": 3055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019447366400338116,
+      "loss": 0.9609,
+      "step": 3056
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019447005680208086,
+      "loss": 1.0112,
+      "step": 3057
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019446644845737797,
+      "loss": 1.1495,
+      "step": 3058
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019446283896931612,
+      "loss": 1.1379,
+      "step": 3059
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019445922833793907,
+      "loss": 1.14,
+      "step": 3060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019445561656329045,
+      "loss": 1.1522,
+      "step": 3061
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019445200364541404,
+      "loss": 1.0554,
+      "step": 3062
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001944483895843535,
+      "loss": 1.0909,
+      "step": 3063
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019444477438015264,
+      "loss": 1.1634,
+      "step": 3064
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019444115803285517,
+      "loss": 1.1936,
+      "step": 3065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019443754054250488,
+      "loss": 1.1463,
+      "step": 3066
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001944339219091455,
+      "loss": 1.074,
+      "step": 3067
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019443030213282093,
+      "loss": 1.1925,
+      "step": 3068
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019442668121357493,
+      "loss": 0.9103,
+      "step": 3069
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019442305915145128,
+      "loss": 1.2136,
+      "step": 3070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019441943594649387,
+      "loss": 1.2422,
+      "step": 3071
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019441581159874654,
+      "loss": 1.1261,
+      "step": 3072
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001944121861082532,
+      "loss": 1.1682,
+      "step": 3073
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019440855947505762,
+      "loss": 1.1842,
+      "step": 3074
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001944049316992038,
+      "loss": 1.2524,
+      "step": 3075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019440130278073565,
+      "loss": 1.2052,
+      "step": 3076
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000194397672719697,
+      "loss": 1.0225,
+      "step": 3077
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019439404151613186,
+      "loss": 1.0987,
+      "step": 3078
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019439040917008417,
+      "loss": 0.9847,
+      "step": 3079
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001943867756815979,
+      "loss": 1.0544,
+      "step": 3080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019438314105071696,
+      "loss": 1.0936,
+      "step": 3081
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001943795052774854,
+      "loss": 1.1314,
+      "step": 3082
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019437586836194726,
+      "loss": 0.9923,
+      "step": 3083
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019437223030414647,
+      "loss": 1.2512,
+      "step": 3084
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019436859110412712,
+      "loss": 1.1198,
+      "step": 3085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019436495076193325,
+      "loss": 1.2049,
+      "step": 3086
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001943613092776089,
+      "loss": 1.0074,
+      "step": 3087
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001943576666511982,
+      "loss": 1.0521,
+      "step": 3088
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019435402288274517,
+      "loss": 1.1274,
+      "step": 3089
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019435037797229393,
+      "loss": 1.0742,
+      "step": 3090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019434673191988863,
+      "loss": 1.1236,
+      "step": 3091
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019434308472557338,
+      "loss": 1.1058,
+      "step": 3092
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019433943638939228,
+      "loss": 0.9016,
+      "step": 3093
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019433578691138953,
+      "loss": 1.0182,
+      "step": 3094
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019433213629160931,
+      "loss": 1.1331,
+      "step": 3095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001943284845300958,
+      "loss": 1.0057,
+      "step": 3096
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019432483162689317,
+      "loss": 1.0513,
+      "step": 3097
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019432117758204565,
+      "loss": 1.0399,
+      "step": 3098
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019431752239559746,
+      "loss": 1.0564,
+      "step": 3099
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019431386606759285,
+      "loss": 1.2004,
+      "step": 3100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019431020859807607,
+      "loss": 1.0436,
+      "step": 3101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001943065499870914,
+      "loss": 1.0796,
+      "step": 3102
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001943028902346831,
+      "loss": 0.9842,
+      "step": 3103
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001942992293408955,
+      "loss": 1.0441,
+      "step": 3104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019429556730577284,
+      "loss": 1.2376,
+      "step": 3105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001942919041293595,
+      "loss": 1.3289,
+      "step": 3106
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001942882398116998,
+      "loss": 1.1729,
+      "step": 3107
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001942845743528381,
+      "loss": 1.0813,
+      "step": 3108
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019428090775281876,
+      "loss": 1.1516,
+      "step": 3109
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019427724001168617,
+      "loss": 1.1429,
+      "step": 3110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019427357112948468,
+      "loss": 1.0675,
+      "step": 3111
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019426990110625874,
+      "loss": 1.2527,
+      "step": 3112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019426622994205274,
+      "loss": 1.08,
+      "step": 3113
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019426255763691113,
+      "loss": 1.0354,
+      "step": 3114
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019425888419087833,
+      "loss": 1.1981,
+      "step": 3115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019425520960399887,
+      "loss": 1.268,
+      "step": 3116
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019425153387631717,
+      "loss": 1.0352,
+      "step": 3117
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001942478570078777,
+      "loss": 0.9873,
+      "step": 3118
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019424417899872501,
+      "loss": 1.2128,
+      "step": 3119
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019424049984890357,
+      "loss": 1.1387,
+      "step": 3120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000194236819558458,
+      "loss": 1.105,
+      "step": 3121
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001942331381274327,
+      "loss": 1.1279,
+      "step": 3122
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019422945555587235,
+      "loss": 1.2552,
+      "step": 3123
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001942257718438215,
+      "loss": 1.2241,
+      "step": 3124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019422208699132468,
+      "loss": 1.1437,
+      "step": 3125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001942184009984265,
+      "loss": 1.1657,
+      "step": 3126
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019421471386517165,
+      "loss": 1.0278,
+      "step": 3127
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019421102559160467,
+      "loss": 1.1094,
+      "step": 3128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019420733617777022,
+      "loss": 1.1334,
+      "step": 3129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019420364562371302,
+      "loss": 1.0799,
+      "step": 3130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019419995392947766,
+      "loss": 1.0748,
+      "step": 3131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019419626109510883,
+      "loss": 0.9226,
+      "step": 3132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019419256712065124,
+      "loss": 1.0101,
+      "step": 3133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019418887200614958,
+      "loss": 1.0934,
+      "step": 3134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019418517575164865,
+      "loss": 1.1301,
+      "step": 3135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001941814783571931,
+      "loss": 1.1914,
+      "step": 3136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019417777982282775,
+      "loss": 1.1913,
+      "step": 3137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001941740801485973,
+      "loss": 1.1118,
+      "step": 3138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019417037933454654,
+      "loss": 1.1619,
+      "step": 3139
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001941666773807203,
+      "loss": 1.1621,
+      "step": 3140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019416297428716335,
+      "loss": 1.1743,
+      "step": 3141
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001941592700539205,
+      "loss": 1.1773,
+      "step": 3142
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019415556468103663,
+      "loss": 1.109,
+      "step": 3143
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019415185816855658,
+      "loss": 1.0852,
+      "step": 3144
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019414815051652519,
+      "loss": 1.0737,
+      "step": 3145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019414444172498736,
+      "loss": 1.0537,
+      "step": 3146
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019414073179398794,
+      "loss": 1.0847,
+      "step": 3147
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019413702072357183,
+      "loss": 1.0102,
+      "step": 3148
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019413330851378397,
+      "loss": 1.1172,
+      "step": 3149
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019412959516466934,
+      "loss": 1.0538,
+      "step": 3150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019412588067627278,
+      "loss": 1.182,
+      "step": 3151
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001941221650486393,
+      "loss": 1.1033,
+      "step": 3152
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019411844828181392,
+      "loss": 0.9333,
+      "step": 3153
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019411473037584154,
+      "loss": 1.182,
+      "step": 3154
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001941110113307672,
+      "loss": 1.0011,
+      "step": 3155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001941072911466359,
+      "loss": 1.0151,
+      "step": 3156
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019410356982349267,
+      "loss": 1.1803,
+      "step": 3157
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019409984736138257,
+      "loss": 1.1228,
+      "step": 3158
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019409612376035065,
+      "loss": 1.0267,
+      "step": 3159
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019409239902044197,
+      "loss": 1.0116,
+      "step": 3160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001940886731417016,
+      "loss": 1.0406,
+      "step": 3161
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019408494612417463,
+      "loss": 1.1281,
+      "step": 3162
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001940812179679062,
+      "loss": 1.2818,
+      "step": 3163
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019407748867294143,
+      "loss": 1.1045,
+      "step": 3164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019407375823932542,
+      "loss": 1.1726,
+      "step": 3165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019407002666710336,
+      "loss": 1.0054,
+      "step": 3166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019406629395632037,
+      "loss": 1.2409,
+      "step": 3167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001940625601070217,
+      "loss": 1.1038,
+      "step": 3168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019405882511925246,
+      "loss": 1.0929,
+      "step": 3169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001940550889930579,
+      "loss": 1.2542,
+      "step": 3170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019405135172848324,
+      "loss": 1.0921,
+      "step": 3171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019404761332557374,
+      "loss": 1.1459,
+      "step": 3172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001940438737843746,
+      "loss": 1.0762,
+      "step": 3173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001940401331049311,
+      "loss": 1.0332,
+      "step": 3174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001940363912872885,
+      "loss": 1.0032,
+      "step": 3175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001940326483314921,
+      "loss": 1.0416,
+      "step": 3176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001940289042375872,
+      "loss": 1.101,
+      "step": 3177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019402515900561915,
+      "loss": 0.971,
+      "step": 3178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019402141263563321,
+      "loss": 1.1458,
+      "step": 3179
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019401766512767478,
+      "loss": 1.0686,
+      "step": 3180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019401391648178918,
+      "loss": 1.1388,
+      "step": 3181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001940101666980218,
+      "loss": 1.055,
+      "step": 3182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019400641577641805,
+      "loss": 1.0236,
+      "step": 3183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001940026637170233,
+      "loss": 1.1445,
+      "step": 3184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019399891051988296,
+      "loss": 1.0831,
+      "step": 3185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019399515618504244,
+      "loss": 1.0652,
+      "step": 3186
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019399140071254723,
+      "loss": 1.2038,
+      "step": 3187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019398764410244275,
+      "loss": 0.9836,
+      "step": 3188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019398388635477446,
+      "loss": 1.0857,
+      "step": 3189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019398012746958787,
+      "loss": 1.1307,
+      "step": 3190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019397636744692846,
+      "loss": 1.0145,
+      "step": 3191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019397260628684171,
+      "loss": 1.0247,
+      "step": 3192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019396884398937322,
+      "loss": 1.0474,
+      "step": 3193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019396508055456843,
+      "loss": 0.9898,
+      "step": 3194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019396131598247297,
+      "loss": 1.106,
+      "step": 3195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019395755027313234,
+      "loss": 1.1064,
+      "step": 3196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019395378342659217,
+      "loss": 1.0011,
+      "step": 3197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019395001544289804,
+      "loss": 0.9246,
+      "step": 3198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019394624632209552,
+      "loss": 1.1676,
+      "step": 3199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019394247606423028,
+      "loss": 1.0502,
+      "step": 3200
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 1.0920900106430054,
+      "eval_runtime": 844.9902,
+      "eval_samples_per_second": 28.474,
+      "eval_steps_per_second": 14.237,
+      "step": 3200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001939387046693479,
+      "loss": 1.057,
+      "step": 3201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019393493213749408,
+      "loss": 1.053,
+      "step": 3202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019393115846871447,
+      "loss": 1.152,
+      "step": 3203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019392738366305468,
+      "loss": 1.1656,
+      "step": 3204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019392360772056043,
+      "loss": 0.9668,
+      "step": 3205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001939198306412775,
+      "loss": 1.1738,
+      "step": 3206
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001939160524252515,
+      "loss": 0.9855,
+      "step": 3207
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019391227307252824,
+      "loss": 1.0651,
+      "step": 3208
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001939084925831534,
+      "loss": 1.1709,
+      "step": 3209
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019390471095717272,
+      "loss": 1.1075,
+      "step": 3210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019390092819463207,
+      "loss": 0.9611,
+      "step": 3211
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019389714429557717,
+      "loss": 1.0731,
+      "step": 3212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001938933592600538,
+      "loss": 1.1695,
+      "step": 3213
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001938895730881078,
+      "loss": 1.0119,
+      "step": 3214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000193885785779785,
+      "loss": 1.1301,
+      "step": 3215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019388199733513122,
+      "loss": 1.0596,
+      "step": 3216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019387820775419232,
+      "loss": 1.0105,
+      "step": 3217
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019387441703701416,
+      "loss": 1.0497,
+      "step": 3218
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019387062518364263,
+      "loss": 1.1516,
+      "step": 3219
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019386683219412362,
+      "loss": 1.1037,
+      "step": 3220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019386303806850306,
+      "loss": 1.1953,
+      "step": 3221
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019385924280682687,
+      "loss": 1.0255,
+      "step": 3222
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019385544640914093,
+      "loss": 1.1283,
+      "step": 3223
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019385164887549122,
+      "loss": 1.0284,
+      "step": 3224
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019384785020592373,
+      "loss": 1.1004,
+      "step": 3225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019384405040048441,
+      "loss": 1.1756,
+      "step": 3226
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019384024945921926,
+      "loss": 1.0442,
+      "step": 3227
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019383644738217427,
+      "loss": 1.0568,
+      "step": 3228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001938326441693955,
+      "loss": 1.1316,
+      "step": 3229
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019382883982092887,
+      "loss": 1.2433,
+      "step": 3230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019382503433682059,
+      "loss": 0.9837,
+      "step": 3231
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019382122771711657,
+      "loss": 1.1008,
+      "step": 3232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019381741996186297,
+      "loss": 1.0656,
+      "step": 3233
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019381361107110586,
+      "loss": 1.08,
+      "step": 3234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019380980104489134,
+      "loss": 1.1468,
+      "step": 3235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019380598988326552,
+      "loss": 1.0001,
+      "step": 3236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019380217758627451,
+      "loss": 1.1175,
+      "step": 3237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019379836415396446,
+      "loss": 1.0975,
+      "step": 3238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019379454958638154,
+      "loss": 1.1229,
+      "step": 3239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019379073388357191,
+      "loss": 1.0404,
+      "step": 3240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019378691704558177,
+      "loss": 1.1687,
+      "step": 3241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019378309907245727,
+      "loss": 1.125,
+      "step": 3242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019377927996424467,
+      "loss": 1.1039,
+      "step": 3243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019377545972099015,
+      "loss": 1.0538,
+      "step": 3244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019377163834274002,
+      "loss": 1.2528,
+      "step": 3245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019376781582954044,
+      "loss": 1.1194,
+      "step": 3246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019376399218143774,
+      "loss": 1.0652,
+      "step": 3247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019376016739847818,
+      "loss": 1.1091,
+      "step": 3248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019375634148070803,
+      "loss": 1.0257,
+      "step": 3249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019375251442817361,
+      "loss": 1.1894,
+      "step": 3250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019374868624092125,
+      "loss": 1.0178,
+      "step": 3251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019374485691899727,
+      "loss": 1.0366,
+      "step": 3252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001937410264624481,
+      "loss": 0.9366,
+      "step": 3253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019373719487131993,
+      "loss": 0.9447,
+      "step": 3254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019373336214565932,
+      "loss": 1.0903,
+      "step": 3255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019372952828551252,
+      "loss": 1.1042,
+      "step": 3256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019372569329092598,
+      "loss": 1.1533,
+      "step": 3257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019372185716194616,
+      "loss": 1.1439,
+      "step": 3258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019371801989861943,
+      "loss": 1.0503,
+      "step": 3259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019371418150099224,
+      "loss": 1.025,
+      "step": 3260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019371034196911108,
+      "loss": 1.158,
+      "step": 3261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019370650130302243,
+      "loss": 1.1832,
+      "step": 3262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001937026595027727,
+      "loss": 1.0048,
+      "step": 3263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019369881656840845,
+      "loss": 1.07,
+      "step": 3264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001936949724999762,
+      "loss": 0.9839,
+      "step": 3265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019369112729752242,
+      "loss": 1.2569,
+      "step": 3266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019368728096109372,
+      "loss": 0.9984,
+      "step": 3267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019368343349073657,
+      "loss": 1.0472,
+      "step": 3268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019367958488649765,
+      "loss": 1.0992,
+      "step": 3269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019367573514842342,
+      "loss": 1.0902,
+      "step": 3270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019367188427656058,
+      "loss": 1.0851,
+      "step": 3271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019366803227095566,
+      "loss": 1.1138,
+      "step": 3272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001936641791316553,
+      "loss": 1.2493,
+      "step": 3273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001936603248587062,
+      "loss": 1.1676,
+      "step": 3274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001936564694521549,
+      "loss": 1.2084,
+      "step": 3275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019365261291204818,
+      "loss": 1.0786,
+      "step": 3276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001936487552384326,
+      "loss": 1.0102,
+      "step": 3277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019364489643135493,
+      "loss": 1.3231,
+      "step": 3278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019364103649086186,
+      "loss": 1.0319,
+      "step": 3279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019363717541700007,
+      "loss": 1.0396,
+      "step": 3280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019363331320981634,
+      "loss": 1.0755,
+      "step": 3281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019362944986935744,
+      "loss": 1.0072,
+      "step": 3282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019362558539567,
+      "loss": 1.0103,
+      "step": 3283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019362171978880096,
+      "loss": 1.0965,
+      "step": 3284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000193617853048797,
+      "loss": 1.0966,
+      "step": 3285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019361398517570494,
+      "loss": 0.9887,
+      "step": 3286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019361011616957164,
+      "loss": 1.0843,
+      "step": 3287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019360624603044385,
+      "loss": 0.9756,
+      "step": 3288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019360237475836846,
+      "loss": 1.1278,
+      "step": 3289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001935985023533923,
+      "loss": 1.1243,
+      "step": 3290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001935946288155623,
+      "loss": 1.1025,
+      "step": 3291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019359075414492524,
+      "loss": 1.065,
+      "step": 3292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001935868783415281,
+      "loss": 1.0408,
+      "step": 3293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001935830014054178,
+      "loss": 1.1904,
+      "step": 3294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019357912333664118,
+      "loss": 1.1152,
+      "step": 3295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019357524413524523,
+      "loss": 0.9257,
+      "step": 3296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001935713638012769,
+      "loss": 1.0878,
+      "step": 3297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019356748233478318,
+      "loss": 1.1027,
+      "step": 3298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000193563599735811,
+      "loss": 1.1303,
+      "step": 3299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001935597160044074,
+      "loss": 1.1334,
+      "step": 3300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001935558311406193,
+      "loss": 1.1362,
+      "step": 3301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019355194514449387,
+      "loss": 1.1377,
+      "step": 3302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000193548058016078,
+      "loss": 1.0449,
+      "step": 3303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001935441697554188,
+      "loss": 1.0883,
+      "step": 3304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019354028036256332,
+      "loss": 1.1185,
+      "step": 3305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019353638983755862,
+      "loss": 0.8931,
+      "step": 3306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001935324981804518,
+      "loss": 1.1006,
+      "step": 3307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019352860539129,
+      "loss": 1.1003,
+      "step": 3308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019352471147012027,
+      "loss": 1.1878,
+      "step": 3309
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019352081641698976,
+      "loss": 1.0512,
+      "step": 3310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019351692023194565,
+      "loss": 1.1444,
+      "step": 3311
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019351302291503508,
+      "loss": 1.007,
+      "step": 3312
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001935091244663052,
+      "loss": 0.8646,
+      "step": 3313
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019350522488580321,
+      "loss": 1.0993,
+      "step": 3314
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001935013241735763,
+      "loss": 1.2049,
+      "step": 3315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019349742232967166,
+      "loss": 0.9133,
+      "step": 3316
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019349351935413659,
+      "loss": 1.045,
+      "step": 3317
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019348961524701824,
+      "loss": 1.1606,
+      "step": 3318
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019348571000836392,
+      "loss": 1.1096,
+      "step": 3319
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019348180363822087,
+      "loss": 1.0212,
+      "step": 3320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019347789613663642,
+      "loss": 1.0945,
+      "step": 3321
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001934739875036578,
+      "loss": 1.1912,
+      "step": 3322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001934700777393323,
+      "loss": 1.2063,
+      "step": 3323
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001934661668437073,
+      "loss": 1.0743,
+      "step": 3324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019346225481683016,
+      "loss": 1.1479,
+      "step": 3325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019345834165874814,
+      "loss": 1.0918,
+      "step": 3326
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019345442736950865,
+      "loss": 0.9858,
+      "step": 3327
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019345051194915908,
+      "loss": 1.0322,
+      "step": 3328
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019344659539774682,
+      "loss": 1.0811,
+      "step": 3329
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019344267771531922,
+      "loss": 1.1228,
+      "step": 3330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019343875890192377,
+      "loss": 1.1375,
+      "step": 3331
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019343483895760782,
+      "loss": 0.9796,
+      "step": 3332
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019343091788241889,
+      "loss": 1.0799,
+      "step": 3333
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019342699567640438,
+      "loss": 1.0027,
+      "step": 3334
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019342307233961178,
+      "loss": 1.0962,
+      "step": 3335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001934191478720886,
+      "loss": 1.0769,
+      "step": 3336
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019341522227388236,
+      "loss": 1.2401,
+      "step": 3337
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001934112955450405,
+      "loss": 1.1265,
+      "step": 3338
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019340736768561057,
+      "loss": 1.1508,
+      "step": 3339
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019340343869564012,
+      "loss": 1.1532,
+      "step": 3340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019339950857517676,
+      "loss": 1.0062,
+      "step": 3341
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019339557732426794,
+      "loss": 1.0763,
+      "step": 3342
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019339164494296135,
+      "loss": 0.9777,
+      "step": 3343
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019338771143130452,
+      "loss": 1.1863,
+      "step": 3344
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019338377678934507,
+      "loss": 1.0079,
+      "step": 3345
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019337984101713067,
+      "loss": 1.166,
+      "step": 3346
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019337590411470888,
+      "loss": 1.0829,
+      "step": 3347
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019337196608212737,
+      "loss": 1.1437,
+      "step": 3348
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019336802691943388,
+      "loss": 1.0744,
+      "step": 3349
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019336408662667598,
+      "loss": 1.0389,
+      "step": 3350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019336014520390143,
+      "loss": 1.043,
+      "step": 3351
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019335620265115787,
+      "loss": 0.9763,
+      "step": 3352
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001933522589684931,
+      "loss": 1.0498,
+      "step": 3353
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001933483141559548,
+      "loss": 1.0793,
+      "step": 3354
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019334436821359075,
+      "loss": 1.1102,
+      "step": 3355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019334042114144863,
+      "loss": 1.0284,
+      "step": 3356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019333647293957632,
+      "loss": 1.037,
+      "step": 3357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019333252360802155,
+      "loss": 1.1484,
+      "step": 3358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001933285731468321,
+      "loss": 0.978,
+      "step": 3359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001933246215560558,
+      "loss": 1.0006,
+      "step": 3360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001933206688357405,
+      "loss": 1.1357,
+      "step": 3361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019331671498593406,
+      "loss": 0.9815,
+      "step": 3362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019331276000668428,
+      "loss": 1.1572,
+      "step": 3363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019330880389803904,
+      "loss": 1.0969,
+      "step": 3364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019330484666004624,
+      "loss": 0.957,
+      "step": 3365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019330088829275376,
+      "loss": 1.0276,
+      "step": 3366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019329692879620954,
+      "loss": 1.1461,
+      "step": 3367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019329296817046146,
+      "loss": 0.9863,
+      "step": 3368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019328900641555748,
+      "loss": 1.1304,
+      "step": 3369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019328504353154555,
+      "loss": 1.2431,
+      "step": 3370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019328107951847363,
+      "loss": 1.1357,
+      "step": 3371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001932771143763897,
+      "loss": 1.0482,
+      "step": 3372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019327314810534176,
+      "loss": 1.0833,
+      "step": 3373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019326918070537784,
+      "loss": 1.0157,
+      "step": 3374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019326521217654586,
+      "loss": 1.1208,
+      "step": 3375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019326124251889396,
+      "loss": 1.1825,
+      "step": 3376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001932572717324701,
+      "loss": 1.0277,
+      "step": 3377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019325329981732242,
+      "loss": 1.1416,
+      "step": 3378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019324932677349896,
+      "loss": 1.1654,
+      "step": 3379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001932453526010478,
+      "loss": 1.1526,
+      "step": 3380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019324137730001702,
+      "loss": 1.0743,
+      "step": 3381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001932374008704548,
+      "loss": 0.9998,
+      "step": 3382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001932334233124092,
+      "loss": 1.0602,
+      "step": 3383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001932294446259284,
+      "loss": 1.1022,
+      "step": 3384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001932254648110605,
+      "loss": 1.0914,
+      "step": 3385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019322148386785378,
+      "loss": 0.9888,
+      "step": 3386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019321750179635634,
+      "loss": 1.176,
+      "step": 3387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019321351859661636,
+      "loss": 1.0764,
+      "step": 3388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019320953426868212,
+      "loss": 1.1707,
+      "step": 3389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019320554881260179,
+      "loss": 1.0183,
+      "step": 3390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019320156222842362,
+      "loss": 1.1262,
+      "step": 3391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019319757451619586,
+      "loss": 1.0683,
+      "step": 3392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001931935856759668,
+      "loss": 1.0411,
+      "step": 3393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019318959570778467,
+      "loss": 1.0954,
+      "step": 3394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001931856046116978,
+      "loss": 1.0365,
+      "step": 3395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001931816123877545,
+      "loss": 0.9688,
+      "step": 3396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019317761903600306,
+      "loss": 1.0993,
+      "step": 3397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019317362455649183,
+      "loss": 1.0006,
+      "step": 3398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019316962894926914,
+      "loss": 0.9175,
+      "step": 3399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019316563221438338,
+      "loss": 1.0453,
+      "step": 3400
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 1.0716861486434937,
+      "eval_runtime": 853.5879,
+      "eval_samples_per_second": 28.187,
+      "eval_steps_per_second": 14.093,
+      "step": 3400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019316163435188291,
+      "loss": 1.1467,
+      "step": 3401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001931576353618161,
+      "loss": 1.0506,
+      "step": 3402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019315363524423135,
+      "loss": 0.9897,
+      "step": 3403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019314963399917715,
+      "loss": 0.9979,
+      "step": 3404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001931456316267018,
+      "loss": 1.1712,
+      "step": 3405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019314162812685383,
+      "loss": 1.1198,
+      "step": 3406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001931376234996817,
+      "loss": 1.0017,
+      "step": 3407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019313361774523385,
+      "loss": 1.0798,
+      "step": 3408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019312961086355877,
+      "loss": 1.0917,
+      "step": 3409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019312560285470495,
+      "loss": 1.0893,
+      "step": 3410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019312159371872093,
+      "loss": 1.0983,
+      "step": 3411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019311758345565516,
+      "loss": 1.0146,
+      "step": 3412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019311357206555624,
+      "loss": 1.038,
+      "step": 3413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019310955954847275,
+      "loss": 1.1594,
+      "step": 3414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019310554590445315,
+      "loss": 1.0256,
+      "step": 3415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019310153113354615,
+      "loss": 1.1076,
+      "step": 3416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019309751523580024,
+      "loss": 0.9228,
+      "step": 3417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019309349821126402,
+      "loss": 1.1514,
+      "step": 3418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019308948005998622,
+      "loss": 0.9999,
+      "step": 3419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019308546078201537,
+      "loss": 1.0922,
+      "step": 3420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019308144037740016,
+      "loss": 1.1305,
+      "step": 3421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019307741884618923,
+      "loss": 1.0648,
+      "step": 3422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019307339618843122,
+      "loss": 1.089,
+      "step": 3423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019306937240417494,
+      "loss": 1.0016,
+      "step": 3424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019306534749346897,
+      "loss": 1.0433,
+      "step": 3425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019306132145636206,
+      "loss": 1.1934,
+      "step": 3426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019305729429290299,
+      "loss": 1.1799,
+      "step": 3427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001930532660031404,
+      "loss": 1.1137,
+      "step": 3428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019304923658712312,
+      "loss": 1.0996,
+      "step": 3429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001930452060448999,
+      "loss": 0.986,
+      "step": 3430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019304117437651957,
+      "loss": 0.9728,
+      "step": 3431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019303714158203084,
+      "loss": 1.0176,
+      "step": 3432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019303310766148255,
+      "loss": 0.9793,
+      "step": 3433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019302907261492357,
+      "loss": 1.0731,
+      "step": 3434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001930250364424027,
+      "loss": 0.9942,
+      "step": 3435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019302099914396875,
+      "loss": 1.149,
+      "step": 3436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019301696071967068,
+      "loss": 1.0811,
+      "step": 3437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001930129211695573,
+      "loss": 1.0341,
+      "step": 3438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001930088804936775,
+      "loss": 1.0554,
+      "step": 3439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019300483869208023,
+      "loss": 1.0621,
+      "step": 3440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019300079576481437,
+      "loss": 1.1503,
+      "step": 3441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001929967517119289,
+      "loss": 1.0676,
+      "step": 3442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019299270653347272,
+      "loss": 1.1235,
+      "step": 3443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001929886602294948,
+      "loss": 1.0439,
+      "step": 3444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001929846128000441,
+      "loss": 1.0261,
+      "step": 3445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019298056424516964,
+      "loss": 1.0488,
+      "step": 3446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001929765145649204,
+      "loss": 1.1577,
+      "step": 3447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019297246375934543,
+      "loss": 1.0806,
+      "step": 3448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001929684118284937,
+      "loss": 1.1822,
+      "step": 3449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001929643587724143,
+      "loss": 1.1242,
+      "step": 3450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019296030459115625,
+      "loss": 1.0035,
+      "step": 3451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019295624928476865,
+      "loss": 0.9429,
+      "step": 3452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019295219285330055,
+      "loss": 1.1046,
+      "step": 3453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001929481352968011,
+      "loss": 1.0966,
+      "step": 3454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019294407661531936,
+      "loss": 1.0636,
+      "step": 3455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019294001680890443,
+      "loss": 0.9899,
+      "step": 3456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001929359558776055,
+      "loss": 1.1226,
+      "step": 3457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019293189382147175,
+      "loss": 1.0403,
+      "step": 3458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019292783064055226,
+      "loss": 1.1344,
+      "step": 3459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019292376633489627,
+      "loss": 1.0684,
+      "step": 3460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019291970090455293,
+      "loss": 1.0704,
+      "step": 3461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001929156343495715,
+      "loss": 1.1219,
+      "step": 3462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001929115666700011,
+      "loss": 1.0733,
+      "step": 3463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019290749786589112,
+      "loss": 1.035,
+      "step": 3464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019290342793729063,
+      "loss": 1.0221,
+      "step": 3465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019289935688424903,
+      "loss": 1.0735,
+      "step": 3466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001928952847068155,
+      "loss": 1.1136,
+      "step": 3467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019289121140503938,
+      "loss": 1.1234,
+      "step": 3468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019288713697896993,
+      "loss": 1.1268,
+      "step": 3469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019288306142865654,
+      "loss": 1.2428,
+      "step": 3470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019287898475414843,
+      "loss": 1.0987,
+      "step": 3471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019287490695549505,
+      "loss": 1.0408,
+      "step": 3472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019287082803274566,
+      "loss": 0.8496,
+      "step": 3473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001928667479859497,
+      "loss": 0.9686,
+      "step": 3474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019286266681515647,
+      "loss": 0.9923,
+      "step": 3475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019285858452041548,
+      "loss": 1.2527,
+      "step": 3476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019285450110177602,
+      "loss": 0.9608,
+      "step": 3477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001928504165592876,
+      "loss": 1.1205,
+      "step": 3478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019284633089299965,
+      "loss": 0.9029,
+      "step": 3479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001928422441029616,
+      "loss": 1.0265,
+      "step": 3480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019283815618922288,
+      "loss": 1.082,
+      "step": 3481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019283406715183302,
+      "loss": 1.141,
+      "step": 3482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019282997699084147,
+      "loss": 0.9515,
+      "step": 3483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019282588570629776,
+      "loss": 1.094,
+      "step": 3484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019282179329825142,
+      "loss": 1.1793,
+      "step": 3485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019281769976675192,
+      "loss": 1.0929,
+      "step": 3486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001928136051118489,
+      "loss": 0.9976,
+      "step": 3487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019280950933359183,
+      "loss": 1.0098,
+      "step": 3488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019280541243203034,
+      "loss": 0.9926,
+      "step": 3489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000192801314407214,
+      "loss": 1.0901,
+      "step": 3490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019279721525919242,
+      "loss": 1.0639,
+      "step": 3491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019279311498801514,
+      "loss": 1.2033,
+      "step": 3492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019278901359373192,
+      "loss": 1.1765,
+      "step": 3493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001927849110763923,
+      "loss": 0.9782,
+      "step": 3494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019278080743604596,
+      "loss": 1.0982,
+      "step": 3495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019277670267274258,
+      "loss": 1.0381,
+      "step": 3496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001927725967865318,
+      "loss": 1.1775,
+      "step": 3497
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019276848977746339,
+      "loss": 1.0864,
+      "step": 3498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019276438164558699,
+      "loss": 1.1548,
+      "step": 3499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019276027239095234,
+      "loss": 1.0988,
+      "step": 3500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001927561620136092,
+      "loss": 1.0612,
+      "step": 3501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001927520505136073,
+      "loss": 1.1453,
+      "step": 3502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001927479378909964,
+      "loss": 1.181,
+      "step": 3503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019274382414582628,
+      "loss": 1.0555,
+      "step": 3504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001927397092781467,
+      "loss": 1.028,
+      "step": 3505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019273559328800754,
+      "loss": 1.1194,
+      "step": 3506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019273147617545857,
+      "loss": 0.9524,
+      "step": 3507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001927273579405496,
+      "loss": 1.0574,
+      "step": 3508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001927232385833305,
+      "loss": 1.0124,
+      "step": 3509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019271911810385116,
+      "loss": 1.0419,
+      "step": 3510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019271499650216136,
+      "loss": 0.954,
+      "step": 3511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019271087377831106,
+      "loss": 1.0101,
+      "step": 3512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019270674993235017,
+      "loss": 1.0418,
+      "step": 3513
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019270262496432852,
+      "loss": 1.0996,
+      "step": 3514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019269849887429614,
+      "loss": 1.1668,
+      "step": 3515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001926943716623029,
+      "loss": 1.0286,
+      "step": 3516
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019269024332839874,
+      "loss": 1.1343,
+      "step": 3517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001926861138726337,
+      "loss": 1.1964,
+      "step": 3518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001926819832950577,
+      "loss": 1.1769,
+      "step": 3519
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001926778515957208,
+      "loss": 1.1694,
+      "step": 3520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001926737187746729,
+      "loss": 1.1519,
+      "step": 3521
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001926695848319641,
+      "loss": 1.1607,
+      "step": 3522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001926654497676444,
+      "loss": 1.0662,
+      "step": 3523
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019266131358176386,
+      "loss": 1.1091,
+      "step": 3524
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019265717627437254,
+      "loss": 1.1525,
+      "step": 3525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019265303784552054,
+      "loss": 1.0758,
+      "step": 3526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001926488982952579,
+      "loss": 1.1325,
+      "step": 3527
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001926447576236348,
+      "loss": 1.116,
+      "step": 3528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019264061583070127,
+      "loss": 1.1746,
+      "step": 3529
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019263647291650746,
+      "loss": 1.011,
+      "step": 3530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019263232888110357,
+      "loss": 1.1648,
+      "step": 3531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001926281837245397,
+      "loss": 1.203,
+      "step": 3532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019262403744686602,
+      "loss": 1.0162,
+      "step": 3533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019261989004813274,
+      "loss": 0.9895,
+      "step": 3534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019261574152839005,
+      "loss": 1.0592,
+      "step": 3535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019261159188768817,
+      "loss": 1.0211,
+      "step": 3536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019260744112607728,
+      "loss": 1.0954,
+      "step": 3537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019260328924360767,
+      "loss": 1.048,
+      "step": 3538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019259913624032957,
+      "loss": 1.122,
+      "step": 3539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019259498211629325,
+      "loss": 0.9357,
+      "step": 3540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019259082687154896,
+      "loss": 1.1351,
+      "step": 3541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019258667050614704,
+      "loss": 1.2462,
+      "step": 3542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001925825130201378,
+      "loss": 1.0325,
+      "step": 3543
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001925783544135715,
+      "loss": 1.1324,
+      "step": 3544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019257419468649852,
+      "loss": 1.1316,
+      "step": 3545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019257003383896918,
+      "loss": 1.108,
+      "step": 3546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019256587187103385,
+      "loss": 1.0851,
+      "step": 3547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019256170878274292,
+      "loss": 1.15,
+      "step": 3548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019255754457414676,
+      "loss": 1.0791,
+      "step": 3549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019255337924529577,
+      "loss": 1.0511,
+      "step": 3550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019254921279624038,
+      "loss": 0.9547,
+      "step": 3551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000192545045227031,
+      "loss": 0.9466,
+      "step": 3552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019254087653771808,
+      "loss": 1.0698,
+      "step": 3553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019253670672835207,
+      "loss": 1.0563,
+      "step": 3554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019253253579898342,
+      "loss": 0.9909,
+      "step": 3555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019252836374966268,
+      "loss": 0.965,
+      "step": 3556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019252419058044028,
+      "loss": 1.0673,
+      "step": 3557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019252001629136673,
+      "loss": 1.0956,
+      "step": 3558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001925158408824926,
+      "loss": 0.9443,
+      "step": 3559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001925116643538684,
+      "loss": 1.0702,
+      "step": 3560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019250748670554465,
+      "loss": 1.0875,
+      "step": 3561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019250330793757196,
+      "loss": 1.1252,
+      "step": 3562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001924991280500009,
+      "loss": 1.1798,
+      "step": 3563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019249494704288203,
+      "loss": 0.9046,
+      "step": 3564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019249076491626598,
+      "loss": 0.9977,
+      "step": 3565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019248658167020336,
+      "loss": 1.0716,
+      "step": 3566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001924823973047448,
+      "loss": 0.9208,
+      "step": 3567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019247821181994095,
+      "loss": 1.0207,
+      "step": 3568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019247402521584248,
+      "loss": 0.9459,
+      "step": 3569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019246983749250004,
+      "loss": 1.017,
+      "step": 3570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019246564864996432,
+      "loss": 1.0529,
+      "step": 3571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019246145868828597,
+      "loss": 1.0225,
+      "step": 3572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001924572676075158,
+      "loss": 1.0144,
+      "step": 3573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019245307540770453,
+      "loss": 1.0438,
+      "step": 3574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019244888208890283,
+      "loss": 0.9887,
+      "step": 3575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019244468765116147,
+      "loss": 0.9543,
+      "step": 3576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019244049209453121,
+      "loss": 1.1155,
+      "step": 3577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019243629541906287,
+      "loss": 1.0581,
+      "step": 3578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019243209762480725,
+      "loss": 1.0451,
+      "step": 3579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001924278987118151,
+      "loss": 1.1545,
+      "step": 3580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001924236986801373,
+      "loss": 1.0211,
+      "step": 3581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019241949752982462,
+      "loss": 1.0969,
+      "step": 3582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019241529526092796,
+      "loss": 0.9938,
+      "step": 3583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019241109187349817,
+      "loss": 1.1423,
+      "step": 3584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019240688736758612,
+      "loss": 1.1605,
+      "step": 3585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001924026817432427,
+      "loss": 1.0947,
+      "step": 3586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001923984750005188,
+      "loss": 0.9577,
+      "step": 3587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019239426713946536,
+      "loss": 0.9852,
+      "step": 3588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019239005816013328,
+      "loss": 1.07,
+      "step": 3589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019238584806257353,
+      "loss": 1.0771,
+      "step": 3590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019238163684683705,
+      "loss": 1.0588,
+      "step": 3591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019237742451297483,
+      "loss": 0.9481,
+      "step": 3592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019237321106103784,
+      "loss": 1.0367,
+      "step": 3593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019236899649107704,
+      "loss": 0.9339,
+      "step": 3594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001923647808031435,
+      "loss": 1.0085,
+      "step": 3595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019236056399728822,
+      "loss": 1.2252,
+      "step": 3596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019235634607356222,
+      "loss": 1.06,
+      "step": 3597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001923521270320166,
+      "loss": 0.9961,
+      "step": 3598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019234790687270236,
+      "loss": 0.9829,
+      "step": 3599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019234368559567063,
+      "loss": 0.9332,
+      "step": 3600
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 1.0462063550949097,
+      "eval_runtime": 846.6254,
+      "eval_samples_per_second": 28.419,
+      "eval_steps_per_second": 14.209,
+      "step": 3600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019233946320097248,
+      "loss": 1.0344,
+      "step": 3601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000192335239688659,
+      "loss": 0.974,
+      "step": 3602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019233101505878133,
+      "loss": 1.0219,
+      "step": 3603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001923267893113906,
+      "loss": 1.0631,
+      "step": 3604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019232256244653794,
+      "loss": 1.1274,
+      "step": 3605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019231833446427455,
+      "loss": 1.0497,
+      "step": 3606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019231410536465156,
+      "loss": 0.9779,
+      "step": 3607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019230987514772015,
+      "loss": 1.0902,
+      "step": 3608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019230564381353155,
+      "loss": 0.8773,
+      "step": 3609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000192301411362137,
+      "loss": 1.0416,
+      "step": 3610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019229717779358765,
+      "loss": 1.0082,
+      "step": 3611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001922929431079348,
+      "loss": 1.0015,
+      "step": 3612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019228870730522967,
+      "loss": 1.0325,
+      "step": 3613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019228447038552356,
+      "loss": 0.9132,
+      "step": 3614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001922802323488677,
+      "loss": 1.1197,
+      "step": 3615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019227599319531345,
+      "loss": 1.0374,
+      "step": 3616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019227175292491208,
+      "loss": 1.0379,
+      "step": 3617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019226751153771492,
+      "loss": 1.1421,
+      "step": 3618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019226326903377329,
+      "loss": 0.9614,
+      "step": 3619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019225902541313854,
+      "loss": 1.0584,
+      "step": 3620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019225478067586206,
+      "loss": 1.2806,
+      "step": 3621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001922505348219952,
+      "loss": 1.2382,
+      "step": 3622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019224628785158935,
+      "loss": 1.093,
+      "step": 3623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019224203976469592,
+      "loss": 1.2161,
+      "step": 3624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019223779056136634,
+      "loss": 0.9348,
+      "step": 3625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019223354024165202,
+      "loss": 0.9119,
+      "step": 3626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001922292888056044,
+      "loss": 1.0229,
+      "step": 3627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019222503625327496,
+      "loss": 0.957,
+      "step": 3628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019222078258471513,
+      "loss": 1.0542,
+      "step": 3629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019221652779997643,
+      "loss": 0.9397,
+      "step": 3630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019221227189911033,
+      "loss": 0.9356,
+      "step": 3631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001922080148821684,
+      "loss": 1.0189,
+      "step": 3632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019220375674920207,
+      "loss": 0.9661,
+      "step": 3633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001921994975002629,
+      "loss": 1.029,
+      "step": 3634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019219523713540255,
+      "loss": 0.9846,
+      "step": 3635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019219097565467244,
+      "loss": 1.0429,
+      "step": 3636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019218671305812424,
+      "loss": 0.9622,
+      "step": 3637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001921824493458095,
+      "loss": 1.1022,
+      "step": 3638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019217818451777984,
+      "loss": 1.0574,
+      "step": 3639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019217391857408688,
+      "loss": 0.9119,
+      "step": 3640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019216965151478226,
+      "loss": 0.9637,
+      "step": 3641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019216538333991762,
+      "loss": 1.1038,
+      "step": 3642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019216111404954457,
+      "loss": 1.0068,
+      "step": 3643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019215684364371486,
+      "loss": 1.0384,
+      "step": 3644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019215257212248013,
+      "loss": 0.8676,
+      "step": 3645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001921482994858921,
+      "loss": 1.063,
+      "step": 3646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019214402573400246,
+      "loss": 1.1603,
+      "step": 3647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019213975086686293,
+      "loss": 1.1397,
+      "step": 3648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001921354748845253,
+      "loss": 1.0761,
+      "step": 3649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019213119778704128,
+      "loss": 1.0122,
+      "step": 3650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001921269195744627,
+      "loss": 1.065,
+      "step": 3651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019212264024684125,
+      "loss": 1.0969,
+      "step": 3652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019211835980422878,
+      "loss": 1.1283,
+      "step": 3653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019211407824667708,
+      "loss": 1.0622,
+      "step": 3654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019210979557423796,
+      "loss": 1.0647,
+      "step": 3655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001921055117869633,
+      "loss": 1.1335,
+      "step": 3656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001921012268849049,
+      "loss": 1.0467,
+      "step": 3657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019209694086811465,
+      "loss": 1.015,
+      "step": 3658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001920926537366444,
+      "loss": 1.1146,
+      "step": 3659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019208836549054604,
+      "loss": 0.945,
+      "step": 3660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019208407612987153,
+      "loss": 0.9834,
+      "step": 3661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019207978565467272,
+      "loss": 0.9372,
+      "step": 3662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019207549406500155,
+      "loss": 0.9405,
+      "step": 3663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019207120136091,
+      "loss": 1.0224,
+      "step": 3664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019206690754244998,
+      "loss": 1.0592,
+      "step": 3665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019206261260967347,
+      "loss": 0.9802,
+      "step": 3666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019205831656263247,
+      "loss": 1.1358,
+      "step": 3667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019205401940137898,
+      "loss": 1.0785,
+      "step": 3668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019204972112596498,
+      "loss": 0.9726,
+      "step": 3669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019204542173644252,
+      "loss": 0.941,
+      "step": 3670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019204112123286362,
+      "loss": 1.2271,
+      "step": 3671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019203681961528032,
+      "loss": 1.0313,
+      "step": 3672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019203251688374473,
+      "loss": 1.0052,
+      "step": 3673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001920282130383089,
+      "loss": 0.886,
+      "step": 3674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019202390807902494,
+      "loss": 0.99,
+      "step": 3675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001920196020059449,
+      "loss": 0.8889,
+      "step": 3676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019201529481912096,
+      "loss": 1.076,
+      "step": 3677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001920109865186052,
+      "loss": 1.0522,
+      "step": 3678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019200667710444978,
+      "loss": 1.1085,
+      "step": 3679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001920023665767069,
+      "loss": 0.9024,
+      "step": 3680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019199805493542868,
+      "loss": 1.0001,
+      "step": 3681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019199374218066732,
+      "loss": 1.0279,
+      "step": 3682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019198942831247505,
+      "loss": 1.1763,
+      "step": 3683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019198511333090403,
+      "loss": 1.1354,
+      "step": 3684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019198079723600653,
+      "loss": 1.082,
+      "step": 3685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019197648002783474,
+      "loss": 1.0873,
+      "step": 3686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019197216170644096,
+      "loss": 0.9289,
+      "step": 3687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019196784227187746,
+      "loss": 0.9368,
+      "step": 3688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019196352172419647,
+      "loss": 0.9993,
+      "step": 3689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019195920006345033,
+      "loss": 1.1853,
+      "step": 3690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019195487728969132,
+      "loss": 0.9467,
+      "step": 3691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019195055340297175,
+      "loss": 0.9649,
+      "step": 3692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019194622840334398,
+      "loss": 1.0438,
+      "step": 3693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019194190229086035,
+      "loss": 1.0719,
+      "step": 3694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019193757506557325,
+      "loss": 1.0849,
+      "step": 3695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019193324672753498,
+      "loss": 1.1921,
+      "step": 3696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000191928917276798,
+      "loss": 0.981,
+      "step": 3697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019192458671341464,
+      "loss": 0.9351,
+      "step": 3698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001919202550374374,
+      "loss": 1.0307,
+      "step": 3699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019191592224891864,
+      "loss": 1.0681,
+      "step": 3700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019191158834791082,
+      "loss": 0.9062,
+      "step": 3701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001919072533344664,
+      "loss": 1.0028,
+      "step": 3702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019190291720863788,
+      "loss": 0.9841,
+      "step": 3703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019189857997047767,
+      "loss": 1.0437,
+      "step": 3704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001918942416200383,
+      "loss": 1.1022,
+      "step": 3705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001918899021573723,
+      "loss": 1.0232,
+      "step": 3706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019188556158253217,
+      "loss": 1.1263,
+      "step": 3707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019188121989557045,
+      "loss": 1.0155,
+      "step": 3708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001918768770965397,
+      "loss": 1.0305,
+      "step": 3709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019187253318549244,
+      "loss": 1.1253,
+      "step": 3710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001918681881624813,
+      "loss": 0.9073,
+      "step": 3711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019186384202755883,
+      "loss": 1.166,
+      "step": 3712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019185949478077764,
+      "loss": 1.0675,
+      "step": 3713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019185514642219037,
+      "loss": 0.9438,
+      "step": 3714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019185079695184962,
+      "loss": 1.0524,
+      "step": 3715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019184644636980805,
+      "loss": 0.9474,
+      "step": 3716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019184209467611836,
+      "loss": 1.143,
+      "step": 3717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019183774187083312,
+      "loss": 0.9973,
+      "step": 3718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019183338795400508,
+      "loss": 1.0385,
+      "step": 3719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001918290329256869,
+      "loss": 1.0118,
+      "step": 3720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019182467678593136,
+      "loss": 1.0423,
+      "step": 3721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019182031953479114,
+      "loss": 1.0451,
+      "step": 3722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019181596117231892,
+      "loss": 1.136,
+      "step": 3723
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019181160169856755,
+      "loss": 1.065,
+      "step": 3724
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019180724111358973,
+      "loss": 1.0437,
+      "step": 3725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001918028794174383,
+      "loss": 1.0328,
+      "step": 3726
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019179851661016594,
+      "loss": 1.0492,
+      "step": 3727
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019179415269182556,
+      "loss": 1.0187,
+      "step": 3728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019178978766246996,
+      "loss": 1.1842,
+      "step": 3729
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019178542152215193,
+      "loss": 0.988,
+      "step": 3730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019178105427092434,
+      "loss": 1.0271,
+      "step": 3731
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019177668590884004,
+      "loss": 0.9069,
+      "step": 3732
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001917723164359519,
+      "loss": 1.0589,
+      "step": 3733
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019176794585231286,
+      "loss": 1.0319,
+      "step": 3734
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019176357415797573,
+      "loss": 1.1033,
+      "step": 3735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019175920135299345,
+      "loss": 1.1298,
+      "step": 3736
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000191754827437419,
+      "loss": 1.0026,
+      "step": 3737
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019175045241130525,
+      "loss": 0.9862,
+      "step": 3738
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001917460762747052,
+      "loss": 1.1468,
+      "step": 3739
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019174169902767179,
+      "loss": 0.8989,
+      "step": 3740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000191737320670258,
+      "loss": 0.9588,
+      "step": 3741
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019173294120251682,
+      "loss": 0.8869,
+      "step": 3742
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019172856062450125,
+      "loss": 0.9236,
+      "step": 3743
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019172417893626435,
+      "loss": 1.0753,
+      "step": 3744
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019171979613785914,
+      "loss": 1.0891,
+      "step": 3745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019171541222933863,
+      "loss": 0.8487,
+      "step": 3746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001917110272107559,
+      "loss": 1.0694,
+      "step": 3747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019170664108216406,
+      "loss": 0.9686,
+      "step": 3748
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019170225384361613,
+      "loss": 0.9249,
+      "step": 3749
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019169786549516523,
+      "loss": 1.138,
+      "step": 3750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019169347603686455,
+      "loss": 1.0633,
+      "step": 3751
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001916890854687671,
+      "loss": 1.0122,
+      "step": 3752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001916846937909261,
+      "loss": 1.0612,
+      "step": 3753
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019168030100339463,
+      "loss": 1.0863,
+      "step": 3754
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019167590710622597,
+      "loss": 1.1437,
+      "step": 3755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001916715120994732,
+      "loss": 0.9529,
+      "step": 3756
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019166711598318957,
+      "loss": 0.9574,
+      "step": 3757
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019166271875742825,
+      "loss": 1.0919,
+      "step": 3758
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001916583204222425,
+      "loss": 0.998,
+      "step": 3759
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019165392097768554,
+      "loss": 1.2614,
+      "step": 3760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019164952042381058,
+      "loss": 1.0893,
+      "step": 3761
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019164511876067092,
+      "loss": 0.9897,
+      "step": 3762
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019164071598831988,
+      "loss": 1.0783,
+      "step": 3763
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019163631210681063,
+      "loss": 1.0338,
+      "step": 3764
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019163190711619658,
+      "loss": 1.1914,
+      "step": 3765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019162750101653098,
+      "loss": 1.0637,
+      "step": 3766
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001916230938078672,
+      "loss": 1.0014,
+      "step": 3767
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019161868549025853,
+      "loss": 1.0992,
+      "step": 3768
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019161427606375839,
+      "loss": 1.0297,
+      "step": 3769
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019160986552842012,
+      "loss": 0.9527,
+      "step": 3770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019160545388429708,
+      "loss": 1.0691,
+      "step": 3771
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019160104113144272,
+      "loss": 1.0516,
+      "step": 3772
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001915966272699104,
+      "loss": 1.0687,
+      "step": 3773
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019159221229975355,
+      "loss": 1.1103,
+      "step": 3774
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019158779622102562,
+      "loss": 1.1914,
+      "step": 3775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019158337903378003,
+      "loss": 1.0514,
+      "step": 3776
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001915789607380703,
+      "loss": 0.9703,
+      "step": 3777
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019157454133394985,
+      "loss": 0.9798,
+      "step": 3778
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001915701208214722,
+      "loss": 0.9659,
+      "step": 3779
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001915656992006908,
+      "loss": 1.1045,
+      "step": 3780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019156127647165926,
+      "loss": 1.1718,
+      "step": 3781
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019155685263443103,
+      "loss": 0.9483,
+      "step": 3782
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019155242768905973,
+      "loss": 1.1009,
+      "step": 3783
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019154800163559882,
+      "loss": 1.146,
+      "step": 3784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019154357447410196,
+      "loss": 1.0493,
+      "step": 3785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019153914620462266,
+      "loss": 0.9993,
+      "step": 3786
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019153471682721454,
+      "loss": 0.8781,
+      "step": 3787
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019153028634193126,
+      "loss": 0.9832,
+      "step": 3788
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019152585474882638,
+      "loss": 1.0902,
+      "step": 3789
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019152142204795358,
+      "loss": 1.1147,
+      "step": 3790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019151698823936645,
+      "loss": 1.1492,
+      "step": 3791
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019151255332311876,
+      "loss": 1.1733,
+      "step": 3792
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019150811729926406,
+      "loss": 1.0254,
+      "step": 3793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001915036801678561,
+      "loss": 0.957,
+      "step": 3794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019149924192894867,
+      "loss": 1.0986,
+      "step": 3795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019149480258259533,
+      "loss": 1.0345,
+      "step": 3796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001914903621288499,
+      "loss": 0.9731,
+      "step": 3797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019148592056776611,
+      "loss": 1.0944,
+      "step": 3798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019148147789939773,
+      "loss": 0.9847,
+      "step": 3799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001914770341237985,
+      "loss": 1.1569,
+      "step": 3800
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 1.0284124612808228,
+      "eval_runtime": 844.6709,
+      "eval_samples_per_second": 28.484,
+      "eval_steps_per_second": 14.242,
+      "step": 3800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019147258924102225,
+      "loss": 1.0692,
+      "step": 3801
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019146814325112272,
+      "loss": 1.1191,
+      "step": 3802
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019146369615415375,
+      "loss": 0.9268,
+      "step": 3803
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001914592479501692,
+      "loss": 1.0052,
+      "step": 3804
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019145479863922287,
+      "loss": 0.9974,
+      "step": 3805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001914503482213686,
+      "loss": 1.1277,
+      "step": 3806
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019144589669666028,
+      "loss": 0.9946,
+      "step": 3807
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019144144406515174,
+      "loss": 1.0722,
+      "step": 3808
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019143699032689697,
+      "loss": 1.122,
+      "step": 3809
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001914325354819498,
+      "loss": 1.0779,
+      "step": 3810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019142807953036414,
+      "loss": 1.0159,
+      "step": 3811
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019142362247219396,
+      "loss": 1.0387,
+      "step": 3812
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001914191643074932,
+      "loss": 0.9668,
+      "step": 3813
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001914147050363158,
+      "loss": 0.979,
+      "step": 3814
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019141024465871577,
+      "loss": 1.2342,
+      "step": 3815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019140578317474705,
+      "loss": 0.9985,
+      "step": 3816
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019140132058446366,
+      "loss": 0.9996,
+      "step": 3817
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001913968568879196,
+      "loss": 0.908,
+      "step": 3818
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001913923920851689,
+      "loss": 1.254,
+      "step": 3819
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019138792617626562,
+      "loss": 1.0811,
+      "step": 3820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019138345916126377,
+      "loss": 1.0658,
+      "step": 3821
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019137899104021747,
+      "loss": 1.0617,
+      "step": 3822
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019137452181318076,
+      "loss": 1.0065,
+      "step": 3823
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019137005148020774,
+      "loss": 1.0117,
+      "step": 3824
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019136558004135252,
+      "loss": 0.8707,
+      "step": 3825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019136110749666925,
+      "loss": 0.9782,
+      "step": 3826
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000191356633846212,
+      "loss": 0.9121,
+      "step": 3827
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019135215909003495,
+      "loss": 1.0509,
+      "step": 3828
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019134768322819225,
+      "loss": 0.9673,
+      "step": 3829
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001913432062607381,
+      "loss": 1.008,
+      "step": 3830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019133872818772666,
+      "loss": 1.1044,
+      "step": 3831
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019133424900921213,
+      "loss": 0.8671,
+      "step": 3832
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019132976872524872,
+      "loss": 1.046,
+      "step": 3833
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001913252873358907,
+      "loss": 0.8531,
+      "step": 3834
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019132080484119223,
+      "loss": 0.9255,
+      "step": 3835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019131632124120765,
+      "loss": 1.0342,
+      "step": 3836
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019131183653599117,
+      "loss": 0.9437,
+      "step": 3837
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019130735072559707,
+      "loss": 1.0382,
+      "step": 3838
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019130286381007964,
+      "loss": 0.9301,
+      "step": 3839
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019129837578949323,
+      "loss": 0.9509,
+      "step": 3840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019129388666389217,
+      "loss": 0.9743,
+      "step": 3841
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001912893964333307,
+      "loss": 0.9347,
+      "step": 3842
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019128490509786326,
+      "loss": 0.9321,
+      "step": 3843
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019128041265754417,
+      "loss": 1.003,
+      "step": 3844
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001912759191124278,
+      "loss": 0.9689,
+      "step": 3845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019127142446256855,
+      "loss": 1.1861,
+      "step": 3846
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001912669287080208,
+      "loss": 1.054,
+      "step": 3847
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019126243184883898,
+      "loss": 0.8804,
+      "step": 3848
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019125793388507752,
+      "loss": 0.9738,
+      "step": 3849
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019125343481679087,
+      "loss": 1.0167,
+      "step": 3850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019124893464403346,
+      "loss": 0.9549,
+      "step": 3851
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001912444333668598,
+      "loss": 1.0799,
+      "step": 3852
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001912399309853243,
+      "loss": 0.9105,
+      "step": 3853
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019123542749948147,
+      "loss": 1.0237,
+      "step": 3854
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019123092290938587,
+      "loss": 1.0471,
+      "step": 3855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019122641721509197,
+      "loss": 1.0597,
+      "step": 3856
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019122191041665436,
+      "loss": 0.9414,
+      "step": 3857
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001912174025141275,
+      "loss": 1.0573,
+      "step": 3858
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019121289350756607,
+      "loss": 1.1582,
+      "step": 3859
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019120838339702453,
+      "loss": 0.8603,
+      "step": 3860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019120387218255752,
+      "loss": 1.0839,
+      "step": 3861
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001911993598642196,
+      "loss": 1.1865,
+      "step": 3862
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019119484644206547,
+      "loss": 0.9953,
+      "step": 3863
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019119033191614967,
+      "loss": 0.9777,
+      "step": 3864
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001911858162865269,
+      "loss": 0.9597,
+      "step": 3865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019118129955325176,
+      "loss": 1.0399,
+      "step": 3866
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019117678171637898,
+      "loss": 1.0074,
+      "step": 3867
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019117226277596319,
+      "loss": 1.0126,
+      "step": 3868
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001911677427320591,
+      "loss": 1.0872,
+      "step": 3869
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019116322158472144,
+      "loss": 1.0233,
+      "step": 3870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019115869933400488,
+      "loss": 1.0739,
+      "step": 3871
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001911541759799642,
+      "loss": 0.8729,
+      "step": 3872
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019114965152265413,
+      "loss": 1.1225,
+      "step": 3873
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019114512596212944,
+      "loss": 1.0517,
+      "step": 3874
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019114059929844487,
+      "loss": 1.1168,
+      "step": 3875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019113607153165528,
+      "loss": 0.9634,
+      "step": 3876
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019113154266181542,
+      "loss": 0.9531,
+      "step": 3877
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001911270126889801,
+      "loss": 0.9806,
+      "step": 3878
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019112248161320413,
+      "loss": 1.074,
+      "step": 3879
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019111794943454244,
+      "loss": 0.9431,
+      "step": 3880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001911134161530498,
+      "loss": 1.1319,
+      "step": 3881
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019110888176878107,
+      "loss": 1.1149,
+      "step": 3882
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019110434628179118,
+      "loss": 1.0296,
+      "step": 3883
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019109980969213503,
+      "loss": 0.9882,
+      "step": 3884
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001910952719998675,
+      "loss": 0.9413,
+      "step": 3885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019109073320504352,
+      "loss": 1.0402,
+      "step": 3886
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000191086193307718,
+      "loss": 1.1124,
+      "step": 3887
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001910816523079459,
+      "loss": 1.0121,
+      "step": 3888
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019107711020578222,
+      "loss": 0.9588,
+      "step": 3889
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019107256700128192,
+      "loss": 0.9018,
+      "step": 3890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019106802269449995,
+      "loss": 1.0994,
+      "step": 3891
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019106347728549135,
+      "loss": 0.9928,
+      "step": 3892
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019105893077431109,
+      "loss": 1.0894,
+      "step": 3893
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019105438316101422,
+      "loss": 1.0351,
+      "step": 3894
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019104983444565582,
+      "loss": 1.004,
+      "step": 3895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001910452846282909,
+      "loss": 1.0875,
+      "step": 3896
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019104073370897459,
+      "loss": 1.0228,
+      "step": 3897
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019103618168776185,
+      "loss": 0.9486,
+      "step": 3898
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019103162856470789,
+      "loss": 1.0668,
+      "step": 3899
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019102707433986776,
+      "loss": 1.1191,
+      "step": 3900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019102251901329656,
+      "loss": 0.8782,
+      "step": 3901
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019101796258504951,
+      "loss": 1.0519,
+      "step": 3902
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001910134050551817,
+      "loss": 0.8821,
+      "step": 3903
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001910088464237483,
+      "loss": 0.8848,
+      "step": 3904
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019100428669080447,
+      "loss": 0.9808,
+      "step": 3905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019099972585640544,
+      "loss": 0.9899,
+      "step": 3906
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019099516392060637,
+      "loss": 1.0979,
+      "step": 3907
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019099060088346247,
+      "loss": 1.1008,
+      "step": 3908
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000190986036745029,
+      "loss": 0.9751,
+      "step": 3909
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001909814715053612,
+      "loss": 0.9166,
+      "step": 3910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019097690516451431,
+      "loss": 0.8834,
+      "step": 3911
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001909723377225436,
+      "loss": 0.9766,
+      "step": 3912
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019096776917950435,
+      "loss": 1.2002,
+      "step": 3913
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019096319953545185,
+      "loss": 0.9769,
+      "step": 3914
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019095862879044142,
+      "loss": 0.9617,
+      "step": 3915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019095405694452837,
+      "loss": 1.0476,
+      "step": 3916
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019094948399776804,
+      "loss": 1.1321,
+      "step": 3917
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019094490995021576,
+      "loss": 0.9789,
+      "step": 3918
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019094033480192697,
+      "loss": 1.0187,
+      "step": 3919
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019093575855295693,
+      "loss": 0.9151,
+      "step": 3920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019093118120336108,
+      "loss": 1.0211,
+      "step": 3921
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019092660275319486,
+      "loss": 1.061,
+      "step": 3922
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001909220232025136,
+      "loss": 0.989,
+      "step": 3923
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019091744255137283,
+      "loss": 1.0248,
+      "step": 3924
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019091286079982788,
+      "loss": 0.9934,
+      "step": 3925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001909082779479343,
+      "loss": 1.0908,
+      "step": 3926
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001909036939957475,
+      "loss": 0.9434,
+      "step": 3927
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019089910894332302,
+      "loss": 0.929,
+      "step": 3928
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019089452279071628,
+      "loss": 1.0731,
+      "step": 3929
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019088993553798282,
+      "loss": 0.9836,
+      "step": 3930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019088534718517813,
+      "loss": 1.1103,
+      "step": 3931
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019088075773235783,
+      "loss": 1.0644,
+      "step": 3932
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001908761671795774,
+      "loss": 1.0564,
+      "step": 3933
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019087157552689243,
+      "loss": 1.0663,
+      "step": 3934
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019086698277435846,
+      "loss": 0.9552,
+      "step": 3935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001908623889220311,
+      "loss": 1.0949,
+      "step": 3936
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019085779396996595,
+      "loss": 1.1169,
+      "step": 3937
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019085319791821862,
+      "loss": 1.0862,
+      "step": 3938
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019084860076684477,
+      "loss": 0.9381,
+      "step": 3939
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019084400251589997,
+      "loss": 1.0678,
+      "step": 3940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019083940316543994,
+      "loss": 1.0736,
+      "step": 3941
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019083480271552032,
+      "loss": 0.8937,
+      "step": 3942
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019083020116619678,
+      "loss": 1.1058,
+      "step": 3943
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019082559851752503,
+      "loss": 1.0089,
+      "step": 3944
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001908209947695608,
+      "loss": 0.9709,
+      "step": 3945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019081638992235974,
+      "loss": 1.0073,
+      "step": 3946
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019081178397597768,
+      "loss": 0.9122,
+      "step": 3947
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019080717693047028,
+      "loss": 1.0871,
+      "step": 3948
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019080256878589336,
+      "loss": 1.0064,
+      "step": 3949
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019079795954230267,
+      "loss": 0.9618,
+      "step": 3950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000190793349199754,
+      "loss": 1.2609,
+      "step": 3951
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019078873775830313,
+      "loss": 1.0517,
+      "step": 3952
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001907841252180059,
+      "loss": 1.0045,
+      "step": 3953
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019077951157891817,
+      "loss": 1.0031,
+      "step": 3954
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019077489684109572,
+      "loss": 0.9489,
+      "step": 3955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019077028100459437,
+      "loss": 0.9885,
+      "step": 3956
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019076566406947012,
+      "loss": 1.1361,
+      "step": 3957
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019076104603577872,
+      "loss": 1.0526,
+      "step": 3958
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019075642690357614,
+      "loss": 1.0916,
+      "step": 3959
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019075180667291826,
+      "loss": 0.9713,
+      "step": 3960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019074718534386098,
+      "loss": 1.0537,
+      "step": 3961
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019074256291646028,
+      "loss": 0.9735,
+      "step": 3962
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019073793939077208,
+      "loss": 0.8726,
+      "step": 3963
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019073331476685234,
+      "loss": 0.9387,
+      "step": 3964
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019072868904475705,
+      "loss": 0.9787,
+      "step": 3965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019072406222454217,
+      "loss": 0.9895,
+      "step": 3966
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019071943430626372,
+      "loss": 1.0434,
+      "step": 3967
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001907148052899777,
+      "loss": 1.222,
+      "step": 3968
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019071017517574012,
+      "loss": 0.9867,
+      "step": 3969
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019070554396360707,
+      "loss": 1.1127,
+      "step": 3970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019070091165363458,
+      "loss": 0.9163,
+      "step": 3971
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019069627824587867,
+      "loss": 0.9801,
+      "step": 3972
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001906916437403955,
+      "loss": 1.0048,
+      "step": 3973
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001906870081372411,
+      "loss": 1.1581,
+      "step": 3974
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019068237143647166,
+      "loss": 1.0176,
+      "step": 3975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001906777336381432,
+      "loss": 0.9728,
+      "step": 3976
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019067309474231188,
+      "loss": 1.1814,
+      "step": 3977
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019066845474903388,
+      "loss": 0.8797,
+      "step": 3978
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019066381365836533,
+      "loss": 0.9827,
+      "step": 3979
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001906591714703624,
+      "loss": 0.9732,
+      "step": 3980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019065452818508132,
+      "loss": 1.0577,
+      "step": 3981
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019064988380257822,
+      "loss": 0.9607,
+      "step": 3982
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019064523832290937,
+      "loss": 1.1409,
+      "step": 3983
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019064059174613097,
+      "loss": 1.0741,
+      "step": 3984
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019063594407229927,
+      "loss": 1.174,
+      "step": 3985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019063129530147053,
+      "loss": 0.9464,
+      "step": 3986
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019062664543370098,
+      "loss": 0.9867,
+      "step": 3987
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001906219944690469,
+      "loss": 0.915,
+      "step": 3988
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019061734240756464,
+      "loss": 0.9442,
+      "step": 3989
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019061268924931043,
+      "loss": 0.9718,
+      "step": 3990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019060803499434067,
+      "loss": 0.9064,
+      "step": 3991
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019060337964271165,
+      "loss": 1.0964,
+      "step": 3992
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001905987231944797,
+      "loss": 1.1744,
+      "step": 3993
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019059406564970118,
+      "loss": 1.1332,
+      "step": 3994
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019058940700843248,
+      "loss": 0.9942,
+      "step": 3995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019058474727072996,
+      "loss": 1.1864,
+      "step": 3996
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019058008643665008,
+      "loss": 0.9033,
+      "step": 3997
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001905754245062492,
+      "loss": 0.9648,
+      "step": 3998
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019057076147958375,
+      "loss": 1.0346,
+      "step": 3999
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019056609735671018,
+      "loss": 1.146,
+      "step": 4000
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 1.0092287063598633,
+      "eval_runtime": 849.6129,
+      "eval_samples_per_second": 28.319,
+      "eval_steps_per_second": 14.159,
+      "step": 4000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001905614321376849,
+      "loss": 1.0128,
+      "step": 4001
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019055676582256444,
+      "loss": 1.0876,
+      "step": 4002
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019055209841140525,
+      "loss": 1.0209,
+      "step": 4003
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019054742990426383,
+      "loss": 0.9088,
+      "step": 4004
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019054276030119663,
+      "loss": 0.937,
+      "step": 4005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019053808960226023,
+      "loss": 1.0074,
+      "step": 4006
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019053341780751116,
+      "loss": 0.9237,
+      "step": 4007
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019052874491700593,
+      "loss": 0.921,
+      "step": 4008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001905240709308011,
+      "loss": 0.9931,
+      "step": 4009
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019051939584895328,
+      "loss": 0.8661,
+      "step": 4010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000190514719671519,
+      "loss": 0.9725,
+      "step": 4011
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001905100423985549,
+      "loss": 0.9658,
+      "step": 4012
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001905053640301176,
+      "loss": 0.9382,
+      "step": 4013
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019050068456626367,
+      "loss": 1.05,
+      "step": 4014
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019049600400704978,
+      "loss": 0.928,
+      "step": 4015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001904913223525326,
+      "loss": 0.9643,
+      "step": 4016
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019048663960276874,
+      "loss": 0.9391,
+      "step": 4017
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019048195575781493,
+      "loss": 1.1249,
+      "step": 4018
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019047727081772782,
+      "loss": 1.0511,
+      "step": 4019
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019047258478256413,
+      "loss": 1.0085,
+      "step": 4020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001904678976523806,
+      "loss": 0.9143,
+      "step": 4021
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019046320942723392,
+      "loss": 0.8507,
+      "step": 4022
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001904585201071809,
+      "loss": 1.0112,
+      "step": 4023
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001904538296922782,
+      "loss": 1.1175,
+      "step": 4024
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019044913818258262,
+      "loss": 0.8132,
+      "step": 4025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000190444445578151,
+      "loss": 0.9932,
+      "step": 4026
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019043975187904008,
+      "loss": 0.9078,
+      "step": 4027
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001904350570853067,
+      "loss": 0.9561,
+      "step": 4028
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019043036119700766,
+      "loss": 0.928,
+      "step": 4029
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001904256642141998,
+      "loss": 1.0588,
+      "step": 4030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019042096613693996,
+      "loss": 1.0687,
+      "step": 4031
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019041626696528503,
+      "loss": 0.9369,
+      "step": 4032
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019041156669929188,
+      "loss": 1.0472,
+      "step": 4033
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019040686533901738,
+      "loss": 1.0708,
+      "step": 4034
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019040216288451845,
+      "loss": 1.1213,
+      "step": 4035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019039745933585202,
+      "loss": 1.0089,
+      "step": 4036
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019039275469307494,
+      "loss": 0.8902,
+      "step": 4037
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019038804895624422,
+      "loss": 0.9091,
+      "step": 4038
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019038334212541683,
+      "loss": 1.041,
+      "step": 4039
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019037863420064973,
+      "loss": 1.1809,
+      "step": 4040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019037392518199986,
+      "loss": 0.965,
+      "step": 4041
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019036921506952422,
+      "loss": 1.0663,
+      "step": 4042
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019036450386327984,
+      "loss": 1.2294,
+      "step": 4043
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019035979156332378,
+      "loss": 1.1152,
+      "step": 4044
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000190355078169713,
+      "loss": 1.0931,
+      "step": 4045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001903503636825046,
+      "loss": 0.986,
+      "step": 4046
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019034564810175558,
+      "loss": 1.0883,
+      "step": 4047
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019034093142752308,
+      "loss": 1.0075,
+      "step": 4048
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019033621365986415,
+      "loss": 0.9116,
+      "step": 4049
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019033149479883593,
+      "loss": 1.0325,
+      "step": 4050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019032677484449547,
+      "loss": 0.9356,
+      "step": 4051
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019032205379689997,
+      "loss": 1.0435,
+      "step": 4052
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019031733165610654,
+      "loss": 0.975,
+      "step": 4053
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019031260842217228,
+      "loss": 1.072,
+      "step": 4054
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019030788409515442,
+      "loss": 0.9761,
+      "step": 4055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019030315867511017,
+      "loss": 1.0225,
+      "step": 4056
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019029843216209665,
+      "loss": 1.0785,
+      "step": 4057
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001902937045561711,
+      "loss": 0.8986,
+      "step": 4058
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019028897585739074,
+      "loss": 1.0376,
+      "step": 4059
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019028424606581278,
+      "loss": 0.9006,
+      "step": 4060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019027951518149448,
+      "loss": 0.9989,
+      "step": 4061
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019027478320449314,
+      "loss": 0.9098,
+      "step": 4062
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019027005013486596,
+      "loss": 1.0252,
+      "step": 4063
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001902653159726703,
+      "loss": 1.027,
+      "step": 4064
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019026058071796337,
+      "loss": 1.0379,
+      "step": 4065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001902558443708026,
+      "loss": 1.0254,
+      "step": 4066
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001902511069312452,
+      "loss": 0.9162,
+      "step": 4067
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019024636839934855,
+      "loss": 1.168,
+      "step": 4068
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019024162877517004,
+      "loss": 1.0539,
+      "step": 4069
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000190236888058767,
+      "loss": 1.1106,
+      "step": 4070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001902321462501968,
+      "loss": 0.9683,
+      "step": 4071
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019022740334951688,
+      "loss": 0.7768,
+      "step": 4072
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001902226593567846,
+      "loss": 1.0579,
+      "step": 4073
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019021791427205738,
+      "loss": 1.1112,
+      "step": 4074
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019021316809539264,
+      "loss": 1.0454,
+      "step": 4075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019020842082684788,
+      "loss": 1.0213,
+      "step": 4076
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019020367246648048,
+      "loss": 0.9152,
+      "step": 4077
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019019892301434802,
+      "loss": 0.9715,
+      "step": 4078
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019019417247050785,
+      "loss": 0.9433,
+      "step": 4079
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019018942083501758,
+      "loss": 1.0704,
+      "step": 4080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019018466810793468,
+      "loss": 0.9485,
+      "step": 4081
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019017991428931663,
+      "loss": 0.9724,
+      "step": 4082
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019017515937922103,
+      "loss": 1.0827,
+      "step": 4083
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019017040337770543,
+      "loss": 1.1208,
+      "step": 4084
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001901656462848273,
+      "loss": 0.8624,
+      "step": 4085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019016088810064433,
+      "loss": 0.9649,
+      "step": 4086
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001901561288252141,
+      "loss": 1.0864,
+      "step": 4087
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019015136845859415,
+      "loss": 1.2034,
+      "step": 4088
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001901466070008421,
+      "loss": 1.0551,
+      "step": 4089
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019014184445201565,
+      "loss": 0.9019,
+      "step": 4090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019013708081217238,
+      "loss": 1.1146,
+      "step": 4091
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019013231608136996,
+      "loss": 0.9925,
+      "step": 4092
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001901275502596661,
+      "loss": 0.9437,
+      "step": 4093
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001901227833471184,
+      "loss": 0.9916,
+      "step": 4094
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019011801534378462,
+      "loss": 0.9644,
+      "step": 4095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019011324624972244,
+      "loss": 1.1118,
+      "step": 4096
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001901084760649896,
+      "loss": 0.9111,
+      "step": 4097
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019010370478964383,
+      "loss": 0.9302,
+      "step": 4098
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019009893242374285,
+      "loss": 1.0585,
+      "step": 4099
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019009415896734447,
+      "loss": 1.0481,
+      "step": 4100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019008938442050644,
+      "loss": 1.0457,
+      "step": 4101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001900846087832866,
+      "loss": 1.0454,
+      "step": 4102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019007983205574265,
+      "loss": 0.9182,
+      "step": 4103
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019007505423793245,
+      "loss": 0.901,
+      "step": 4104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019007027532991384,
+      "loss": 1.041,
+      "step": 4105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019006549533174465,
+      "loss": 1.0371,
+      "step": 4106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019006071424348275,
+      "loss": 0.8508,
+      "step": 4107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019005593206518597,
+      "loss": 1.0068,
+      "step": 4108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019005114879691224,
+      "loss": 0.9752,
+      "step": 4109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019004636443871944,
+      "loss": 0.9415,
+      "step": 4110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019004157899066545,
+      "loss": 0.959,
+      "step": 4111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001900367924528082,
+      "loss": 1.1121,
+      "step": 4112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019003200482520564,
+      "loss": 0.9275,
+      "step": 4113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001900272161079157,
+      "loss": 1.0844,
+      "step": 4114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019002242630099633,
+      "loss": 0.9965,
+      "step": 4115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019001763540450552,
+      "loss": 1.0158,
+      "step": 4116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019001284341850125,
+      "loss": 1.0039,
+      "step": 4117
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019000805034304153,
+      "loss": 0.9078,
+      "step": 4118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019000325617818438,
+      "loss": 0.9352,
+      "step": 4119
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018999846092398779,
+      "loss": 0.8697,
+      "step": 4120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018999366458050983,
+      "loss": 1.0056,
+      "step": 4121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001899888671478085,
+      "loss": 1.0566,
+      "step": 4122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018998406862594193,
+      "loss": 1.0999,
+      "step": 4123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001899792690149682,
+      "loss": 1.0435,
+      "step": 4124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018997446831494534,
+      "loss": 0.9974,
+      "step": 4125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001899696665259315,
+      "loss": 0.9922,
+      "step": 4126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001899648636479848,
+      "loss": 1.0019,
+      "step": 4127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018996005968116334,
+      "loss": 0.9487,
+      "step": 4128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018995525462552526,
+      "loss": 1.069,
+      "step": 4129
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018995044848112876,
+      "loss": 1.0885,
+      "step": 4130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000189945641248032,
+      "loss": 0.9978,
+      "step": 4131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001899408329262931,
+      "loss": 1.0339,
+      "step": 4132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018993602351597037,
+      "loss": 1.0339,
+      "step": 4133
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018993121301712193,
+      "loss": 0.8591,
+      "step": 4134
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018992640142980602,
+      "loss": 0.8985,
+      "step": 4135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018992158875408093,
+      "loss": 0.9628,
+      "step": 4136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018991677499000483,
+      "loss": 1.0842,
+      "step": 4137
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000189911960137636,
+      "loss": 0.9816,
+      "step": 4138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001899071441970328,
+      "loss": 1.0378,
+      "step": 4139
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001899023271682534,
+      "loss": 1.0693,
+      "step": 4140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018989750905135618,
+      "loss": 1.0266,
+      "step": 4141
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018989268984639942,
+      "loss": 1.0267,
+      "step": 4142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018988786955344148,
+      "loss": 1.0241,
+      "step": 4143
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018988304817254065,
+      "loss": 1.0087,
+      "step": 4144
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018987822570375534,
+      "loss": 1.016,
+      "step": 4145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018987340214714392,
+      "loss": 0.9822,
+      "step": 4146
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001898685775027647,
+      "loss": 0.9831,
+      "step": 4147
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018986375177067614,
+      "loss": 0.9742,
+      "step": 4148
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018985892495093661,
+      "loss": 1.034,
+      "step": 4149
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018985409704360456,
+      "loss": 0.9638,
+      "step": 4150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018984926804873842,
+      "loss": 0.9327,
+      "step": 4151
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018984443796639662,
+      "loss": 1.0087,
+      "step": 4152
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018983960679663762,
+      "loss": 0.9028,
+      "step": 4153
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018983477453951992,
+      "loss": 1.0778,
+      "step": 4154
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018982994119510197,
+      "loss": 1.0517,
+      "step": 4155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001898251067634423,
+      "loss": 1.1624,
+      "step": 4156
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001898202712445994,
+      "loss": 1.0591,
+      "step": 4157
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018981543463863183,
+      "loss": 1.043,
+      "step": 4158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001898105969455981,
+      "loss": 0.8838,
+      "step": 4159
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018980575816555673,
+      "loss": 0.971,
+      "step": 4160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018980091829856637,
+      "loss": 0.9973,
+      "step": 4161
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018979607734468553,
+      "loss": 0.9556,
+      "step": 4162
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001897912353039728,
+      "loss": 1.0951,
+      "step": 4163
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018978639217648687,
+      "loss": 1.0012,
+      "step": 4164
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018978154796228625,
+      "loss": 1.0018,
+      "step": 4165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018977670266142962,
+      "loss": 1.0277,
+      "step": 4166
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018977185627397562,
+      "loss": 1.014,
+      "step": 4167
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018976700879998293,
+      "loss": 1.0896,
+      "step": 4168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001897621602395102,
+      "loss": 0.902,
+      "step": 4169
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018975731059261609,
+      "loss": 0.9249,
+      "step": 4170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018975245985935931,
+      "loss": 1.14,
+      "step": 4171
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018974760803979862,
+      "loss": 1.1724,
+      "step": 4172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018974275513399268,
+      "loss": 1.0717,
+      "step": 4173
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018973790114200028,
+      "loss": 1.1755,
+      "step": 4174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001897330460638801,
+      "loss": 0.9043,
+      "step": 4175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018972818989969095,
+      "loss": 0.9613,
+      "step": 4176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001897233326494916,
+      "loss": 0.9201,
+      "step": 4177
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018971847431334084,
+      "loss": 0.994,
+      "step": 4178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018971361489129748,
+      "loss": 1.0479,
+      "step": 4179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001897087543834203,
+      "loss": 0.9257,
+      "step": 4180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018970389278976816,
+      "loss": 0.9709,
+      "step": 4181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018969903011039995,
+      "loss": 0.9386,
+      "step": 4182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001896941663453744,
+      "loss": 1.1184,
+      "step": 4183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018968930149475046,
+      "loss": 0.9727,
+      "step": 4184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000189684435558587,
+      "loss": 1.079,
+      "step": 4185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018967956853694292,
+      "loss": 0.9781,
+      "step": 4186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001896747004298771,
+      "loss": 0.8908,
+      "step": 4187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018966983123744848,
+      "loss": 0.8664,
+      "step": 4188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018966496095971602,
+      "loss": 1.0718,
+      "step": 4189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018966008959673864,
+      "loss": 0.9925,
+      "step": 4190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018965521714857527,
+      "loss": 1.0588,
+      "step": 4191
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001896503436152849,
+      "loss": 0.9328,
+      "step": 4192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018964546899692656,
+      "loss": 0.9915,
+      "step": 4193
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001896405932935592,
+      "loss": 0.9344,
+      "step": 4194
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018963571650524187,
+      "loss": 0.9727,
+      "step": 4195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001896308386320335,
+      "loss": 0.8959,
+      "step": 4196
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001896259596739933,
+      "loss": 1.0883,
+      "step": 4197
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018962107963118018,
+      "loss": 1.0782,
+      "step": 4198
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018961619850365328,
+      "loss": 0.9819,
+      "step": 4199
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001896113162914716,
+      "loss": 1.0325,
+      "step": 4200
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 0.9915627241134644,
+      "eval_runtime": 846.154,
+      "eval_samples_per_second": 28.435,
+      "eval_steps_per_second": 14.217,
+      "step": 4200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018960643299469429,
+      "loss": 1.0247,
+      "step": 4201
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018960154861338046,
+      "loss": 1.02,
+      "step": 4202
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001895966631475892,
+      "loss": 1.0614,
+      "step": 4203
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018959177659737963,
+      "loss": 1.1039,
+      "step": 4204
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018958688896281097,
+      "loss": 0.9647,
+      "step": 4205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018958200024394225,
+      "loss": 1.1061,
+      "step": 4206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018957711044083277,
+      "loss": 1.0317,
+      "step": 4207
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001895722195535416,
+      "loss": 0.9838,
+      "step": 4208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018956732758212803,
+      "loss": 0.9502,
+      "step": 4209
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001895624345266512,
+      "loss": 1.0479,
+      "step": 4210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018955754038717037,
+      "loss": 0.9007,
+      "step": 4211
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018955264516374478,
+      "loss": 1.1046,
+      "step": 4212
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018954774885643368,
+      "loss": 1.0314,
+      "step": 4213
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018954285146529628,
+      "loss": 1.0543,
+      "step": 4214
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001895379529903919,
+      "loss": 0.9254,
+      "step": 4215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001895330534317798,
+      "loss": 0.9072,
+      "step": 4216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018952815278951932,
+      "loss": 0.9752,
+      "step": 4217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018952325106366978,
+      "loss": 0.9867,
+      "step": 4218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018951834825429043,
+      "loss": 1.1124,
+      "step": 4219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001895134443614407,
+      "loss": 1.0068,
+      "step": 4220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018950853938517987,
+      "loss": 0.8788,
+      "step": 4221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018950363332556738,
+      "loss": 1.0237,
+      "step": 4222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018949872618266253,
+      "loss": 1.0671,
+      "step": 4223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001894938179565248,
+      "loss": 0.9232,
+      "step": 4224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018948890864721351,
+      "loss": 0.9557,
+      "step": 4225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018948399825478813,
+      "loss": 1.0874,
+      "step": 4226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001894790867793081,
+      "loss": 0.9787,
+      "step": 4227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001894741742208328,
+      "loss": 0.9442,
+      "step": 4228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018946926057942178,
+      "loss": 1.0548,
+      "step": 4229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018946434585513444,
+      "loss": 1.0203,
+      "step": 4230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001894594300480303,
+      "loss": 0.9628,
+      "step": 4231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018945451315816888,
+      "loss": 0.9662,
+      "step": 4232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018944959518560957,
+      "loss": 1.0031,
+      "step": 4233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018944467613041206,
+      "loss": 0.9358,
+      "step": 4234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001894397559926358,
+      "loss": 1.0207,
+      "step": 4235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018943483477234035,
+      "loss": 0.9795,
+      "step": 4236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018942991246958526,
+      "loss": 0.9464,
+      "step": 4237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001894249890844301,
+      "loss": 1.0108,
+      "step": 4238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018942006461693448,
+      "loss": 0.9688,
+      "step": 4239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018941513906715803,
+      "loss": 0.9682,
+      "step": 4240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018941021243516036,
+      "loss": 1.049,
+      "step": 4241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000189405284721001,
+      "loss": 0.9187,
+      "step": 4242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018940035592473974,
+      "loss": 0.9876,
+      "step": 4243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001893954260464361,
+      "loss": 0.9735,
+      "step": 4244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018939049508614987,
+      "loss": 0.8615,
+      "step": 4245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001893855630439406,
+      "loss": 0.9716,
+      "step": 4246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018938062991986812,
+      "loss": 1.1257,
+      "step": 4247
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018937569571399204,
+      "loss": 1.0075,
+      "step": 4248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001893707604263721,
+      "loss": 0.9112,
+      "step": 4249
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001893658240570681,
+      "loss": 0.9587,
+      "step": 4250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018936088660613968,
+      "loss": 0.9934,
+      "step": 4251
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018935594807364668,
+      "loss": 1.0154,
+      "step": 4252
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018935100845964882,
+      "loss": 1.0583,
+      "step": 4253
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001893460677642059,
+      "loss": 0.9417,
+      "step": 4254
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018934112598737777,
+      "loss": 1.1004,
+      "step": 4255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018933618312922418,
+      "loss": 1.0589,
+      "step": 4256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000189331239189805,
+      "loss": 1.0491,
+      "step": 4257
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018932629416918,
+      "loss": 1.0369,
+      "step": 4258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018932134806740912,
+      "loss": 1.0134,
+      "step": 4259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018931640088455215,
+      "loss": 1.0248,
+      "step": 4260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018931145262066903,
+      "loss": 1.0832,
+      "step": 4261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018930650327581957,
+      "loss": 1.002,
+      "step": 4262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018930155285006378,
+      "loss": 1.0118,
+      "step": 4263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018929660134346147,
+      "loss": 0.9638,
+      "step": 4264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018929164875607264,
+      "loss": 1.0043,
+      "step": 4265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018928669508795723,
+      "loss": 0.9984,
+      "step": 4266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018928174033917513,
+      "loss": 1.074,
+      "step": 4267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001892767845097864,
+      "loss": 1.0062,
+      "step": 4268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018927182759985093,
+      "loss": 0.9754,
+      "step": 4269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001892668696094288,
+      "loss": 1.0483,
+      "step": 4270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018926191053857995,
+      "loss": 0.9599,
+      "step": 4271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018925695038736446,
+      "loss": 1.0774,
+      "step": 4272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001892519891558423,
+      "loss": 0.93,
+      "step": 4273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018924702684407362,
+      "loss": 1.0279,
+      "step": 4274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018924206345211837,
+      "loss": 1.0695,
+      "step": 4275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018923709898003668,
+      "loss": 1.0434,
+      "step": 4276
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018923213342788864,
+      "loss": 0.9481,
+      "step": 4277
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001892271667957343,
+      "loss": 1.0183,
+      "step": 4278
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018922219908363381,
+      "loss": 0.94,
+      "step": 4279
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018921723029164734,
+      "loss": 1.1096,
+      "step": 4280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018921226041983494,
+      "loss": 1.1363,
+      "step": 4281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018920728946825678,
+      "loss": 1.0925,
+      "step": 4282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018920231743697313,
+      "loss": 0.7946,
+      "step": 4283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018919734432604402,
+      "loss": 1.0101,
+      "step": 4284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018919237013552973,
+      "loss": 1.0092,
+      "step": 4285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018918739486549043,
+      "loss": 0.9208,
+      "step": 4286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018918241851598636,
+      "loss": 0.9853,
+      "step": 4287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018917744108707776,
+      "loss": 0.9521,
+      "step": 4288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018917246257882484,
+      "loss": 1.0395,
+      "step": 4289
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018916748299128785,
+      "loss": 0.8611,
+      "step": 4290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001891625023245271,
+      "loss": 1.0068,
+      "step": 4291
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018915752057860284,
+      "loss": 1.0297,
+      "step": 4292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001891525377535754,
+      "loss": 1.0133,
+      "step": 4293
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018914755384950507,
+      "loss": 1.0189,
+      "step": 4294
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018914256886645215,
+      "loss": 0.9729,
+      "step": 4295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018913758280447702,
+      "loss": 1.064,
+      "step": 4296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018913259566363997,
+      "loss": 0.8679,
+      "step": 4297
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001891276074440014,
+      "loss": 1.0356,
+      "step": 4298
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001891226181456217,
+      "loss": 0.97,
+      "step": 4299
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001891176277685612,
+      "loss": 0.9281,
+      "step": 4300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018911263631288039,
+      "loss": 1.0598,
+      "step": 4301
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018910764377863957,
+      "loss": 0.9908,
+      "step": 4302
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018910265016589925,
+      "loss": 0.9407,
+      "step": 4303
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018909765547471985,
+      "loss": 0.885,
+      "step": 4304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018909265970516178,
+      "loss": 1.0826,
+      "step": 4305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001890876628572856,
+      "loss": 1.0124,
+      "step": 4306
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018908266493115167,
+      "loss": 0.9569,
+      "step": 4307
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018907766592682058,
+      "loss": 0.9234,
+      "step": 4308
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001890726658443528,
+      "loss": 0.9667,
+      "step": 4309
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018906766468380885,
+      "loss": 1.1332,
+      "step": 4310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018906266244524922,
+      "loss": 0.9849,
+      "step": 4311
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001890576591287345,
+      "loss": 1.0374,
+      "step": 4312
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018905265473432526,
+      "loss": 0.9773,
+      "step": 4313
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018904764926208203,
+      "loss": 1.0219,
+      "step": 4314
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018904264271206537,
+      "loss": 1.0077,
+      "step": 4315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018903763508433595,
+      "loss": 1.1194,
+      "step": 4316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018903262637895435,
+      "loss": 0.8842,
+      "step": 4317
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018902761659598117,
+      "loss": 1.0471,
+      "step": 4318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018902260573547708,
+      "loss": 1.0691,
+      "step": 4319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001890175937975027,
+      "loss": 0.9522,
+      "step": 4320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018901258078211867,
+      "loss": 0.9619,
+      "step": 4321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018900756668938573,
+      "loss": 0.9244,
+      "step": 4322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001890025515193645,
+      "loss": 0.9354,
+      "step": 4323
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018899753527211574,
+      "loss": 0.9762,
+      "step": 4324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018899251794770013,
+      "loss": 1.0492,
+      "step": 4325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001889874995461784,
+      "loss": 1.0638,
+      "step": 4326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001889824800676113,
+      "loss": 0.9881,
+      "step": 4327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001889774595120595,
+      "loss": 0.9809,
+      "step": 4328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018897243787958392,
+      "loss": 1.0742,
+      "step": 4329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018896741517024522,
+      "loss": 0.891,
+      "step": 4330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018896239138410425,
+      "loss": 0.9591,
+      "step": 4331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018895736652122178,
+      "loss": 0.9385,
+      "step": 4332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018895234058165865,
+      "loss": 0.9042,
+      "step": 4333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018894731356547566,
+      "loss": 0.9631,
+      "step": 4334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001889422854727337,
+      "loss": 1.037,
+      "step": 4335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001889372563034936,
+      "loss": 0.9093,
+      "step": 4336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001889322260578162,
+      "loss": 0.9671,
+      "step": 4337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018892719473576243,
+      "loss": 0.9287,
+      "step": 4338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018892216233739318,
+      "loss": 0.948,
+      "step": 4339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018891712886276934,
+      "loss": 1.0003,
+      "step": 4340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001889120943119518,
+      "loss": 1.0264,
+      "step": 4341
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018890705868500161,
+      "loss": 1.0116,
+      "step": 4342
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018890202198197962,
+      "loss": 0.9965,
+      "step": 4343
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001888969842029468,
+      "loss": 0.8983,
+      "step": 4344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018889194534796414,
+      "loss": 0.9729,
+      "step": 4345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018888690541709263,
+      "loss": 1.0018,
+      "step": 4346
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018888186441039324,
+      "loss": 0.9803,
+      "step": 4347
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018887682232792704,
+      "loss": 1.0428,
+      "step": 4348
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000188871779169755,
+      "loss": 0.9743,
+      "step": 4349
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001888667349359382,
+      "loss": 0.8995,
+      "step": 4350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018886168962653764,
+      "loss": 0.9032,
+      "step": 4351
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018885664324161446,
+      "loss": 1.0686,
+      "step": 4352
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018885159578122966,
+      "loss": 1.1188,
+      "step": 4353
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001888465472454444,
+      "loss": 0.9579,
+      "step": 4354
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018884149763431974,
+      "loss": 0.9771,
+      "step": 4355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018883644694791683,
+      "loss": 0.9186,
+      "step": 4356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018883139518629675,
+      "loss": 0.9369,
+      "step": 4357
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018882634234952068,
+      "loss": 0.983,
+      "step": 4358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018882128843764976,
+      "loss": 0.9679,
+      "step": 4359
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001888162334507452,
+      "loss": 0.9676,
+      "step": 4360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018881117738886815,
+      "loss": 0.9582,
+      "step": 4361
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001888061202520798,
+      "loss": 1.1154,
+      "step": 4362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018880106204044136,
+      "loss": 0.9394,
+      "step": 4363
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018879600275401403,
+      "loss": 0.8632,
+      "step": 4364
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001887909423928591,
+      "loss": 0.9431,
+      "step": 4365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018878588095703778,
+      "loss": 1.0023,
+      "step": 4366
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018878081844661134,
+      "loss": 1.0292,
+      "step": 4367
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018877575486164102,
+      "loss": 0.9394,
+      "step": 4368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018877069020218818,
+      "loss": 0.9888,
+      "step": 4369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018876562446831405,
+      "loss": 1.0589,
+      "step": 4370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018876055766007998,
+      "loss": 0.9334,
+      "step": 4371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001887554897775473,
+      "loss": 0.9819,
+      "step": 4372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018875042082077727,
+      "loss": 0.9021,
+      "step": 4373
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018874535078983136,
+      "loss": 1.0997,
+      "step": 4374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018874027968477085,
+      "loss": 0.9839,
+      "step": 4375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018873520750565718,
+      "loss": 0.8616,
+      "step": 4376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001887301342525517,
+      "loss": 1.0367,
+      "step": 4377
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018872505992551577,
+      "loss": 1.0144,
+      "step": 4378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018871998452461088,
+      "loss": 1.0249,
+      "step": 4379
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018871490804989844,
+      "loss": 1.0385,
+      "step": 4380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018870983050143988,
+      "loss": 0.9967,
+      "step": 4381
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018870475187929666,
+      "loss": 0.9786,
+      "step": 4382
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018869967218353026,
+      "loss": 1.0004,
+      "step": 4383
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018869459141420214,
+      "loss": 0.9239,
+      "step": 4384
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001886895095713738,
+      "loss": 0.8897,
+      "step": 4385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018868442665510678,
+      "loss": 0.9649,
+      "step": 4386
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018867934266546255,
+      "loss": 0.9858,
+      "step": 4387
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018867425760250267,
+      "loss": 1.0414,
+      "step": 4388
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018866917146628866,
+      "loss": 0.9509,
+      "step": 4389
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018866408425688213,
+      "loss": 1.0658,
+      "step": 4390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018865899597434463,
+      "loss": 1.0116,
+      "step": 4391
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001886539066187377,
+      "loss": 0.9144,
+      "step": 4392
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018864881619012303,
+      "loss": 0.9852,
+      "step": 4393
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018864372468856212,
+      "loss": 1.1328,
+      "step": 4394
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001886386321141167,
+      "loss": 0.9944,
+      "step": 4395
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018863353846684832,
+      "loss": 0.9748,
+      "step": 4396
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001886284437468187,
+      "loss": 0.9094,
+      "step": 4397
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018862334795408944,
+      "loss": 0.9204,
+      "step": 4398
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018861825108872229,
+      "loss": 1.0731,
+      "step": 4399
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001886131531507789,
+      "loss": 0.9936,
+      "step": 4400
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 0.9748373627662659,
+      "eval_runtime": 847.0797,
+      "eval_samples_per_second": 28.403,
+      "eval_steps_per_second": 14.202,
+      "step": 4400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001886080541403209,
+      "loss": 0.9745,
+      "step": 4401
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018860295405741012,
+      "loss": 1.0297,
+      "step": 4402
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018859785290210824,
+      "loss": 0.92,
+      "step": 4403
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018859275067447702,
+      "loss": 0.8888,
+      "step": 4404
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001885876473745782,
+      "loss": 1.0486,
+      "step": 4405
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018858254300247352,
+      "loss": 0.8369,
+      "step": 4406
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018857743755822478,
+      "loss": 0.9965,
+      "step": 4407
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001885723310418938,
+      "loss": 1.013,
+      "step": 4408
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018856722345354232,
+      "loss": 0.9782,
+      "step": 4409
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018856211479323224,
+      "loss": 0.9791,
+      "step": 4410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018855700506102533,
+      "loss": 0.9406,
+      "step": 4411
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001885518942569835,
+      "loss": 1.0845,
+      "step": 4412
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018854678238116855,
+      "loss": 0.953,
+      "step": 4413
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018854166943364233,
+      "loss": 0.8909,
+      "step": 4414
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018853655541446683,
+      "loss": 0.8835,
+      "step": 4415
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018853144032370382,
+      "loss": 0.9691,
+      "step": 4416
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018852632416141527,
+      "loss": 0.931,
+      "step": 4417
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018852120692766313,
+      "loss": 1.0215,
+      "step": 4418
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018851608862250927,
+      "loss": 1.0481,
+      "step": 4419
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001885109692460157,
+      "loss": 0.838,
+      "step": 4420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018850584879824434,
+      "loss": 1.0713,
+      "step": 4421
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001885007272792572,
+      "loss": 1.0277,
+      "step": 4422
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001884956046891162,
+      "loss": 1.0262,
+      "step": 4423
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018849048102788344,
+      "loss": 0.9291,
+      "step": 4424
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018848535629562086,
+      "loss": 1.0222,
+      "step": 4425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018848023049239048,
+      "loss": 0.8341,
+      "step": 4426
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018847510361825443,
+      "loss": 1.1322,
+      "step": 4427
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018846997567327468,
+      "loss": 0.9901,
+      "step": 4428
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018846484665751328,
+      "loss": 0.9427,
+      "step": 4429
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018845971657103236,
+      "loss": 0.8918,
+      "step": 4430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000188454585413894,
+      "loss": 1.0635,
+      "step": 4431
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001884494531861603,
+      "loss": 0.9769,
+      "step": 4432
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018844431988789334,
+      "loss": 0.9044,
+      "step": 4433
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001884391855191553,
+      "loss": 0.8859,
+      "step": 4434
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018843405008000833,
+      "loss": 0.8311,
+      "step": 4435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018842891357051457,
+      "loss": 0.8582,
+      "step": 4436
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018842377599073614,
+      "loss": 0.8703,
+      "step": 4437
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001884186373407353,
+      "loss": 0.937,
+      "step": 4438
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001884134976205742,
+      "loss": 1.0276,
+      "step": 4439
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018840835683031507,
+      "loss": 0.9289,
+      "step": 4440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001884032149700201,
+      "loss": 0.9789,
+      "step": 4441
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018839807203975152,
+      "loss": 0.8913,
+      "step": 4442
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018839292803957163,
+      "loss": 1.0152,
+      "step": 4443
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018838778296954265,
+      "loss": 0.9982,
+      "step": 4444
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018838263682972688,
+      "loss": 1.0912,
+      "step": 4445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018837748962018656,
+      "loss": 0.9535,
+      "step": 4446
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000188372341340984,
+      "loss": 0.9088,
+      "step": 4447
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018836719199218153,
+      "loss": 0.882,
+      "step": 4448
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001883620415738415,
+      "loss": 1.0183,
+      "step": 4449
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001883568900860262,
+      "loss": 1.0692,
+      "step": 4450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018835173752879798,
+      "loss": 0.8981,
+      "step": 4451
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018834658390221923,
+      "loss": 0.9078,
+      "step": 4452
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018834142920635234,
+      "loss": 0.9055,
+      "step": 4453
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018833627344125965,
+      "loss": 0.9906,
+      "step": 4454
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001883311166070036,
+      "loss": 0.9512,
+      "step": 4455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018832595870364656,
+      "loss": 0.9949,
+      "step": 4456
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018832079973125104,
+      "loss": 1.0008,
+      "step": 4457
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001883156396898794,
+      "loss": 0.9291,
+      "step": 4458
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018831047857959413,
+      "loss": 1.0733,
+      "step": 4459
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001883053164004577,
+      "loss": 0.9953,
+      "step": 4460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018830015315253256,
+      "loss": 1.0203,
+      "step": 4461
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018829498883588124,
+      "loss": 0.9398,
+      "step": 4462
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001882898234505662,
+      "loss": 0.8836,
+      "step": 4463
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018828465699665,
+      "loss": 0.9356,
+      "step": 4464
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001882794894741952,
+      "loss": 0.8849,
+      "step": 4465
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018827432088326423,
+      "loss": 0.965,
+      "step": 4466
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018826915122391978,
+      "loss": 1.0099,
+      "step": 4467
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018826398049622432,
+      "loss": 1.0109,
+      "step": 4468
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001882588087002405,
+      "loss": 1.0605,
+      "step": 4469
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018825363583603087,
+      "loss": 0.8364,
+      "step": 4470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018824846190365805,
+      "loss": 1.1074,
+      "step": 4471
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001882432869031847,
+      "loss": 0.9953,
+      "step": 4472
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018823811083467338,
+      "loss": 0.9254,
+      "step": 4473
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001882329336981868,
+      "loss": 0.9534,
+      "step": 4474
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018822775549378763,
+      "loss": 1.0404,
+      "step": 4475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001882225762215385,
+      "loss": 0.9594,
+      "step": 4476
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001882173958815021,
+      "loss": 0.9411,
+      "step": 4477
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018821221447374114,
+      "loss": 1.0929,
+      "step": 4478
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018820703199831835,
+      "loss": 0.8383,
+      "step": 4479
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018820184845529645,
+      "loss": 0.9206,
+      "step": 4480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018819666384473812,
+      "loss": 0.8852,
+      "step": 4481
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001881914781667062,
+      "loss": 0.9805,
+      "step": 4482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001881862914212634,
+      "loss": 0.8562,
+      "step": 4483
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001881811036084725,
+      "loss": 1.1335,
+      "step": 4484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001881759147283963,
+      "loss": 0.9804,
+      "step": 4485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018817072478109764,
+      "loss": 0.9785,
+      "step": 4486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018816553376663924,
+      "loss": 1.0985,
+      "step": 4487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018816034168508402,
+      "loss": 0.8665,
+      "step": 4488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018815514853649478,
+      "loss": 0.9524,
+      "step": 4489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001881499543209344,
+      "loss": 0.9504,
+      "step": 4490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018814475903846571,
+      "loss": 0.8933,
+      "step": 4491
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001881395626891516,
+      "loss": 0.9703,
+      "step": 4492
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018813436527305498,
+      "loss": 1.0086,
+      "step": 4493
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018812916679023878,
+      "loss": 1.0828,
+      "step": 4494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018812396724076584,
+      "loss": 1.0564,
+      "step": 4495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018811876662469916,
+      "loss": 0.9712,
+      "step": 4496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018811356494210165,
+      "loss": 0.9603,
+      "step": 4497
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018810836219303631,
+      "loss": 0.9974,
+      "step": 4498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001881031583775661,
+      "loss": 1.0331,
+      "step": 4499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018809795349575392,
+      "loss": 0.8699,
+      "step": 4500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018809274754766288,
+      "loss": 0.8747,
+      "step": 4501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018808754053335593,
+      "loss": 0.9962,
+      "step": 4502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001880823324528961,
+      "loss": 1.1062,
+      "step": 4503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018807712330634642,
+      "loss": 0.9702,
+      "step": 4504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018807191309377,
+      "loss": 1.0357,
+      "step": 4505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001880667018152298,
+      "loss": 0.8985,
+      "step": 4506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018806148947078893,
+      "loss": 1.1088,
+      "step": 4507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001880562760605105,
+      "loss": 0.9203,
+      "step": 4508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018805106158445762,
+      "loss": 0.9571,
+      "step": 4509
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018804584604269336,
+      "loss": 0.9367,
+      "step": 4510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018804062943528088,
+      "loss": 1.0321,
+      "step": 4511
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018803541176228328,
+      "loss": 0.8771,
+      "step": 4512
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001880301930237638,
+      "loss": 1.1079,
+      "step": 4513
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018802497321978548,
+      "loss": 0.9207,
+      "step": 4514
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001880197523504116,
+      "loss": 0.9717,
+      "step": 4515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018801453041570527,
+      "loss": 1.0445,
+      "step": 4516
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018800930741572972,
+      "loss": 0.9096,
+      "step": 4517
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018800408335054822,
+      "loss": 0.9441,
+      "step": 4518
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018799885822022393,
+      "loss": 0.9961,
+      "step": 4519
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001879936320248201,
+      "loss": 0.9115,
+      "step": 4520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018798840476440002,
+      "loss": 0.927,
+      "step": 4521
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018798317643902696,
+      "loss": 0.9839,
+      "step": 4522
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018797794704876415,
+      "loss": 1.0527,
+      "step": 4523
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018797271659367487,
+      "loss": 0.9524,
+      "step": 4524
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018796748507382254,
+      "loss": 0.8994,
+      "step": 4525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018796225248927036,
+      "loss": 1.0268,
+      "step": 4526
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018795701884008172,
+      "loss": 1.092,
+      "step": 4527
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018795178412631995,
+      "loss": 0.9522,
+      "step": 4528
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018794654834804842,
+      "loss": 1.0246,
+      "step": 4529
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018794131150533046,
+      "loss": 0.9826,
+      "step": 4530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018793607359822955,
+      "loss": 0.9515,
+      "step": 4531
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018793083462680896,
+      "loss": 0.9531,
+      "step": 4532
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018792559459113216,
+      "loss": 0.9456,
+      "step": 4533
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018792035349126258,
+      "loss": 0.9017,
+      "step": 4534
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018791511132726366,
+      "loss": 0.9937,
+      "step": 4535
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018790986809919883,
+      "loss": 0.9006,
+      "step": 4536
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018790462380713152,
+      "loss": 1.0272,
+      "step": 4537
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018789937845112525,
+      "loss": 0.9667,
+      "step": 4538
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001878941320312435,
+      "loss": 0.9789,
+      "step": 4539
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018788888454754975,
+      "loss": 0.963,
+      "step": 4540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018788363600010758,
+      "loss": 0.9022,
+      "step": 4541
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001878783863889804,
+      "loss": 1.0151,
+      "step": 4542
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001878731357142318,
+      "loss": 0.9094,
+      "step": 4543
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018786788397592537,
+      "loss": 0.9095,
+      "step": 4544
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018786263117412464,
+      "loss": 0.9365,
+      "step": 4545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018785737730889315,
+      "loss": 1.0336,
+      "step": 4546
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018785212238029458,
+      "loss": 1.0233,
+      "step": 4547
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018784686638839243,
+      "loss": 0.8775,
+      "step": 4548
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001878416093332504,
+      "loss": 0.968,
+      "step": 4549
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018783635121493206,
+      "loss": 0.8774,
+      "step": 4550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018783109203350108,
+      "loss": 0.8719,
+      "step": 4551
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018782583178902112,
+      "loss": 0.9864,
+      "step": 4552
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001878205704815558,
+      "loss": 1.0316,
+      "step": 4553
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018781530811116885,
+      "loss": 1.0154,
+      "step": 4554
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018781004467792394,
+      "loss": 0.9704,
+      "step": 4555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018780478018188477,
+      "loss": 0.9533,
+      "step": 4556
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001877995146231151,
+      "loss": 0.8941,
+      "step": 4557
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001877942480016786,
+      "loss": 0.9719,
+      "step": 4558
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018778898031763906,
+      "loss": 1.0751,
+      "step": 4559
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001877837115710602,
+      "loss": 0.9326,
+      "step": 4560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018777844176200582,
+      "loss": 0.8688,
+      "step": 4561
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018777317089053964,
+      "loss": 0.9366,
+      "step": 4562
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018776789895672558,
+      "loss": 1.0209,
+      "step": 4563
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018776262596062734,
+      "loss": 0.8542,
+      "step": 4564
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018775735190230876,
+      "loss": 0.9055,
+      "step": 4565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001877520767818337,
+      "loss": 0.9388,
+      "step": 4566
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000187746800599266,
+      "loss": 0.9359,
+      "step": 4567
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001877415233546695,
+      "loss": 1.0285,
+      "step": 4568
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001877362450481081,
+      "loss": 0.9129,
+      "step": 4569
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018773096567964567,
+      "loss": 0.9485,
+      "step": 4570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018772568524934609,
+      "loss": 0.9575,
+      "step": 4571
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001877204037572733,
+      "loss": 1.0443,
+      "step": 4572
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018771512120349123,
+      "loss": 1.0791,
+      "step": 4573
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018770983758806378,
+      "loss": 0.8434,
+      "step": 4574
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001877045529110549,
+      "loss": 0.8524,
+      "step": 4575
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018769926717252861,
+      "loss": 0.9237,
+      "step": 4576
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018769398037254883,
+      "loss": 0.9717,
+      "step": 4577
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018768869251117954,
+      "loss": 0.9974,
+      "step": 4578
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018768340358848483,
+      "loss": 0.8414,
+      "step": 4579
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018767811360452858,
+      "loss": 0.9428,
+      "step": 4580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018767282255937492,
+      "loss": 0.8683,
+      "step": 4581
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018766753045308784,
+      "loss": 1.022,
+      "step": 4582
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018766223728573144,
+      "loss": 0.9996,
+      "step": 4583
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001876569430573697,
+      "loss": 0.9497,
+      "step": 4584
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018765164776806678,
+      "loss": 1.034,
+      "step": 4585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018764635141788673,
+      "loss": 1.0402,
+      "step": 4586
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018764105400689366,
+      "loss": 0.9136,
+      "step": 4587
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018763575553515168,
+      "loss": 0.9838,
+      "step": 4588
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018763045600272496,
+      "loss": 0.9481,
+      "step": 4589
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001876251554096776,
+      "loss": 1.0254,
+      "step": 4590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018761985375607374,
+      "loss": 0.8504,
+      "step": 4591
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001876145510419776,
+      "loss": 0.918,
+      "step": 4592
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018760924726745332,
+      "loss": 0.9689,
+      "step": 4593
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018760394243256512,
+      "loss": 0.9218,
+      "step": 4594
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001875986365373772,
+      "loss": 0.9156,
+      "step": 4595
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018759332958195375,
+      "loss": 0.8724,
+      "step": 4596
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018758802156635903,
+      "loss": 0.9466,
+      "step": 4597
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001875827124906573,
+      "loss": 1.0284,
+      "step": 4598
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018757740235491276,
+      "loss": 0.8998,
+      "step": 4599
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018757209115918973,
+      "loss": 0.9131,
+      "step": 4600
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 0.9592324495315552,
+      "eval_runtime": 849.2893,
+      "eval_samples_per_second": 28.33,
+      "eval_steps_per_second": 14.165,
+      "step": 4600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001875667789035525,
+      "loss": 0.9899,
+      "step": 4601
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018756146558806532,
+      "loss": 1.0264,
+      "step": 4602
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018755615121279252,
+      "loss": 0.9189,
+      "step": 4603
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018755083577779846,
+      "loss": 0.9238,
+      "step": 4604
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018754551928314741,
+      "loss": 1.0522,
+      "step": 4605
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018754020172890377,
+      "loss": 1.0212,
+      "step": 4606
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018753488311513185,
+      "loss": 0.9743,
+      "step": 4607
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018752956344189609,
+      "loss": 0.9411,
+      "step": 4608
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001875242427092608,
+      "loss": 0.8786,
+      "step": 4609
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001875189209172904,
+      "loss": 0.9333,
+      "step": 4610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018751359806604937,
+      "loss": 1.0269,
+      "step": 4611
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018750827415560205,
+      "loss": 0.9445,
+      "step": 4612
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018750294918601292,
+      "loss": 1.0317,
+      "step": 4613
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018749762315734643,
+      "loss": 0.8315,
+      "step": 4614
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000187492296069667,
+      "loss": 0.9585,
+      "step": 4615
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018748696792303916,
+      "loss": 1.0658,
+      "step": 4616
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018748163871752738,
+      "loss": 1.0233,
+      "step": 4617
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018747630845319612,
+      "loss": 0.9415,
+      "step": 4618
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018747097713010997,
+      "loss": 1.0696,
+      "step": 4619
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018746564474833339,
+      "loss": 0.918,
+      "step": 4620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018746031130793096,
+      "loss": 0.9617,
+      "step": 4621
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018745497680896722,
+      "loss": 0.9976,
+      "step": 4622
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018744964125150672,
+      "loss": 0.9222,
+      "step": 4623
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018744430463561406,
+      "loss": 0.8952,
+      "step": 4624
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018743896696135385,
+      "loss": 0.9629,
+      "step": 4625
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018743362822879064,
+      "loss": 0.8452,
+      "step": 4626
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018742828843798905,
+      "loss": 1.0081,
+      "step": 4627
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001874229475890138,
+      "loss": 1.053,
+      "step": 4628
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018741760568192936,
+      "loss": 0.9919,
+      "step": 4629
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018741226271680057,
+      "loss": 0.9957,
+      "step": 4630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000187406918693692,
+      "loss": 0.9833,
+      "step": 4631
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001874015736126683,
+      "loss": 0.9682,
+      "step": 4632
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001873962274737943,
+      "loss": 0.9726,
+      "step": 4633
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018739088027713456,
+      "loss": 0.827,
+      "step": 4634
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018738553202275385,
+      "loss": 1.0662,
+      "step": 4635
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018738018271071693,
+      "loss": 1.0308,
+      "step": 4636
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018737483234108852,
+      "loss": 1.0279,
+      "step": 4637
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018736948091393335,
+      "loss": 0.7995,
+      "step": 4638
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018736412842931626,
+      "loss": 0.9496,
+      "step": 4639
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000187358774887302,
+      "loss": 0.9813,
+      "step": 4640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001873534202879553,
+      "loss": 1.0276,
+      "step": 4641
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018734806463134107,
+      "loss": 0.9522,
+      "step": 4642
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018734270791752406,
+      "loss": 0.9464,
+      "step": 4643
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018733735014656914,
+      "loss": 0.8327,
+      "step": 4644
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018733199131854114,
+      "loss": 1.0919,
+      "step": 4645
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018732663143350499,
+      "loss": 0.8807,
+      "step": 4646
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018732127049152542,
+      "loss": 0.8097,
+      "step": 4647
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018731590849266745,
+      "loss": 1.0206,
+      "step": 4648
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018731054543699588,
+      "loss": 0.9439,
+      "step": 4649
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018730518132457568,
+      "loss": 0.9626,
+      "step": 4650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001872998161554718,
+      "loss": 0.9968,
+      "step": 4651
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001872944499297491,
+      "loss": 0.9116,
+      "step": 4652
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018728908264747257,
+      "loss": 0.8924,
+      "step": 4653
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018728371430870716,
+      "loss": 0.9613,
+      "step": 4654
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018727834491351786,
+      "loss": 1.0663,
+      "step": 4655
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018727297446196963,
+      "loss": 1.0328,
+      "step": 4656
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001872676029541275,
+      "loss": 0.8906,
+      "step": 4657
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018726223039005645,
+      "loss": 1.0003,
+      "step": 4658
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018725685676982158,
+      "loss": 1.112,
+      "step": 4659
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018725148209348784,
+      "loss": 1.0247,
+      "step": 4660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001872461063611203,
+      "loss": 1.0122,
+      "step": 4661
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018724072957278408,
+      "loss": 0.9988,
+      "step": 4662
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018723535172854422,
+      "loss": 1.0297,
+      "step": 4663
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001872299728284658,
+      "loss": 1.0075,
+      "step": 4664
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018722459287261393,
+      "loss": 0.9347,
+      "step": 4665
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001872192118610537,
+      "loss": 0.8986,
+      "step": 4666
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018721382979385032,
+      "loss": 1.0307,
+      "step": 4667
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018720844667106885,
+      "loss": 0.8891,
+      "step": 4668
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018720306249277448,
+      "loss": 1.0566,
+      "step": 4669
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018719767725903233,
+      "loss": 1.1641,
+      "step": 4670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018719229096990768,
+      "loss": 0.9725,
+      "step": 4671
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001871869036254656,
+      "loss": 1.084,
+      "step": 4672
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018718151522577139,
+      "loss": 0.9565,
+      "step": 4673
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001871761257708902,
+      "loss": 1.0134,
+      "step": 4674
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018717073526088732,
+      "loss": 0.8585,
+      "step": 4675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018716534369582792,
+      "loss": 0.9449,
+      "step": 4676
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018715995107577733,
+      "loss": 1.0932,
+      "step": 4677
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018715455740080079,
+      "loss": 1.0729,
+      "step": 4678
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001871491626709636,
+      "loss": 0.9365,
+      "step": 4679
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000187143766886331,
+      "loss": 0.9431,
+      "step": 4680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018713837004696835,
+      "loss": 0.8612,
+      "step": 4681
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018713297215294095,
+      "loss": 0.9349,
+      "step": 4682
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001871275732043141,
+      "loss": 1.0564,
+      "step": 4683
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018712217320115322,
+      "loss": 0.8877,
+      "step": 4684
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001871167721435236,
+      "loss": 0.9459,
+      "step": 4685
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018711137003149067,
+      "loss": 1.0086,
+      "step": 4686
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018710596686511975,
+      "loss": 0.8848,
+      "step": 4687
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018710056264447628,
+      "loss": 0.9587,
+      "step": 4688
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018709515736962566,
+      "loss": 0.9416,
+      "step": 4689
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018708975104063331,
+      "loss": 0.945,
+      "step": 4690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018708434365756466,
+      "loss": 0.9529,
+      "step": 4691
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018707893522048515,
+      "loss": 0.9557,
+      "step": 4692
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018707352572946027,
+      "loss": 1.0141,
+      "step": 4693
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018706811518455548,
+      "loss": 1.0187,
+      "step": 4694
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018706270358583623,
+      "loss": 0.853,
+      "step": 4695
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018705729093336805,
+      "loss": 0.9307,
+      "step": 4696
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018705187722721647,
+      "loss": 1.0048,
+      "step": 4697
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018704646246744697,
+      "loss": 0.8684,
+      "step": 4698
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018704104665412513,
+      "loss": 1.1039,
+      "step": 4699
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018703562978731645,
+      "loss": 0.8841,
+      "step": 4700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018703021186708654,
+      "loss": 0.9199,
+      "step": 4701
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018702479289350095,
+      "loss": 0.9595,
+      "step": 4702
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018701937286662526,
+      "loss": 0.9344,
+      "step": 4703
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018701395178652512,
+      "loss": 0.8918,
+      "step": 4704
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001870085296532661,
+      "loss": 0.8322,
+      "step": 4705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001870031064669138,
+      "loss": 0.9751,
+      "step": 4706
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018699768222753393,
+      "loss": 0.9617,
+      "step": 4707
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018699225693519206,
+      "loss": 1.0233,
+      "step": 4708
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018698683058995395,
+      "loss": 0.9139,
+      "step": 4709
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001869814031918852,
+      "loss": 0.9689,
+      "step": 4710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001869759747410515,
+      "loss": 0.8646,
+      "step": 4711
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001869705452375186,
+      "loss": 1.077,
+      "step": 4712
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018696511468135215,
+      "loss": 0.9889,
+      "step": 4713
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018695968307261797,
+      "loss": 0.9614,
+      "step": 4714
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018695425041138176,
+      "loss": 0.9392,
+      "step": 4715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018694881669770923,
+      "loss": 1.0558,
+      "step": 4716
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018694338193166616,
+      "loss": 0.9009,
+      "step": 4717
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018693794611331836,
+      "loss": 0.8497,
+      "step": 4718
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018693250924273165,
+      "loss": 0.9878,
+      "step": 4719
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018692707131997175,
+      "loss": 0.8438,
+      "step": 4720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018692163234510452,
+      "loss": 0.9656,
+      "step": 4721
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018691619231819582,
+      "loss": 0.903,
+      "step": 4722
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018691075123931144,
+      "loss": 1.0485,
+      "step": 4723
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018690530910851725,
+      "loss": 1.006,
+      "step": 4724
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018689986592587912,
+      "loss": 0.9543,
+      "step": 4725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018689442169146297,
+      "loss": 1.0292,
+      "step": 4726
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001868889764053346,
+      "loss": 0.9369,
+      "step": 4727
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018688353006756004,
+      "loss": 0.9354,
+      "step": 4728
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018687808267820515,
+      "loss": 0.9002,
+      "step": 4729
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018687263423733584,
+      "loss": 1.0221,
+      "step": 4730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018686718474501804,
+      "loss": 0.9158,
+      "step": 4731
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018686173420131774,
+      "loss": 0.947,
+      "step": 4732
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018685628260630094,
+      "loss": 0.927,
+      "step": 4733
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018685082996003355,
+      "loss": 1.0254,
+      "step": 4734
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018684537626258163,
+      "loss": 0.9625,
+      "step": 4735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018683992151401115,
+      "loss": 0.9368,
+      "step": 4736
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018683446571438815,
+      "loss": 0.8902,
+      "step": 4737
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018682900886377864,
+      "loss": 0.963,
+      "step": 4738
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018682355096224872,
+      "loss": 1.0627,
+      "step": 4739
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001868180920098644,
+      "loss": 0.9409,
+      "step": 4740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018681263200669176,
+      "loss": 1.0003,
+      "step": 4741
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018680717095279688,
+      "loss": 0.9284,
+      "step": 4742
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018680170884824586,
+      "loss": 0.9829,
+      "step": 4743
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001867962456931048,
+      "loss": 0.9655,
+      "step": 4744
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018679078148743986,
+      "loss": 0.9267,
+      "step": 4745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018678531623131717,
+      "loss": 0.9596,
+      "step": 4746
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018677984992480283,
+      "loss": 0.9461,
+      "step": 4747
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018677438256796302,
+      "loss": 1.0686,
+      "step": 4748
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018676891416086395,
+      "loss": 0.9055,
+      "step": 4749
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018676344470357176,
+      "loss": 1.139,
+      "step": 4750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018675797419615266,
+      "loss": 0.9731,
+      "step": 4751
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001867525026386729,
+      "loss": 0.8782,
+      "step": 4752
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018674703003119866,
+      "loss": 1.051,
+      "step": 4753
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001867415563737962,
+      "loss": 0.9259,
+      "step": 4754
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018673608166653173,
+      "loss": 1.0749,
+      "step": 4755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018673060590947154,
+      "loss": 1.0045,
+      "step": 4756
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018672512910268194,
+      "loss": 0.9505,
+      "step": 4757
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018671965124622916,
+      "loss": 0.9295,
+      "step": 4758
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018671417234017955,
+      "loss": 1.0023,
+      "step": 4759
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018670869238459936,
+      "loss": 0.8447,
+      "step": 4760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018670321137955498,
+      "loss": 0.9335,
+      "step": 4761
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018669772932511272,
+      "loss": 0.9676,
+      "step": 4762
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001866922462213389,
+      "loss": 0.9127,
+      "step": 4763
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018668676206829997,
+      "loss": 1.0474,
+      "step": 4764
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018668127686606223,
+      "loss": 0.8608,
+      "step": 4765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018667579061469207,
+      "loss": 0.9203,
+      "step": 4766
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018667030331425593,
+      "loss": 0.9972,
+      "step": 4767
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001866648149648202,
+      "loss": 0.8644,
+      "step": 4768
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018665932556645134,
+      "loss": 1.0234,
+      "step": 4769
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018665383511921575,
+      "loss": 0.9633,
+      "step": 4770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018664834362317988,
+      "loss": 0.852,
+      "step": 4771
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018664285107841027,
+      "loss": 1.032,
+      "step": 4772
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018663735748497331,
+      "loss": 0.8523,
+      "step": 4773
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001866318628429355,
+      "loss": 0.9754,
+      "step": 4774
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001866263671523634,
+      "loss": 1.0178,
+      "step": 4775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018662087041332347,
+      "loss": 0.8971,
+      "step": 4776
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018661537262588227,
+      "loss": 0.9568,
+      "step": 4777
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018660987379010634,
+      "loss": 0.9698,
+      "step": 4778
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018660437390606222,
+      "loss": 1.0179,
+      "step": 4779
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018659887297381648,
+      "loss": 0.859,
+      "step": 4780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018659337099343571,
+      "loss": 1.0174,
+      "step": 4781
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001865878679649865,
+      "loss": 0.9528,
+      "step": 4782
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018658236388853548,
+      "loss": 0.9638,
+      "step": 4783
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018657685876414919,
+      "loss": 0.8453,
+      "step": 4784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018657135259189433,
+      "loss": 1.0754,
+      "step": 4785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018656584537183753,
+      "loss": 1.0255,
+      "step": 4786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018656033710404542,
+      "loss": 1.0923,
+      "step": 4787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018655482778858468,
+      "loss": 1.0803,
+      "step": 4788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018654931742552203,
+      "loss": 0.977,
+      "step": 4789
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018654380601492408,
+      "loss": 0.953,
+      "step": 4790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018653829355685762,
+      "loss": 1.0893,
+      "step": 4791
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018653278005138936,
+      "loss": 0.8578,
+      "step": 4792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018652726549858598,
+      "loss": 0.7379,
+      "step": 4793
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018652174989851428,
+      "loss": 0.9477,
+      "step": 4794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018651623325124095,
+      "loss": 1.0051,
+      "step": 4795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001865107155568328,
+      "loss": 0.8709,
+      "step": 4796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018650519681535666,
+      "loss": 0.9567,
+      "step": 4797
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018649967702687924,
+      "loss": 0.846,
+      "step": 4798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001864941561914674,
+      "loss": 1.0119,
+      "step": 4799
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018648863430918795,
+      "loss": 0.9426,
+      "step": 4800
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 0.9415798187255859,
+      "eval_runtime": 850.1562,
+      "eval_samples_per_second": 28.301,
+      "eval_steps_per_second": 14.15,
+      "step": 4800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018648311138010773,
+      "loss": 1.032,
+      "step": 4801
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018647758740429352,
+      "loss": 0.9823,
+      "step": 4802
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001864720623818123,
+      "loss": 0.9794,
+      "step": 4803
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018646653631273084,
+      "loss": 0.8485,
+      "step": 4804
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018646100919711604,
+      "loss": 1.0537,
+      "step": 4805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018645548103503486,
+      "loss": 0.9541,
+      "step": 4806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018644995182655417,
+      "loss": 0.8379,
+      "step": 4807
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018644442157174083,
+      "loss": 1.0188,
+      "step": 4808
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018643889027066186,
+      "loss": 0.8834,
+      "step": 4809
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018643335792338416,
+      "loss": 0.8775,
+      "step": 4810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018642782452997473,
+      "loss": 1.0026,
+      "step": 4811
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018642229009050053,
+      "loss": 0.9389,
+      "step": 4812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018641675460502851,
+      "loss": 0.9206,
+      "step": 4813
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018641121807362571,
+      "loss": 0.7628,
+      "step": 4814
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018640568049635913,
+      "loss": 0.9022,
+      "step": 4815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018640014187329578,
+      "loss": 1.0668,
+      "step": 4816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001863946022045027,
+      "loss": 0.9727,
+      "step": 4817
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018638906149004693,
+      "loss": 1.0893,
+      "step": 4818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018638351972999557,
+      "loss": 1.0994,
+      "step": 4819
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018637797692441565,
+      "loss": 0.9108,
+      "step": 4820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018637243307337425,
+      "loss": 0.9566,
+      "step": 4821
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018636688817693853,
+      "loss": 0.9435,
+      "step": 4822
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018636134223517554,
+      "loss": 0.8338,
+      "step": 4823
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018635579524815246,
+      "loss": 0.9371,
+      "step": 4824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018635024721593636,
+      "loss": 0.8549,
+      "step": 4825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018634469813859442,
+      "loss": 0.9612,
+      "step": 4826
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018633914801619383,
+      "loss": 0.8485,
+      "step": 4827
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018633359684880172,
+      "loss": 1.0441,
+      "step": 4828
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018632804463648533,
+      "loss": 0.8438,
+      "step": 4829
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001863224913793118,
+      "loss": 0.9736,
+      "step": 4830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001863169370773484,
+      "loss": 0.9097,
+      "step": 4831
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001863113817306623,
+      "loss": 1.0442,
+      "step": 4832
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018630582533932078,
+      "loss": 1.0446,
+      "step": 4833
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018630026790339107,
+      "loss": 1.0225,
+      "step": 4834
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018629470942294044,
+      "loss": 1.0273,
+      "step": 4835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001862891498980362,
+      "loss": 0.9478,
+      "step": 4836
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018628358932874556,
+      "loss": 0.9346,
+      "step": 4837
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018627802771513589,
+      "loss": 0.9902,
+      "step": 4838
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018627246505727448,
+      "loss": 0.9867,
+      "step": 4839
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018626690135522864,
+      "loss": 0.939,
+      "step": 4840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018626133660906575,
+      "loss": 0.9403,
+      "step": 4841
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018625577081885313,
+      "loss": 0.9121,
+      "step": 4842
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018625020398465819,
+      "loss": 1.118,
+      "step": 4843
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001862446361065482,
+      "loss": 0.9533,
+      "step": 4844
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018623906718459064,
+      "loss": 0.8377,
+      "step": 4845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018623349721885295,
+      "loss": 0.8892,
+      "step": 4846
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018622792620940243,
+      "loss": 0.9534,
+      "step": 4847
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001862223541563066,
+      "loss": 0.9182,
+      "step": 4848
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018621678105963286,
+      "loss": 1.0104,
+      "step": 4849
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018621120691944865,
+      "loss": 0.811,
+      "step": 4850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018620563173582147,
+      "loss": 0.8516,
+      "step": 4851
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001862000555088188,
+      "loss": 0.8788,
+      "step": 4852
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018619447823850808,
+      "loss": 0.9743,
+      "step": 4853
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018618889992495687,
+      "loss": 0.9558,
+      "step": 4854
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018618332056823266,
+      "loss": 0.8442,
+      "step": 4855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018617774016840296,
+      "loss": 0.9631,
+      "step": 4856
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001861721587255354,
+      "loss": 0.9196,
+      "step": 4857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001861665762396974,
+      "loss": 0.9022,
+      "step": 4858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018616099271095662,
+      "loss": 0.8422,
+      "step": 4859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001861554081393806,
+      "loss": 0.85,
+      "step": 4860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000186149822525037,
+      "loss": 1.0507,
+      "step": 4861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018614423586799333,
+      "loss": 0.9016,
+      "step": 4862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018613864816831725,
+      "loss": 0.9938,
+      "step": 4863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001861330594260764,
+      "loss": 0.9589,
+      "step": 4864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001861274696413384,
+      "loss": 0.991,
+      "step": 4865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001861218788141709,
+      "loss": 0.897,
+      "step": 4866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001861162869446416,
+      "loss": 0.9565,
+      "step": 4867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001861106940328182,
+      "loss": 0.987,
+      "step": 4868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001861051000787683,
+      "loss": 0.921,
+      "step": 4869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018609950508255966,
+      "loss": 0.9412,
+      "step": 4870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018609390904426006,
+      "loss": 0.9824,
+      "step": 4871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018608831196393713,
+      "loss": 0.9909,
+      "step": 4872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018608271384165865,
+      "loss": 0.7496,
+      "step": 4873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018607711467749237,
+      "loss": 0.9177,
+      "step": 4874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018607151447150612,
+      "loss": 0.7464,
+      "step": 4875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001860659132237676,
+      "loss": 1.0569,
+      "step": 4876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018606031093434461,
+      "loss": 0.9982,
+      "step": 4877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000186054707603305,
+      "loss": 0.9107,
+      "step": 4878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018604910323071657,
+      "loss": 0.9391,
+      "step": 4879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018604349781664717,
+      "loss": 0.9128,
+      "step": 4880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018603789136116457,
+      "loss": 0.901,
+      "step": 4881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018603228386433672,
+      "loss": 0.9606,
+      "step": 4882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018602667532623146,
+      "loss": 1.0559,
+      "step": 4883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018602106574691667,
+      "loss": 0.9123,
+      "step": 4884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018601545512646021,
+      "loss": 0.9216,
+      "step": 4885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018600984346493004,
+      "loss": 1.0357,
+      "step": 4886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000186004230762394,
+      "loss": 0.8483,
+      "step": 4887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001859986170189201,
+      "loss": 0.9952,
+      "step": 4888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018599300223457629,
+      "loss": 1.0275,
+      "step": 4889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001859873864094305,
+      "loss": 1.0081,
+      "step": 4890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018598176954355068,
+      "loss": 0.9672,
+      "step": 4891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018597615163700484,
+      "loss": 0.9886,
+      "step": 4892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018597053268986098,
+      "loss": 0.9642,
+      "step": 4893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018596491270218707,
+      "loss": 0.9027,
+      "step": 4894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018595929167405118,
+      "loss": 0.9571,
+      "step": 4895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001859536696055213,
+      "loss": 0.8085,
+      "step": 4896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001859480464966655,
+      "loss": 0.9672,
+      "step": 4897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018594242234755185,
+      "loss": 0.8137,
+      "step": 4898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001859367971582484,
+      "loss": 0.934,
+      "step": 4899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018593117092882323,
+      "loss": 0.8616,
+      "step": 4900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018592554365934444,
+      "loss": 0.9274,
+      "step": 4901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018591991534988017,
+      "loss": 0.8808,
+      "step": 4902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018591428600049848,
+      "loss": 0.9727,
+      "step": 4903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018590865561126757,
+      "loss": 0.9702,
+      "step": 4904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018590302418225552,
+      "loss": 0.8641,
+      "step": 4905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018589739171353053,
+      "loss": 0.8937,
+      "step": 4906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018589175820516078,
+      "loss": 0.874,
+      "step": 4907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018588612365721442,
+      "loss": 0.9027,
+      "step": 4908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018588048806975969,
+      "loss": 0.9794,
+      "step": 4909
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018587485144286477,
+      "loss": 0.9252,
+      "step": 4910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001858692137765979,
+      "loss": 0.8876,
+      "step": 4911
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018586357507102726,
+      "loss": 0.9524,
+      "step": 4912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018585793532622117,
+      "loss": 1.0467,
+      "step": 4913
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018585229454224787,
+      "loss": 0.8988,
+      "step": 4914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018584665271917562,
+      "loss": 0.879,
+      "step": 4915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001858410098570727,
+      "loss": 0.9103,
+      "step": 4916
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001858353659560074,
+      "loss": 0.9735,
+      "step": 4917
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018582972101604804,
+      "loss": 0.918,
+      "step": 4918
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018582407503726299,
+      "loss": 0.8887,
+      "step": 4919
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018581842801972053,
+      "loss": 0.9069,
+      "step": 4920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000185812779963489,
+      "loss": 0.8266,
+      "step": 4921
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018580713086863677,
+      "loss": 0.8439,
+      "step": 4922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018580148073523223,
+      "loss": 0.923,
+      "step": 4923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001857958295633438,
+      "loss": 0.9764,
+      "step": 4924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018579017735303977,
+      "loss": 0.9971,
+      "step": 4925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018578452410438864,
+      "loss": 0.9417,
+      "step": 4926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018577886981745883,
+      "loss": 0.9611,
+      "step": 4927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018577321449231873,
+      "loss": 1.0013,
+      "step": 4928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018576755812903684,
+      "loss": 0.9203,
+      "step": 4929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001857619007276816,
+      "loss": 0.9982,
+      "step": 4930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018575624228832146,
+      "loss": 0.9704,
+      "step": 4931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001857505828110249,
+      "loss": 0.9817,
+      "step": 4932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018574492229586045,
+      "loss": 1.0291,
+      "step": 4933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018573926074289663,
+      "loss": 0.9509,
+      "step": 4934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018573359815220197,
+      "loss": 0.8653,
+      "step": 4935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018572793452384495,
+      "loss": 0.9133,
+      "step": 4936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018572226985789415,
+      "loss": 0.8587,
+      "step": 4937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018571660415441817,
+      "loss": 0.849,
+      "step": 4938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001857109374134855,
+      "loss": 0.9333,
+      "step": 4939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018570526963516482,
+      "loss": 0.9773,
+      "step": 4940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018569960081952464,
+      "loss": 0.8843,
+      "step": 4941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018569393096663366,
+      "loss": 0.8298,
+      "step": 4942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018568826007656043,
+      "loss": 0.911,
+      "step": 4943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018568258814937363,
+      "loss": 1.0452,
+      "step": 4944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001856769151851419,
+      "loss": 0.9144,
+      "step": 4945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018567124118393387,
+      "loss": 0.8296,
+      "step": 4946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018566556614581825,
+      "loss": 0.9392,
+      "step": 4947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001856598900708637,
+      "loss": 0.8864,
+      "step": 4948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018565421295913898,
+      "loss": 0.8185,
+      "step": 4949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018564853481071273,
+      "loss": 1.027,
+      "step": 4950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001856428556256537,
+      "loss": 0.9707,
+      "step": 4951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018563717540403065,
+      "loss": 0.9725,
+      "step": 4952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001856314941459123,
+      "loss": 0.9531,
+      "step": 4953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018562581185136744,
+      "loss": 0.9298,
+      "step": 4954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001856201285204648,
+      "loss": 0.9052,
+      "step": 4955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018561444415327323,
+      "loss": 0.9112,
+      "step": 4956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018560875874986147,
+      "loss": 0.9632,
+      "step": 4957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018560307231029835,
+      "loss": 0.9012,
+      "step": 4958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001855973848346527,
+      "loss": 0.9372,
+      "step": 4959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018559169632299338,
+      "loss": 1.0774,
+      "step": 4960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018558600677538923,
+      "loss": 0.9416,
+      "step": 4961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018558031619190905,
+      "loss": 0.879,
+      "step": 4962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001855746245726218,
+      "loss": 0.9104,
+      "step": 4963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018556893191759634,
+      "loss": 1.006,
+      "step": 4964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018556323822690153,
+      "loss": 0.8647,
+      "step": 4965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018555754350060635,
+      "loss": 0.8287,
+      "step": 4966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001855518477387797,
+      "loss": 0.9063,
+      "step": 4967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018554615094149046,
+      "loss": 1.0536,
+      "step": 4968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018554045310880768,
+      "loss": 0.9648,
+      "step": 4969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018553475424080024,
+      "loss": 0.8546,
+      "step": 4970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018552905433753716,
+      "loss": 0.9232,
+      "step": 4971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018552335339908744,
+      "loss": 0.8126,
+      "step": 4972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018551765142552003,
+      "loss": 0.9794,
+      "step": 4973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018551194841690395,
+      "loss": 0.8751,
+      "step": 4974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001855062443733083,
+      "loss": 1.0991,
+      "step": 4975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018550053929480202,
+      "loss": 0.9601,
+      "step": 4976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001854948331814542,
+      "loss": 0.7643,
+      "step": 4977
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018548912603333391,
+      "loss": 0.9928,
+      "step": 4978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018548341785051023,
+      "loss": 0.8997,
+      "step": 4979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018547770863305222,
+      "loss": 0.9498,
+      "step": 4980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018547199838102904,
+      "loss": 0.9925,
+      "step": 4981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018546628709450972,
+      "loss": 0.8018,
+      "step": 4982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018546057477356346,
+      "loss": 0.9219,
+      "step": 4983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018545486141825936,
+      "loss": 0.8363,
+      "step": 4984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018544914702866658,
+      "loss": 0.9755,
+      "step": 4985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018544343160485427,
+      "loss": 0.9339,
+      "step": 4986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018543771514689162,
+      "loss": 0.8997,
+      "step": 4987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018543199765484782,
+      "loss": 0.9129,
+      "step": 4988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018542627912879206,
+      "loss": 0.8797,
+      "step": 4989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018542055956879353,
+      "loss": 0.8235,
+      "step": 4990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018541483897492156,
+      "loss": 0.8798,
+      "step": 4991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018540911734724524,
+      "loss": 0.9633,
+      "step": 4992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018540339468583393,
+      "loss": 1.0441,
+      "step": 4993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001853976709907568,
+      "loss": 0.9882,
+      "step": 4994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018539194626208326,
+      "loss": 0.9157,
+      "step": 4995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001853862204998825,
+      "loss": 0.865,
+      "step": 4996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018538049370422381,
+      "loss": 1.0331,
+      "step": 4997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018537476587517654,
+      "loss": 1.0038,
+      "step": 4998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018536903701281003,
+      "loss": 0.9054,
+      "step": 4999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001853633071171936,
+      "loss": 0.8971,
+      "step": 5000
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 0.9278010129928589,
+      "eval_runtime": 845.3949,
+      "eval_samples_per_second": 28.46,
+      "eval_steps_per_second": 14.23,
+      "step": 5000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001853575761883966,
+      "loss": 0.9364,
+      "step": 5001
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018535184422648838,
+      "loss": 0.8449,
+      "step": 5002
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001853461112315383,
+      "loss": 0.8724,
+      "step": 5003
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018534037720361585,
+      "loss": 0.9286,
+      "step": 5004
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001853346421427903,
+      "loss": 0.8815,
+      "step": 5005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018532890604913113,
+      "loss": 0.9004,
+      "step": 5006
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018532316892270779,
+      "loss": 1.0734,
+      "step": 5007
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018531743076358965,
+      "loss": 1.0016,
+      "step": 5008
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001853116915718462,
+      "loss": 0.8374,
+      "step": 5009
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001853059513475469,
+      "loss": 0.9438,
+      "step": 5010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018530021009076123,
+      "loss": 0.9938,
+      "step": 5011
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018529446780155868,
+      "loss": 0.9373,
+      "step": 5012
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018528872448000876,
+      "loss": 0.9426,
+      "step": 5013
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018528298012618095,
+      "loss": 0.8517,
+      "step": 5014
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001852772347401448,
+      "loss": 0.8542,
+      "step": 5015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001852714883219698,
+      "loss": 1.0322,
+      "step": 5016
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001852657408717256,
+      "loss": 0.861,
+      "step": 5017
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001852599923894817,
+      "loss": 0.9546,
+      "step": 5018
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018525424287530766,
+      "loss": 0.9157,
+      "step": 5019
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018524849232927308,
+      "loss": 0.8747,
+      "step": 5020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001852427407514476,
+      "loss": 0.9298,
+      "step": 5021
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001852369881419008,
+      "loss": 0.9285,
+      "step": 5022
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001852312345007023,
+      "loss": 0.8405,
+      "step": 5023
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018522547982792174,
+      "loss": 0.8102,
+      "step": 5024
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018521972412362882,
+      "loss": 0.8593,
+      "step": 5025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018521396738789312,
+      "loss": 0.8104,
+      "step": 5026
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018520820962078437,
+      "loss": 0.9329,
+      "step": 5027
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018520245082237225,
+      "loss": 0.9529,
+      "step": 5028
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018519669099272646,
+      "loss": 1.0247,
+      "step": 5029
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001851909301319167,
+      "loss": 0.9986,
+      "step": 5030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001851851682400127,
+      "loss": 1.0668,
+      "step": 5031
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018517940531708425,
+      "loss": 0.9306,
+      "step": 5032
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018517364136320098,
+      "loss": 0.992,
+      "step": 5033
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001851678763784328,
+      "loss": 0.8859,
+      "step": 5034
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018516211036284937,
+      "loss": 0.7473,
+      "step": 5035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018515634331652052,
+      "loss": 0.894,
+      "step": 5036
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001851505752395161,
+      "loss": 0.8889,
+      "step": 5037
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018514480613190582,
+      "loss": 0.8734,
+      "step": 5038
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018513903599375956,
+      "loss": 0.9302,
+      "step": 5039
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001851332648251472,
+      "loss": 0.8132,
+      "step": 5040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018512749262613852,
+      "loss": 1.0381,
+      "step": 5041
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001851217193968034,
+      "loss": 0.9709,
+      "step": 5042
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018511594513721173,
+      "loss": 0.8414,
+      "step": 5043
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018511016984743342,
+      "loss": 1.0565,
+      "step": 5044
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001851043935275383,
+      "loss": 0.9436,
+      "step": 5045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018509861617759636,
+      "loss": 0.9115,
+      "step": 5046
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001850928377976775,
+      "loss": 0.9165,
+      "step": 5047
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018508705838785164,
+      "loss": 0.9849,
+      "step": 5048
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018508127794818874,
+      "loss": 1.0031,
+      "step": 5049
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018507549647875875,
+      "loss": 1.0663,
+      "step": 5050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018506971397963166,
+      "loss": 1.0475,
+      "step": 5051
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001850639304508775,
+      "loss": 0.9814,
+      "step": 5052
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018505814589256616,
+      "loss": 0.9399,
+      "step": 5053
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018505236030476777,
+      "loss": 0.9138,
+      "step": 5054
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018504657368755226,
+      "loss": 0.9509,
+      "step": 5055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018504078604098973,
+      "loss": 0.977,
+      "step": 5056
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018503499736515023,
+      "loss": 0.9977,
+      "step": 5057
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018502920766010376,
+      "loss": 0.8667,
+      "step": 5058
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018502341692592044,
+      "loss": 0.839,
+      "step": 5059
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018501762516267038,
+      "loss": 1.0146,
+      "step": 5060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018501183237042365,
+      "loss": 0.9511,
+      "step": 5061
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018500603854925036,
+      "loss": 0.9079,
+      "step": 5062
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018500024369922063,
+      "loss": 0.8824,
+      "step": 5063
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018499444782040466,
+      "loss": 1.0101,
+      "step": 5064
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018498865091287248,
+      "loss": 0.9796,
+      "step": 5065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018498285297669433,
+      "loss": 0.903,
+      "step": 5066
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001849770540119404,
+      "loss": 1.0603,
+      "step": 5067
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018497125401868082,
+      "loss": 0.8697,
+      "step": 5068
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018496545299698583,
+      "loss": 0.8192,
+      "step": 5069
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018495965094692565,
+      "loss": 0.9444,
+      "step": 5070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018495384786857045,
+      "loss": 0.8593,
+      "step": 5071
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018494804376199052,
+      "loss": 0.9285,
+      "step": 5072
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001849422386272561,
+      "loss": 0.8744,
+      "step": 5073
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018493643246443744,
+      "loss": 0.88,
+      "step": 5074
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018493062527360482,
+      "loss": 1.0427,
+      "step": 5075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018492481705482852,
+      "loss": 0.965,
+      "step": 5076
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018491900780817882,
+      "loss": 0.9592,
+      "step": 5077
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018491319753372607,
+      "loss": 0.8366,
+      "step": 5078
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018490738623154056,
+      "loss": 1.0013,
+      "step": 5079
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018490157390169265,
+      "loss": 0.9054,
+      "step": 5080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018489576054425268,
+      "loss": 0.8192,
+      "step": 5081
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000184889946159291,
+      "loss": 0.8872,
+      "step": 5082
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018488413074687802,
+      "loss": 1.0032,
+      "step": 5083
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001848783143070841,
+      "loss": 0.7284,
+      "step": 5084
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018487249683997963,
+      "loss": 0.9269,
+      "step": 5085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018486667834563504,
+      "loss": 0.9688,
+      "step": 5086
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018486085882412072,
+      "loss": 0.9024,
+      "step": 5087
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018485503827550713,
+      "loss": 0.7986,
+      "step": 5088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018484921669986474,
+      "loss": 0.9507,
+      "step": 5089
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018484339409726398,
+      "loss": 0.9697,
+      "step": 5090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018483757046777532,
+      "loss": 1.0527,
+      "step": 5091
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018483174581146927,
+      "loss": 0.9102,
+      "step": 5092
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001848259201284163,
+      "loss": 0.7914,
+      "step": 5093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018482009341868697,
+      "loss": 0.8953,
+      "step": 5094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001848142656823517,
+      "loss": 0.9629,
+      "step": 5095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018480843691948113,
+      "loss": 0.9563,
+      "step": 5096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018480260713014577,
+      "loss": 0.9249,
+      "step": 5097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001847967763144162,
+      "loss": 0.9015,
+      "step": 5098
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018479094447236292,
+      "loss": 0.9487,
+      "step": 5099
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001847851116040566,
+      "loss": 0.9248,
+      "step": 5100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018477927770956779,
+      "loss": 0.9861,
+      "step": 5101
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001847734427889671,
+      "loss": 0.9399,
+      "step": 5102
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018476760684232518,
+      "loss": 0.882,
+      "step": 5103
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018476176986971266,
+      "loss": 0.9265,
+      "step": 5104
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018475593187120015,
+      "loss": 0.949,
+      "step": 5105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018475009284685837,
+      "loss": 0.9342,
+      "step": 5106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001847442527967579,
+      "loss": 1.0091,
+      "step": 5107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018473841172096954,
+      "loss": 0.8778,
+      "step": 5108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001847325696195639,
+      "loss": 1.0708,
+      "step": 5109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018472672649261172,
+      "loss": 0.8983,
+      "step": 5110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001847208823401837,
+      "loss": 0.8944,
+      "step": 5111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001847150371623506,
+      "loss": 1.1054,
+      "step": 5112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018470919095918314,
+      "loss": 0.9318,
+      "step": 5113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018470334373075215,
+      "loss": 0.8798,
+      "step": 5114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018469749547712832,
+      "loss": 0.8654,
+      "step": 5115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018469164619838244,
+      "loss": 0.8897,
+      "step": 5116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018468579589458534,
+      "loss": 0.9559,
+      "step": 5117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018467994456580782,
+      "loss": 1.0848,
+      "step": 5118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018467409221212068,
+      "loss": 0.9095,
+      "step": 5119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018466823883359478,
+      "loss": 0.8872,
+      "step": 5120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018466238443030095,
+      "loss": 1.155,
+      "step": 5121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018465652900231003,
+      "loss": 0.859,
+      "step": 5122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018465067254969296,
+      "loss": 0.902,
+      "step": 5123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018464481507252055,
+      "loss": 0.9044,
+      "step": 5124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001846389565708637,
+      "loss": 1.0646,
+      "step": 5125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018463309704479334,
+      "loss": 0.9248,
+      "step": 5126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018462723649438037,
+      "loss": 0.8546,
+      "step": 5127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001846213749196958,
+      "loss": 0.9435,
+      "step": 5128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018461551232081046,
+      "loss": 0.8878,
+      "step": 5129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018460964869779537,
+      "loss": 0.9783,
+      "step": 5130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001846037840507215,
+      "loss": 0.8421,
+      "step": 5131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018459791837965978,
+      "loss": 1.0088,
+      "step": 5132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001845920516846813,
+      "loss": 0.9185,
+      "step": 5133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018458618396585696,
+      "loss": 1.1094,
+      "step": 5134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018458031522325787,
+      "loss": 1.0847,
+      "step": 5135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018457444545695503,
+      "loss": 1.0193,
+      "step": 5136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018456857466701945,
+      "loss": 0.863,
+      "step": 5137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018456270285352222,
+      "loss": 0.9436,
+      "step": 5138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001845568300165344,
+      "loss": 0.9096,
+      "step": 5139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018455095615612703,
+      "loss": 0.9409,
+      "step": 5140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018454508127237128,
+      "loss": 0.868,
+      "step": 5141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018453920536533824,
+      "loss": 0.8769,
+      "step": 5142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018453332843509899,
+      "loss": 0.9462,
+      "step": 5143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018452745048172465,
+      "loss": 0.9322,
+      "step": 5144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018452157150528645,
+      "loss": 1.022,
+      "step": 5145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018451569150585543,
+      "loss": 1.031,
+      "step": 5146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018450981048350284,
+      "loss": 0.7803,
+      "step": 5147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018450392843829985,
+      "loss": 0.8718,
+      "step": 5148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018449804537031762,
+      "loss": 0.9287,
+      "step": 5149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018449216127962737,
+      "loss": 0.9137,
+      "step": 5150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018448627616630029,
+      "loss": 0.8911,
+      "step": 5151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018448039003040768,
+      "loss": 0.8767,
+      "step": 5152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018447450287202073,
+      "loss": 0.9711,
+      "step": 5153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001844686146912107,
+      "loss": 0.9222,
+      "step": 5154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018446272548804885,
+      "loss": 0.9077,
+      "step": 5155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001844568352626065,
+      "loss": 0.9089,
+      "step": 5156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001844509440149549,
+      "loss": 0.8461,
+      "step": 5157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018444505174516532,
+      "loss": 0.8575,
+      "step": 5158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018443915845330918,
+      "loss": 0.9461,
+      "step": 5159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001844332641394577,
+      "loss": 0.8749,
+      "step": 5160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001844273688036823,
+      "loss": 0.8661,
+      "step": 5161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018442147244605427,
+      "loss": 0.8739,
+      "step": 5162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018441557506664505,
+      "loss": 0.8035,
+      "step": 5163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018440967666552596,
+      "loss": 0.8786,
+      "step": 5164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001844037772427684,
+      "loss": 1.0201,
+      "step": 5165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001843978767984438,
+      "loss": 0.9357,
+      "step": 5166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018439197533262353,
+      "loss": 0.886,
+      "step": 5167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018438607284537907,
+      "loss": 1.0229,
+      "step": 5168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018438016933678179,
+      "loss": 0.7932,
+      "step": 5169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018437426480690322,
+      "loss": 1.0298,
+      "step": 5170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001843683592558148,
+      "loss": 0.8462,
+      "step": 5171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018436245268358797,
+      "loss": 0.8438,
+      "step": 5172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018435654509029424,
+      "loss": 0.8878,
+      "step": 5173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018435063647600513,
+      "loss": 0.8983,
+      "step": 5174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001843447268407921,
+      "loss": 0.7579,
+      "step": 5175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018433881618472677,
+      "loss": 0.9989,
+      "step": 5176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001843329045078806,
+      "loss": 0.9076,
+      "step": 5177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018432699181032518,
+      "loss": 0.7813,
+      "step": 5178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018432107809213206,
+      "loss": 0.8719,
+      "step": 5179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001843151633533728,
+      "loss": 0.9092,
+      "step": 5180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000184309247594119,
+      "loss": 0.9491,
+      "step": 5181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018430333081444227,
+      "loss": 0.9523,
+      "step": 5182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018429741301441421,
+      "loss": 0.8487,
+      "step": 5183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018429149419410648,
+      "loss": 0.9761,
+      "step": 5184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018428557435359064,
+      "loss": 0.9857,
+      "step": 5185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018427965349293842,
+      "loss": 0.9857,
+      "step": 5186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018427373161222145,
+      "loss": 0.9517,
+      "step": 5187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018426780871151142,
+      "loss": 0.9014,
+      "step": 5188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018426188479087998,
+      "loss": 1.039,
+      "step": 5189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018425595985039885,
+      "loss": 0.9014,
+      "step": 5190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018425003389013975,
+      "loss": 0.9736,
+      "step": 5191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001842441069101744,
+      "loss": 1.0209,
+      "step": 5192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018423817891057454,
+      "loss": 0.9801,
+      "step": 5193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001842322498914119,
+      "loss": 0.8946,
+      "step": 5194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018422631985275825,
+      "loss": 0.9727,
+      "step": 5195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018422038879468535,
+      "loss": 1.0126,
+      "step": 5196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018421445671726504,
+      "loss": 0.9202,
+      "step": 5197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018420852362056905,
+      "loss": 0.894,
+      "step": 5198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001842025895046692,
+      "loss": 1.0205,
+      "step": 5199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018419665436963738,
+      "loss": 0.8258,
+      "step": 5200
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 0.9124149084091187,
+      "eval_runtime": 850.6816,
+      "eval_samples_per_second": 28.283,
+      "eval_steps_per_second": 14.142,
+      "step": 5200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018419071821554534,
+      "loss": 0.8546,
+      "step": 5201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000184184781042465,
+      "loss": 1.007,
+      "step": 5202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018417884285046817,
+      "loss": 0.9352,
+      "step": 5203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018417290363962672,
+      "loss": 0.9328,
+      "step": 5204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018416696341001253,
+      "loss": 0.9102,
+      "step": 5205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018416102216169757,
+      "loss": 0.897,
+      "step": 5206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018415507989475366,
+      "loss": 0.9325,
+      "step": 5207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001841491366092528,
+      "loss": 0.9335,
+      "step": 5208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018414319230526682,
+      "loss": 0.8695,
+      "step": 5209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018413724698286777,
+      "loss": 0.8891,
+      "step": 5210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018413130064212754,
+      "loss": 0.9424,
+      "step": 5211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018412535328311814,
+      "loss": 0.7906,
+      "step": 5212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018411940490591154,
+      "loss": 0.9037,
+      "step": 5213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018411345551057969,
+      "loss": 0.9362,
+      "step": 5214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018410750509719468,
+      "loss": 0.9024,
+      "step": 5215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018410155366582848,
+      "loss": 1.0008,
+      "step": 5216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018409560121655314,
+      "loss": 0.9844,
+      "step": 5217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001840896477494407,
+      "loss": 0.7814,
+      "step": 5218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001840836932645632,
+      "loss": 0.8492,
+      "step": 5219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018407773776199274,
+      "loss": 0.9626,
+      "step": 5220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018407178124180135,
+      "loss": 1.0041,
+      "step": 5221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001840658237040612,
+      "loss": 0.9342,
+      "step": 5222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018405986514884434,
+      "loss": 0.7876,
+      "step": 5223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001840539055762229,
+      "loss": 1.0233,
+      "step": 5224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000184047944986269,
+      "loss": 0.8357,
+      "step": 5225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001840419833790548,
+      "loss": 0.8691,
+      "step": 5226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018403602075465244,
+      "loss": 1.06,
+      "step": 5227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001840300571131341,
+      "loss": 0.9024,
+      "step": 5228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018402409245457202,
+      "loss": 0.815,
+      "step": 5229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018401812677903826,
+      "loss": 0.8902,
+      "step": 5230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018401216008660513,
+      "loss": 0.9936,
+      "step": 5231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001840061923773448,
+      "loss": 0.9709,
+      "step": 5232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018400022365132955,
+      "loss": 0.8802,
+      "step": 5233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018399425390863155,
+      "loss": 0.873,
+      "step": 5234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018398828314932305,
+      "loss": 0.9463,
+      "step": 5235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001839823113734764,
+      "loss": 0.7919,
+      "step": 5236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018397633858116384,
+      "loss": 1.0016,
+      "step": 5237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018397036477245762,
+      "loss": 0.9528,
+      "step": 5238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001839643899474301,
+      "loss": 0.8344,
+      "step": 5239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018395841410615352,
+      "loss": 0.9042,
+      "step": 5240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018395243724870033,
+      "loss": 0.8819,
+      "step": 5241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018394645937514276,
+      "loss": 0.955,
+      "step": 5242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001839404804855532,
+      "loss": 0.9498,
+      "step": 5243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018393450058000403,
+      "loss": 0.93,
+      "step": 5244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018392851965856762,
+      "loss": 0.7399,
+      "step": 5245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001839225377213163,
+      "loss": 0.9828,
+      "step": 5246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018391655476832257,
+      "loss": 0.7762,
+      "step": 5247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018391057079965875,
+      "loss": 0.8717,
+      "step": 5248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001839045858153974,
+      "loss": 0.8369,
+      "step": 5249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018389859981561077,
+      "loss": 0.8239,
+      "step": 5250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018389261280037146,
+      "loss": 0.7141,
+      "step": 5251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001838866247697519,
+      "loss": 0.8875,
+      "step": 5252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001838806357238245,
+      "loss": 0.9077,
+      "step": 5253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018387464566266184,
+      "loss": 0.978,
+      "step": 5254
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018386865458633634,
+      "loss": 0.8174,
+      "step": 5255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018386266249492057,
+      "loss": 0.8187,
+      "step": 5256
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018385666938848703,
+      "loss": 0.9888,
+      "step": 5257
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018385067526710823,
+      "loss": 0.8135,
+      "step": 5258
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018384468013085678,
+      "loss": 0.907,
+      "step": 5259
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001838386839798052,
+      "loss": 0.8704,
+      "step": 5260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018383268681402607,
+      "loss": 0.8859,
+      "step": 5261
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000183826688633592,
+      "loss": 0.9141,
+      "step": 5262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018382068943857552,
+      "loss": 0.8527,
+      "step": 5263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018381468922904932,
+      "loss": 0.9306,
+      "step": 5264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018380868800508597,
+      "loss": 1.1486,
+      "step": 5265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018380268576675813,
+      "loss": 0.7843,
+      "step": 5266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018379668251413845,
+      "loss": 0.8891,
+      "step": 5267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018379067824729955,
+      "loss": 0.9216,
+      "step": 5268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018378467296631417,
+      "loss": 0.7999,
+      "step": 5269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018377866667125493,
+      "loss": 0.8996,
+      "step": 5270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018377265936219458,
+      "loss": 0.9794,
+      "step": 5271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018376665103920576,
+      "loss": 1.0184,
+      "step": 5272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018376064170236124,
+      "loss": 0.8341,
+      "step": 5273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018375463135173375,
+      "loss": 0.8243,
+      "step": 5274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018374861998739602,
+      "loss": 0.8888,
+      "step": 5275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018374260760942087,
+      "loss": 0.9581,
+      "step": 5276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018373659421788095,
+      "loss": 0.9553,
+      "step": 5277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018373057981284913,
+      "loss": 0.8866,
+      "step": 5278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001837245643943982,
+      "loss": 0.9081,
+      "step": 5279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018371854796260092,
+      "loss": 0.9216,
+      "step": 5280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018371253051753015,
+      "loss": 0.9328,
+      "step": 5281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018370651205925872,
+      "loss": 0.9197,
+      "step": 5282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018370049258785944,
+      "loss": 0.8961,
+      "step": 5283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001836944721034052,
+      "loss": 0.9182,
+      "step": 5284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018368845060596887,
+      "loss": 0.8341,
+      "step": 5285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001836824280956233,
+      "loss": 1.041,
+      "step": 5286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001836764045724414,
+      "loss": 0.9912,
+      "step": 5287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018367038003649607,
+      "loss": 1.0606,
+      "step": 5288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018366435448786023,
+      "loss": 0.993,
+      "step": 5289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018365832792660683,
+      "loss": 0.7631,
+      "step": 5290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018365230035280878,
+      "loss": 0.9282,
+      "step": 5291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018364627176653902,
+      "loss": 0.9118,
+      "step": 5292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018364024216787058,
+      "loss": 0.8949,
+      "step": 5293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018363421155687638,
+      "loss": 0.9582,
+      "step": 5294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018362817993362944,
+      "loss": 0.8854,
+      "step": 5295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018362214729820275,
+      "loss": 0.9356,
+      "step": 5296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018361611365066933,
+      "loss": 0.9592,
+      "step": 5297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001836100789911022,
+      "loss": 0.9675,
+      "step": 5298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018360404331957438,
+      "loss": 0.9098,
+      "step": 5299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018359800663615897,
+      "loss": 0.7973,
+      "step": 5300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000183591968940929,
+      "loss": 1.0094,
+      "step": 5301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018358593023395757,
+      "loss": 0.9039,
+      "step": 5302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018357989051531775,
+      "loss": 0.8987,
+      "step": 5303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018357384978508266,
+      "loss": 0.905,
+      "step": 5304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018356780804332535,
+      "loss": 0.8152,
+      "step": 5305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018356176529011904,
+      "loss": 0.9327,
+      "step": 5306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018355572152553681,
+      "loss": 0.8407,
+      "step": 5307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018354967674965182,
+      "loss": 0.9301,
+      "step": 5308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018354363096253722,
+      "loss": 0.9479,
+      "step": 5309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001835375841642662,
+      "loss": 0.8751,
+      "step": 5310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018353153635491197,
+      "loss": 0.864,
+      "step": 5311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018352548753454767,
+      "loss": 0.7949,
+      "step": 5312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018351943770324655,
+      "loss": 0.8207,
+      "step": 5313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018351338686108183,
+      "loss": 0.9128,
+      "step": 5314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018350733500812675,
+      "loss": 0.8046,
+      "step": 5315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018350128214445453,
+      "loss": 1.0392,
+      "step": 5316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018349522827013849,
+      "loss": 0.8948,
+      "step": 5317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018348917338525183,
+      "loss": 0.9132,
+      "step": 5318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018348311748986787,
+      "loss": 0.8715,
+      "step": 5319
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018347706058405988,
+      "loss": 0.9139,
+      "step": 5320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001834710026679012,
+      "loss": 0.9285,
+      "step": 5321
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018346494374146515,
+      "loss": 0.9541,
+      "step": 5322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018345888380482506,
+      "loss": 0.8796,
+      "step": 5323
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018345282285805428,
+      "loss": 0.8976,
+      "step": 5324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018344676090122612,
+      "loss": 0.8949,
+      "step": 5325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000183440697934414,
+      "loss": 1.0599,
+      "step": 5326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018343463395769128,
+      "loss": 0.9618,
+      "step": 5327
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001834285689711314,
+      "loss": 0.9201,
+      "step": 5328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018342250297480772,
+      "loss": 0.8956,
+      "step": 5329
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018341643596879367,
+      "loss": 0.8572,
+      "step": 5330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018341036795316266,
+      "loss": 1.006,
+      "step": 5331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018340429892798818,
+      "loss": 0.954,
+      "step": 5332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001833982288933436,
+      "loss": 0.8352,
+      "step": 5333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018339215784930253,
+      "loss": 0.7926,
+      "step": 5334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018338608579593832,
+      "loss": 0.9511,
+      "step": 5335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001833800127333245,
+      "loss": 0.8257,
+      "step": 5336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018337393866153462,
+      "loss": 0.9142,
+      "step": 5337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018336786358064212,
+      "loss": 0.9392,
+      "step": 5338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001833617874907206,
+      "loss": 0.7906,
+      "step": 5339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018335571039184356,
+      "loss": 0.9095,
+      "step": 5340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018334963228408454,
+      "loss": 0.7985,
+      "step": 5341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018334355316751717,
+      "loss": 0.8399,
+      "step": 5342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018333747304221493,
+      "loss": 0.9221,
+      "step": 5343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001833313919082515,
+      "loss": 0.8446,
+      "step": 5344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018332530976570045,
+      "loss": 0.7549,
+      "step": 5345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018331922661463538,
+      "loss": 0.8845,
+      "step": 5346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018331314245512994,
+      "loss": 0.9056,
+      "step": 5347
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018330705728725774,
+      "loss": 0.9228,
+      "step": 5348
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018330097111109246,
+      "loss": 0.9489,
+      "step": 5349
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018329488392670773,
+      "loss": 0.8475,
+      "step": 5350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018328879573417726,
+      "loss": 0.891,
+      "step": 5351
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001832827065335747,
+      "loss": 1.0309,
+      "step": 5352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018327661632497382,
+      "loss": 0.8981,
+      "step": 5353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018327052510844823,
+      "loss": 0.8925,
+      "step": 5354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018326443288407175,
+      "loss": 0.8703,
+      "step": 5355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018325833965191808,
+      "loss": 0.9412,
+      "step": 5356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001832522454120609,
+      "loss": 0.9425,
+      "step": 5357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001832461501645741,
+      "loss": 0.9775,
+      "step": 5358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018324005390953137,
+      "loss": 0.8827,
+      "step": 5359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001832339566470065,
+      "loss": 0.8956,
+      "step": 5360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001832278583770733,
+      "loss": 0.8488,
+      "step": 5361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001832217590998056,
+      "loss": 0.8231,
+      "step": 5362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018321565881527715,
+      "loss": 1.0282,
+      "step": 5363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018320955752356186,
+      "loss": 0.9079,
+      "step": 5364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018320345522473356,
+      "loss": 0.9329,
+      "step": 5365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018319735191886608,
+      "loss": 1.0685,
+      "step": 5366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018319124760603333,
+      "loss": 0.9221,
+      "step": 5367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018318514228630916,
+      "loss": 0.7715,
+      "step": 5368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018317903595976747,
+      "loss": 0.688,
+      "step": 5369
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018317292862648213,
+      "loss": 0.8857,
+      "step": 5370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001831668202865272,
+      "loss": 0.9736,
+      "step": 5371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018316071093997643,
+      "loss": 0.8532,
+      "step": 5372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018315460058690385,
+      "loss": 0.8896,
+      "step": 5373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018314848922738346,
+      "loss": 0.9503,
+      "step": 5374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018314237686148913,
+      "loss": 0.8637,
+      "step": 5375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018313626348929495,
+      "loss": 0.8502,
+      "step": 5376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001831301491108748,
+      "loss": 0.792,
+      "step": 5377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001831240337263027,
+      "loss": 0.9569,
+      "step": 5378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018311791733565275,
+      "loss": 0.8694,
+      "step": 5379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018311179993899893,
+      "loss": 0.7832,
+      "step": 5380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001831056815364153,
+      "loss": 0.9121,
+      "step": 5381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018309956212797588,
+      "loss": 0.773,
+      "step": 5382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018309344171375474,
+      "loss": 0.8741,
+      "step": 5383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018308732029382596,
+      "loss": 0.8675,
+      "step": 5384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018308119786826366,
+      "loss": 0.9131,
+      "step": 5385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001830750744371419,
+      "loss": 0.9246,
+      "step": 5386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018306895000053483,
+      "loss": 0.8092,
+      "step": 5387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018306282455851655,
+      "loss": 0.8884,
+      "step": 5388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001830566981111612,
+      "loss": 1.0288,
+      "step": 5389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018305057065854297,
+      "loss": 0.8366,
+      "step": 5390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018304444220073595,
+      "loss": 0.939,
+      "step": 5391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018303831273781438,
+      "loss": 0.9871,
+      "step": 5392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018303218226985241,
+      "loss": 0.8718,
+      "step": 5393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018302605079692426,
+      "loss": 0.9341,
+      "step": 5394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001830199183191041,
+      "loss": 0.8805,
+      "step": 5395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018301378483646623,
+      "loss": 0.8352,
+      "step": 5396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001830076503490848,
+      "loss": 0.9105,
+      "step": 5397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018300151485703414,
+      "loss": 0.9487,
+      "step": 5398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018299537836038846,
+      "loss": 0.7758,
+      "step": 5399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018298924085922202,
+      "loss": 0.8038,
+      "step": 5400
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 0.899122416973114,
+      "eval_runtime": 849.7173,
+      "eval_samples_per_second": 28.315,
+      "eval_steps_per_second": 14.158,
+      "step": 5400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018298310235360913,
+      "loss": 0.9836,
+      "step": 5401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001829769628436241,
+      "loss": 0.9046,
+      "step": 5402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001829708223293412,
+      "loss": 0.8172,
+      "step": 5403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018296468081083477,
+      "loss": 0.9437,
+      "step": 5404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018295853828817916,
+      "loss": 1.0201,
+      "step": 5405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001829523947614487,
+      "loss": 0.9644,
+      "step": 5406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018294625023071773,
+      "loss": 0.8125,
+      "step": 5407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018294010469606064,
+      "loss": 0.9892,
+      "step": 5408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001829339581575518,
+      "loss": 0.8811,
+      "step": 5409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001829278106152656,
+      "loss": 0.9135,
+      "step": 5410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001829216620692765,
+      "loss": 0.9742,
+      "step": 5411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018291551251965884,
+      "loss": 0.7894,
+      "step": 5412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001829093619664871,
+      "loss": 0.942,
+      "step": 5413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018290321040983568,
+      "loss": 0.9459,
+      "step": 5414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001828970578497791,
+      "loss": 0.8961,
+      "step": 5415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018289090428639173,
+      "loss": 0.8094,
+      "step": 5416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018288474971974816,
+      "loss": 0.8863,
+      "step": 5417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001828785941499228,
+      "loss": 0.9483,
+      "step": 5418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018287243757699018,
+      "loss": 0.9709,
+      "step": 5419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018286628000102484,
+      "loss": 0.8196,
+      "step": 5420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018286012142210124,
+      "loss": 0.8417,
+      "step": 5421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018285396184029398,
+      "loss": 0.9253,
+      "step": 5422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001828478012556776,
+      "loss": 0.7629,
+      "step": 5423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018284163966832666,
+      "loss": 0.8006,
+      "step": 5424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001828354770783157,
+      "loss": 1.0519,
+      "step": 5425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018282931348571936,
+      "loss": 0.8797,
+      "step": 5426
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018282314889061225,
+      "loss": 0.9017,
+      "step": 5427
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018281698329306888,
+      "loss": 0.9655,
+      "step": 5428
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000182810816693164,
+      "loss": 0.7547,
+      "step": 5429
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001828046490909722,
+      "loss": 0.8868,
+      "step": 5430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001827984804865681,
+      "loss": 0.8489,
+      "step": 5431
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018279231088002637,
+      "loss": 0.9692,
+      "step": 5432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018278614027142172,
+      "loss": 0.8963,
+      "step": 5433
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001827799686608288,
+      "loss": 0.8391,
+      "step": 5434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018277379604832226,
+      "loss": 0.8663,
+      "step": 5435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018276762243397694,
+      "loss": 0.9649,
+      "step": 5436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018276144781786746,
+      "loss": 0.8383,
+      "step": 5437
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018275527220006858,
+      "loss": 0.8687,
+      "step": 5438
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018274909558065508,
+      "loss": 0.9568,
+      "step": 5439
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018274291795970164,
+      "loss": 0.8891,
+      "step": 5440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001827367393372831,
+      "loss": 0.9934,
+      "step": 5441
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001827305597134742,
+      "loss": 0.8793,
+      "step": 5442
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018272437908834974,
+      "loss": 0.8411,
+      "step": 5443
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018271819746198454,
+      "loss": 0.9624,
+      "step": 5444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018271201483445343,
+      "loss": 0.8955,
+      "step": 5445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018270583120583124,
+      "loss": 0.9034,
+      "step": 5446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018269964657619274,
+      "loss": 0.9382,
+      "step": 5447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001826934609456129,
+      "loss": 0.8468,
+      "step": 5448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018268727431416647,
+      "loss": 1.0533,
+      "step": 5449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018268108668192842,
+      "loss": 0.9253,
+      "step": 5450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018267489804897358,
+      "loss": 0.8867,
+      "step": 5451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018266870841537693,
+      "loss": 0.8714,
+      "step": 5452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001826625177812133,
+      "loss": 0.9681,
+      "step": 5453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001826563261465577,
+      "loss": 0.8807,
+      "step": 5454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018265013351148496,
+      "loss": 0.9564,
+      "step": 5455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018264393987607014,
+      "loss": 0.9054,
+      "step": 5456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018263774524038815,
+      "loss": 0.8398,
+      "step": 5457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018263154960451394,
+      "loss": 0.873,
+      "step": 5458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018262535296852258,
+      "loss": 0.9424,
+      "step": 5459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000182619155332489,
+      "loss": 0.8369,
+      "step": 5460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018261295669648822,
+      "loss": 0.8855,
+      "step": 5461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001826067570605953,
+      "loss": 0.8199,
+      "step": 5462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018260055642488525,
+      "loss": 0.9115,
+      "step": 5463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001825943547894331,
+      "loss": 0.9044,
+      "step": 5464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018258815215431396,
+      "loss": 0.8973,
+      "step": 5465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018258194851960286,
+      "loss": 0.9324,
+      "step": 5466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001825757438853749,
+      "loss": 0.8681,
+      "step": 5467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001825695382517052,
+      "loss": 0.856,
+      "step": 5468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018256333161866884,
+      "loss": 0.7745,
+      "step": 5469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001825571239863409,
+      "loss": 0.9477,
+      "step": 5470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001825509153547966,
+      "loss": 0.8584,
+      "step": 5471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018254470572411105,
+      "loss": 0.8619,
+      "step": 5472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018253849509435938,
+      "loss": 0.9325,
+      "step": 5473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001825322834656168,
+      "loss": 0.8828,
+      "step": 5474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018252607083795846,
+      "loss": 0.9007,
+      "step": 5475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018251985721145958,
+      "loss": 0.9559,
+      "step": 5476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018251364258619533,
+      "loss": 0.8992,
+      "step": 5477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018250742696224097,
+      "loss": 0.9248,
+      "step": 5478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001825012103396717,
+      "loss": 0.9545,
+      "step": 5479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018249499271856277,
+      "loss": 0.8965,
+      "step": 5480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018248877409898944,
+      "loss": 0.8511,
+      "step": 5481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018248255448102696,
+      "loss": 0.8674,
+      "step": 5482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018247633386475064,
+      "loss": 0.742,
+      "step": 5483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018247011225023572,
+      "loss": 0.8128,
+      "step": 5484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018246388963755758,
+      "loss": 0.8729,
+      "step": 5485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018245766602679146,
+      "loss": 0.9479,
+      "step": 5486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001824514414180127,
+      "loss": 0.9471,
+      "step": 5487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018244521581129667,
+      "loss": 0.7771,
+      "step": 5488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001824389892067187,
+      "loss": 0.8886,
+      "step": 5489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018243276160435413,
+      "loss": 1.0274,
+      "step": 5490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001824265330042784,
+      "loss": 0.9936,
+      "step": 5491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001824203034065668,
+      "loss": 0.8406,
+      "step": 5492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001824140728112948,
+      "loss": 0.8956,
+      "step": 5493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018240784121853785,
+      "loss": 0.9343,
+      "step": 5494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018240160862837126,
+      "loss": 0.8308,
+      "step": 5495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018239537504087053,
+      "loss": 0.8675,
+      "step": 5496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018238914045611116,
+      "loss": 0.8315,
+      "step": 5497
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001823829048741685,
+      "loss": 0.8052,
+      "step": 5498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018237666829511805,
+      "loss": 0.8987,
+      "step": 5499
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018237043071903534,
+      "loss": 0.812,
+      "step": 5500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018236419214599586,
+      "loss": 0.8841,
+      "step": 5501
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018235795257607508,
+      "loss": 0.8221,
+      "step": 5502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018235171200934854,
+      "loss": 0.929,
+      "step": 5503
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018234547044589176,
+      "loss": 0.8337,
+      "step": 5504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018233922788578032,
+      "loss": 0.9436,
+      "step": 5505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018233298432908973,
+      "loss": 0.8612,
+      "step": 5506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018232673977589556,
+      "loss": 0.9478,
+      "step": 5507
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001823204942262734,
+      "loss": 0.927,
+      "step": 5508
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001823142476802989,
+      "loss": 0.8585,
+      "step": 5509
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001823080001380476,
+      "loss": 1.0036,
+      "step": 5510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018230175159959512,
+      "loss": 0.8076,
+      "step": 5511
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001822955020650171,
+      "loss": 0.8486,
+      "step": 5512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018228925153438916,
+      "loss": 0.9984,
+      "step": 5513
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018228300000778695,
+      "loss": 0.9664,
+      "step": 5514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001822767474852862,
+      "loss": 0.7827,
+      "step": 5515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018227049396696252,
+      "loss": 1.0872,
+      "step": 5516
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001822642394528916,
+      "loss": 0.8793,
+      "step": 5517
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018225798394314922,
+      "loss": 0.8649,
+      "step": 5518
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018225172743781097,
+      "loss": 0.9127,
+      "step": 5519
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018224546993695265,
+      "loss": 0.9131,
+      "step": 5520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018223921144064998,
+      "loss": 0.9312,
+      "step": 5521
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018223295194897874,
+      "loss": 0.896,
+      "step": 5522
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018222669146201464,
+      "loss": 0.9066,
+      "step": 5523
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018222042997983348,
+      "loss": 0.7336,
+      "step": 5524
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018221416750251103,
+      "loss": 0.7888,
+      "step": 5525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001822079040301231,
+      "loss": 0.891,
+      "step": 5526
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001822016395627455,
+      "loss": 0.9088,
+      "step": 5527
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018219537410045405,
+      "loss": 0.8895,
+      "step": 5528
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018218910764332457,
+      "loss": 0.9841,
+      "step": 5529
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018218284019143292,
+      "loss": 0.8869,
+      "step": 5530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018217657174485495,
+      "loss": 0.9891,
+      "step": 5531
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018217030230366653,
+      "loss": 0.8869,
+      "step": 5532
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018216403186794352,
+      "loss": 0.8907,
+      "step": 5533
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018215776043776187,
+      "loss": 0.8172,
+      "step": 5534
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018215148801319742,
+      "loss": 0.8386,
+      "step": 5535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018214521459432615,
+      "loss": 0.7674,
+      "step": 5536
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018213894018122395,
+      "loss": 0.9008,
+      "step": 5537
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018213266477396675,
+      "loss": 0.8625,
+      "step": 5538
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001821263883726305,
+      "loss": 0.9909,
+      "step": 5539
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018212011097729124,
+      "loss": 0.7962,
+      "step": 5540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018211383258802486,
+      "loss": 0.9063,
+      "step": 5541
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001821075532049074,
+      "loss": 0.7959,
+      "step": 5542
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018210127282801484,
+      "loss": 0.8746,
+      "step": 5543
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018209499145742318,
+      "loss": 0.9035,
+      "step": 5544
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001820887090932085,
+      "loss": 0.9045,
+      "step": 5545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018208242573544676,
+      "loss": 0.83,
+      "step": 5546
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018207614138421408,
+      "loss": 0.9046,
+      "step": 5547
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018206985603958647,
+      "loss": 0.9445,
+      "step": 5548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018206356970164005,
+      "loss": 0.8474,
+      "step": 5549
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018205728237045087,
+      "loss": 0.8553,
+      "step": 5550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018205099404609504,
+      "loss": 0.8901,
+      "step": 5551
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001820447047286487,
+      "loss": 0.896,
+      "step": 5552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018203841441818787,
+      "loss": 0.8978,
+      "step": 5553
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001820321231147888,
+      "loss": 1.1079,
+      "step": 5554
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001820258308185276,
+      "loss": 0.8314,
+      "step": 5555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018201953752948043,
+      "loss": 0.8897,
+      "step": 5556
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001820132432477234,
+      "loss": 0.8185,
+      "step": 5557
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018200694797333275,
+      "loss": 0.8066,
+      "step": 5558
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018200065170638466,
+      "loss": 0.8868,
+      "step": 5559
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001819943544469554,
+      "loss": 0.8155,
+      "step": 5560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018198805619512107,
+      "loss": 0.9196,
+      "step": 5561
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018198175695095797,
+      "loss": 0.9261,
+      "step": 5562
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018197545671454232,
+      "loss": 0.8924,
+      "step": 5563
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001819691554859504,
+      "loss": 0.8677,
+      "step": 5564
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018196285326525845,
+      "loss": 0.8779,
+      "step": 5565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018195655005254273,
+      "loss": 0.7575,
+      "step": 5566
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018195024584787957,
+      "loss": 0.9815,
+      "step": 5567
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018194394065134528,
+      "loss": 0.8904,
+      "step": 5568
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018193763446301616,
+      "loss": 0.984,
+      "step": 5569
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001819313272829685,
+      "loss": 1.0031,
+      "step": 5570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018192501911127868,
+      "loss": 0.9291,
+      "step": 5571
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018191870994802302,
+      "loss": 0.9262,
+      "step": 5572
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001819123997932779,
+      "loss": 0.9029,
+      "step": 5573
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018190608864711974,
+      "loss": 0.8847,
+      "step": 5574
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018189977650962482,
+      "loss": 0.7796,
+      "step": 5575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018189346338086962,
+      "loss": 0.81,
+      "step": 5576
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001818871492609305,
+      "loss": 0.9532,
+      "step": 5577
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018188083414988392,
+      "loss": 0.9963,
+      "step": 5578
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018187451804780634,
+      "loss": 0.8572,
+      "step": 5579
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018186820095477413,
+      "loss": 0.8624,
+      "step": 5580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018186188287086378,
+      "loss": 0.9649,
+      "step": 5581
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018185556379615177,
+      "loss": 0.8248,
+      "step": 5582
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018184924373071458,
+      "loss": 0.8575,
+      "step": 5583
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018184292267462872,
+      "loss": 1.0309,
+      "step": 5584
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018183660062797064,
+      "loss": 0.8314,
+      "step": 5585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001818302775908169,
+      "loss": 0.9859,
+      "step": 5586
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018182395356324406,
+      "loss": 0.8554,
+      "step": 5587
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018181762854532858,
+      "loss": 0.8087,
+      "step": 5588
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001818113025371471,
+      "loss": 0.8157,
+      "step": 5589
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001818049755387761,
+      "loss": 0.9532,
+      "step": 5590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018179864755029225,
+      "loss": 0.9486,
+      "step": 5591
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018179231857177208,
+      "loss": 0.9124,
+      "step": 5592
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001817859886032922,
+      "loss": 0.9266,
+      "step": 5593
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018177965764492923,
+      "loss": 0.8231,
+      "step": 5594
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018177332569675977,
+      "loss": 0.9301,
+      "step": 5595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001817669927588605,
+      "loss": 0.9619,
+      "step": 5596
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018176065883130805,
+      "loss": 0.8332,
+      "step": 5597
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001817543239141791,
+      "loss": 0.9574,
+      "step": 5598
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001817479880075503,
+      "loss": 0.8777,
+      "step": 5599
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018174165111149832,
+      "loss": 0.9663,
+      "step": 5600
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 0.8840734362602234,
+      "eval_runtime": 851.4939,
+      "eval_samples_per_second": 28.256,
+      "eval_steps_per_second": 14.128,
+      "step": 5600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018173531322609988,
+      "loss": 0.8092,
+      "step": 5601
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018172897435143174,
+      "loss": 0.7464,
+      "step": 5602
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018172263448757052,
+      "loss": 0.7903,
+      "step": 5603
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018171629363459302,
+      "loss": 0.9113,
+      "step": 5604
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018170995179257597,
+      "loss": 0.8363,
+      "step": 5605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018170360896159612,
+      "loss": 0.9035,
+      "step": 5606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018169726514173027,
+      "loss": 0.9214,
+      "step": 5607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018169092033305516,
+      "loss": 0.791,
+      "step": 5608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001816845745356476,
+      "loss": 0.9478,
+      "step": 5609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018167822774958441,
+      "loss": 1.0116,
+      "step": 5610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018167187997494236,
+      "loss": 0.78,
+      "step": 5611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018166553121179833,
+      "loss": 0.9903,
+      "step": 5612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018165918146022917,
+      "loss": 0.9654,
+      "step": 5613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018165283072031166,
+      "loss": 0.8953,
+      "step": 5614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018164647899212278,
+      "loss": 1.0039,
+      "step": 5615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018164012627573927,
+      "loss": 0.9747,
+      "step": 5616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018163377257123813,
+      "loss": 0.8614,
+      "step": 5617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001816274178786962,
+      "loss": 0.8371,
+      "step": 5618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018162106219819045,
+      "loss": 0.8866,
+      "step": 5619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018161470552979772,
+      "loss": 0.7844,
+      "step": 5620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018160834787359502,
+      "loss": 0.8127,
+      "step": 5621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018160198922965925,
+      "loss": 0.7473,
+      "step": 5622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018159562959806745,
+      "loss": 0.7972,
+      "step": 5623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001815892689788965,
+      "loss": 0.862,
+      "step": 5624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001815829073722234,
+      "loss": 0.7944,
+      "step": 5625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001815765447781252,
+      "loss": 0.9376,
+      "step": 5626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018157018119667885,
+      "loss": 0.936,
+      "step": 5627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018156381662796142,
+      "loss": 0.8902,
+      "step": 5628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001815574510720499,
+      "loss": 0.773,
+      "step": 5629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018155108452902138,
+      "loss": 0.7892,
+      "step": 5630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018154471699895287,
+      "loss": 0.9428,
+      "step": 5631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018153834848192148,
+      "loss": 0.8856,
+      "step": 5632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018153197897800423,
+      "loss": 0.9663,
+      "step": 5633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001815256084872783,
+      "loss": 0.8706,
+      "step": 5634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018151923700982075,
+      "loss": 0.8866,
+      "step": 5635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018151286454570864,
+      "loss": 0.8605,
+      "step": 5636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018150649109501918,
+      "loss": 0.8129,
+      "step": 5637
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001815001166578295,
+      "loss": 1.0007,
+      "step": 5638
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018149374123421668,
+      "loss": 0.874,
+      "step": 5639
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018148736482425797,
+      "loss": 0.8776,
+      "step": 5640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001814809874280305,
+      "loss": 0.9111,
+      "step": 5641
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018147460904561146,
+      "loss": 0.8561,
+      "step": 5642
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001814682296770781,
+      "loss": 0.8547,
+      "step": 5643
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018146184932250754,
+      "loss": 0.8553,
+      "step": 5644
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001814554679819771,
+      "loss": 0.8903,
+      "step": 5645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018144908565556394,
+      "loss": 0.8122,
+      "step": 5646
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018144270234334535,
+      "loss": 0.895,
+      "step": 5647
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018143631804539857,
+      "loss": 0.8782,
+      "step": 5648
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018142993276180086,
+      "loss": 0.8569,
+      "step": 5649
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018142354649262953,
+      "loss": 0.832,
+      "step": 5650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018141715923796192,
+      "loss": 0.8547,
+      "step": 5651
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018141077099787523,
+      "loss": 0.8016,
+      "step": 5652
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018140438177244683,
+      "loss": 0.7601,
+      "step": 5653
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001813979915617541,
+      "loss": 0.9216,
+      "step": 5654
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018139160036587427,
+      "loss": 0.8546,
+      "step": 5655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001813852081848848,
+      "loss": 0.9409,
+      "step": 5656
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000181378815018863,
+      "loss": 0.7942,
+      "step": 5657
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001813724208678863,
+      "loss": 0.7999,
+      "step": 5658
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018136602573203198,
+      "loss": 0.841,
+      "step": 5659
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001813596296113776,
+      "loss": 1.0006,
+      "step": 5660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018135323250600047,
+      "loss": 0.9056,
+      "step": 5661
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018134683441597803,
+      "loss": 0.9448,
+      "step": 5662
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018134043534138772,
+      "loss": 0.7664,
+      "step": 5663
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000181334035282307,
+      "loss": 0.931,
+      "step": 5664
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018132763423881336,
+      "loss": 0.8539,
+      "step": 5665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001813212322109842,
+      "loss": 0.9378,
+      "step": 5666
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001813148291988971,
+      "loss": 0.933,
+      "step": 5667
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001813084252026295,
+      "loss": 0.9238,
+      "step": 5668
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001813020202222589,
+      "loss": 0.8028,
+      "step": 5669
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018129561425786283,
+      "loss": 0.9385,
+      "step": 5670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018128920730951884,
+      "loss": 0.9579,
+      "step": 5671
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018128279937730445,
+      "loss": 0.877,
+      "step": 5672
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018127639046129726,
+      "loss": 0.8354,
+      "step": 5673
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018126998056157482,
+      "loss": 0.7487,
+      "step": 5674
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001812635696782147,
+      "loss": 0.9987,
+      "step": 5675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018125715781129448,
+      "loss": 0.875,
+      "step": 5676
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001812507449608918,
+      "loss": 0.8299,
+      "step": 5677
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018124433112708425,
+      "loss": 0.8475,
+      "step": 5678
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001812379163099495,
+      "loss": 0.8287,
+      "step": 5679
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018123150050956512,
+      "loss": 0.7749,
+      "step": 5680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018122508372600882,
+      "loss": 0.8801,
+      "step": 5681
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018121866595935827,
+      "loss": 0.879,
+      "step": 5682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001812122472096911,
+      "loss": 0.9046,
+      "step": 5683
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018120582747708502,
+      "loss": 0.8958,
+      "step": 5684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001811994067616177,
+      "loss": 0.9232,
+      "step": 5685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018119298506336695,
+      "loss": 0.8041,
+      "step": 5686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001811865623824104,
+      "loss": 0.8678,
+      "step": 5687
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001811801387188258,
+      "loss": 0.9042,
+      "step": 5688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018117371407269092,
+      "loss": 0.7886,
+      "step": 5689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018116728844408351,
+      "loss": 1.0048,
+      "step": 5690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018116086183308133,
+      "loss": 0.9407,
+      "step": 5691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018115443423976222,
+      "loss": 0.8704,
+      "step": 5692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018114800566420392,
+      "loss": 0.9065,
+      "step": 5693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018114157610648422,
+      "loss": 0.9413,
+      "step": 5694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000181135145566681,
+      "loss": 0.886,
+      "step": 5695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018112871404487202,
+      "loss": 0.8162,
+      "step": 5696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018112228154113516,
+      "loss": 0.8913,
+      "step": 5697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018111584805554834,
+      "loss": 1.0159,
+      "step": 5698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001811094135881893,
+      "loss": 0.8192,
+      "step": 5699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000181102978139136,
+      "loss": 0.8689,
+      "step": 5700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001810965417084663,
+      "loss": 0.9976,
+      "step": 5701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018109010429625814,
+      "loss": 0.9037,
+      "step": 5702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001810836659025894,
+      "loss": 0.8935,
+      "step": 5703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000181077226527538,
+      "loss": 0.9209,
+      "step": 5704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018107078617118192,
+      "loss": 0.9199,
+      "step": 5705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018106434483359905,
+      "loss": 0.9451,
+      "step": 5706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001810579025148674,
+      "loss": 0.8549,
+      "step": 5707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018105145921506493,
+      "loss": 0.9369,
+      "step": 5708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001810450149342696,
+      "loss": 0.8135,
+      "step": 5709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018103856967255946,
+      "loss": 0.8748,
+      "step": 5710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001810321234300125,
+      "loss": 0.8342,
+      "step": 5711
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018102567620670667,
+      "loss": 0.8916,
+      "step": 5712
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001810192280027201,
+      "loss": 1.0143,
+      "step": 5713
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001810127788181308,
+      "loss": 0.758,
+      "step": 5714
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001810063286530168,
+      "loss": 0.7993,
+      "step": 5715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001809998775074562,
+      "loss": 0.8589,
+      "step": 5716
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018099342538152713,
+      "loss": 0.8219,
+      "step": 5717
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018098697227530755,
+      "loss": 0.7828,
+      "step": 5718
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001809805181888757,
+      "loss": 0.897,
+      "step": 5719
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001809740631223096,
+      "loss": 0.9111,
+      "step": 5720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018096760707568744,
+      "loss": 0.8776,
+      "step": 5721
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001809611500490873,
+      "loss": 0.7717,
+      "step": 5722
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018095469204258737,
+      "loss": 0.8534,
+      "step": 5723
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018094823305626583,
+      "loss": 0.8452,
+      "step": 5724
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018094177309020083,
+      "loss": 0.8393,
+      "step": 5725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018093531214447052,
+      "loss": 0.9796,
+      "step": 5726
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018092885021915315,
+      "loss": 0.7358,
+      "step": 5727
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018092238731432696,
+      "loss": 0.906,
+      "step": 5728
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018091592343007012,
+      "loss": 1.0549,
+      "step": 5729
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018090945856646083,
+      "loss": 0.8178,
+      "step": 5730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018090299272357745,
+      "loss": 0.9351,
+      "step": 5731
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018089652590149813,
+      "loss": 0.8646,
+      "step": 5732
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001808900581003012,
+      "loss": 0.9185,
+      "step": 5733
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018088358932006493,
+      "loss": 0.9723,
+      "step": 5734
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001808771195608676,
+      "loss": 0.8585,
+      "step": 5735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001808706488227875,
+      "loss": 0.8899,
+      "step": 5736
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018086417710590303,
+      "loss": 0.9007,
+      "step": 5737
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001808577044102924,
+      "loss": 0.7464,
+      "step": 5738
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018085123073603403,
+      "loss": 0.9993,
+      "step": 5739
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018084475608320626,
+      "loss": 0.8992,
+      "step": 5740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018083828045188747,
+      "loss": 0.983,
+      "step": 5741
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018083180384215596,
+      "loss": 0.7752,
+      "step": 5742
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018082532625409022,
+      "loss": 0.8704,
+      "step": 5743
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001808188476877686,
+      "loss": 0.8522,
+      "step": 5744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001808123681432695,
+      "loss": 0.8586,
+      "step": 5745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018080588762067138,
+      "loss": 0.8671,
+      "step": 5746
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018079940612005264,
+      "loss": 0.9641,
+      "step": 5747
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018079292364149175,
+      "loss": 0.8975,
+      "step": 5748
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001807864401850672,
+      "loss": 0.848,
+      "step": 5749
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001807799557508574,
+      "loss": 0.9571,
+      "step": 5750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018077347033894082,
+      "loss": 0.9171,
+      "step": 5751
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018076698394939604,
+      "loss": 0.8614,
+      "step": 5752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001807604965823015,
+      "loss": 0.9383,
+      "step": 5753
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018075400823773578,
+      "loss": 0.9193,
+      "step": 5754
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018074751891577733,
+      "loss": 1.0384,
+      "step": 5755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018074102861650475,
+      "loss": 0.722,
+      "step": 5756
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001807345373399966,
+      "loss": 0.891,
+      "step": 5757
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018072804508633138,
+      "loss": 0.9097,
+      "step": 5758
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001807215518555877,
+      "loss": 0.8858,
+      "step": 5759
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001807150576478442,
+      "loss": 0.906,
+      "step": 5760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018070856246317943,
+      "loss": 1.0141,
+      "step": 5761
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018070206630167203,
+      "loss": 0.8753,
+      "step": 5762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001806955691634006,
+      "loss": 0.835,
+      "step": 5763
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001806890710484438,
+      "loss": 0.9788,
+      "step": 5764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018068257195688026,
+      "loss": 0.8399,
+      "step": 5765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018067607188878863,
+      "loss": 0.9498,
+      "step": 5766
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018066957084424763,
+      "loss": 0.8218,
+      "step": 5767
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001806630688233359,
+      "loss": 0.9156,
+      "step": 5768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018065656582613216,
+      "loss": 0.9482,
+      "step": 5769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018065006185271506,
+      "loss": 0.8629,
+      "step": 5770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018064355690316345,
+      "loss": 1.0716,
+      "step": 5771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018063705097755593,
+      "loss": 0.843,
+      "step": 5772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001806305440759713,
+      "loss": 0.8487,
+      "step": 5773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001806240361984883,
+      "loss": 0.9182,
+      "step": 5774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018061752734518574,
+      "loss": 0.9442,
+      "step": 5775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001806110175161423,
+      "loss": 0.9108,
+      "step": 5776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001806045067114369,
+      "loss": 0.7547,
+      "step": 5777
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018059799493114827,
+      "loss": 0.9648,
+      "step": 5778
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001805914821753552,
+      "loss": 0.837,
+      "step": 5779
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018058496844413656,
+      "loss": 0.9964,
+      "step": 5780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018057845373757117,
+      "loss": 0.8578,
+      "step": 5781
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001805719380557379,
+      "loss": 0.9415,
+      "step": 5782
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001805654213987156,
+      "loss": 0.8792,
+      "step": 5783
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018055890376658313,
+      "loss": 0.8345,
+      "step": 5784
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001805523851594194,
+      "loss": 1.0178,
+      "step": 5785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018054586557730327,
+      "loss": 0.917,
+      "step": 5786
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018053934502031368,
+      "loss": 0.8679,
+      "step": 5787
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018053282348852955,
+      "loss": 0.9329,
+      "step": 5788
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001805263009820298,
+      "loss": 0.8404,
+      "step": 5789
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018051977750089336,
+      "loss": 0.9729,
+      "step": 5790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018051325304519923,
+      "loss": 0.811,
+      "step": 5791
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018050672761502635,
+      "loss": 0.8547,
+      "step": 5792
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001805002012104537,
+      "loss": 0.9225,
+      "step": 5793
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018049367383156026,
+      "loss": 0.9105,
+      "step": 5794
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018048714547842502,
+      "loss": 0.8136,
+      "step": 5795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001804806161511271,
+      "loss": 0.9792,
+      "step": 5796
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018047408584974536,
+      "loss": 0.9089,
+      "step": 5797
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018046755457435895,
+      "loss": 1.0294,
+      "step": 5798
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001804610223250469,
+      "loss": 0.9273,
+      "step": 5799
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018045448910188828,
+      "loss": 0.7214,
+      "step": 5800
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 0.8726089596748352,
+      "eval_runtime": 848.972,
+      "eval_samples_per_second": 28.34,
+      "eval_steps_per_second": 14.17,
+      "step": 5800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018044795490496213,
+      "loss": 0.8847,
+      "step": 5801
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018044141973434758,
+      "loss": 0.8943,
+      "step": 5802
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001804348835901237,
+      "loss": 0.8297,
+      "step": 5803
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001804283464723696,
+      "loss": 0.7658,
+      "step": 5804
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001804218083811644,
+      "loss": 0.839,
+      "step": 5805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018041526931658722,
+      "loss": 0.9348,
+      "step": 5806
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018040872927871723,
+      "loss": 0.8186,
+      "step": 5807
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018040218826763358,
+      "loss": 1.0253,
+      "step": 5808
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018039564628341545,
+      "loss": 0.8271,
+      "step": 5809
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018038910332614198,
+      "loss": 0.9569,
+      "step": 5810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001803825593958924,
+      "loss": 0.9408,
+      "step": 5811
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001803760144927459,
+      "loss": 0.7387,
+      "step": 5812
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001803694686167817,
+      "loss": 0.8481,
+      "step": 5813
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018036292176807904,
+      "loss": 0.9052,
+      "step": 5814
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001803563739467171,
+      "loss": 0.8271,
+      "step": 5815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001803498251527752,
+      "loss": 0.8052,
+      "step": 5816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018034327538633254,
+      "loss": 0.8411,
+      "step": 5817
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018033672464746846,
+      "loss": 1.0407,
+      "step": 5818
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001803301729362622,
+      "loss": 0.8075,
+      "step": 5819
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018032362025279308,
+      "loss": 0.8919,
+      "step": 5820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018031706659714042,
+      "loss": 0.9337,
+      "step": 5821
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018031051196938346,
+      "loss": 1.011,
+      "step": 5822
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018030395636960163,
+      "loss": 0.8032,
+      "step": 5823
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018029739979787425,
+      "loss": 0.8447,
+      "step": 5824
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018029084225428067,
+      "loss": 0.8227,
+      "step": 5825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018028428373890022,
+      "loss": 0.7948,
+      "step": 5826
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018027772425181234,
+      "loss": 0.7664,
+      "step": 5827
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018027116379309638,
+      "loss": 0.7733,
+      "step": 5828
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018026460236283175,
+      "loss": 0.8053,
+      "step": 5829
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001802580399610979,
+      "loss": 0.8448,
+      "step": 5830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018025147658797422,
+      "loss": 0.8433,
+      "step": 5831
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018024491224354015,
+      "loss": 0.7909,
+      "step": 5832
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018023834692787515,
+      "loss": 0.8667,
+      "step": 5833
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018023178064105867,
+      "loss": 0.8668,
+      "step": 5834
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018022521338317024,
+      "loss": 0.9581,
+      "step": 5835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001802186451542893,
+      "loss": 0.7939,
+      "step": 5836
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018021207595449529,
+      "loss": 0.8727,
+      "step": 5837
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018020550578386782,
+      "loss": 0.9151,
+      "step": 5838
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018019893464248633,
+      "loss": 0.8099,
+      "step": 5839
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018019236253043042,
+      "loss": 0.7932,
+      "step": 5840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018018578944777963,
+      "loss": 0.8753,
+      "step": 5841
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018017921539461348,
+      "loss": 0.9023,
+      "step": 5842
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018017264037101154,
+      "loss": 0.7624,
+      "step": 5843
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018016606437705338,
+      "loss": 1.0044,
+      "step": 5844
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018015948741281865,
+      "loss": 0.8579,
+      "step": 5845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001801529094783869,
+      "loss": 0.7879,
+      "step": 5846
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018014633057383773,
+      "loss": 0.9312,
+      "step": 5847
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018013975069925084,
+      "loss": 0.9521,
+      "step": 5848
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001801331698547058,
+      "loss": 0.9011,
+      "step": 5849
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001801265880402823,
+      "loss": 0.9471,
+      "step": 5850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018012000525605997,
+      "loss": 0.9244,
+      "step": 5851
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018011342150211853,
+      "loss": 0.9794,
+      "step": 5852
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001801068367785376,
+      "loss": 0.9799,
+      "step": 5853
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018010025108539695,
+      "loss": 0.8919,
+      "step": 5854
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001800936644227762,
+      "loss": 0.832,
+      "step": 5855
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018008707679075515,
+      "loss": 0.8799,
+      "step": 5856
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018008048818941353,
+      "loss": 0.9162,
+      "step": 5857
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000180073898618831,
+      "loss": 0.8358,
+      "step": 5858
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018006730807908742,
+      "loss": 0.8419,
+      "step": 5859
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001800607165702625,
+      "loss": 0.8715,
+      "step": 5860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018005412409243606,
+      "loss": 0.7652,
+      "step": 5861
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001800475306456878,
+      "loss": 0.7997,
+      "step": 5862
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018004093623009764,
+      "loss": 0.8489,
+      "step": 5863
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018003434084574532,
+      "loss": 0.7937,
+      "step": 5864
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001800277444927107,
+      "loss": 0.966,
+      "step": 5865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001800211471710736,
+      "loss": 0.7706,
+      "step": 5866
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018001454888091383,
+      "loss": 1.0362,
+      "step": 5867
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018000794962231133,
+      "loss": 0.8764,
+      "step": 5868
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018000134939534593,
+      "loss": 0.8939,
+      "step": 5869
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017999474820009752,
+      "loss": 0.8748,
+      "step": 5870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000179988146036646,
+      "loss": 0.8807,
+      "step": 5871
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017998154290507128,
+      "loss": 0.7593,
+      "step": 5872
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017997493880545328,
+      "loss": 1.1118,
+      "step": 5873
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017996833373787191,
+      "loss": 0.8465,
+      "step": 5874
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017996172770240714,
+      "loss": 0.8063,
+      "step": 5875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017995512069913892,
+      "loss": 0.8148,
+      "step": 5876
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017994851272814724,
+      "loss": 0.9314,
+      "step": 5877
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017994190378951198,
+      "loss": 0.7953,
+      "step": 5878
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017993529388331326,
+      "loss": 0.935,
+      "step": 5879
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017992868300963102,
+      "loss": 0.76,
+      "step": 5880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017992207116854526,
+      "loss": 0.8275,
+      "step": 5881
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017991545836013603,
+      "loss": 0.9058,
+      "step": 5882
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017990884458448335,
+      "loss": 0.8412,
+      "step": 5883
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001799022298416673,
+      "loss": 0.933,
+      "step": 5884
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017989561413176789,
+      "loss": 0.9466,
+      "step": 5885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017988899745486524,
+      "loss": 0.8743,
+      "step": 5886
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017988237981103944,
+      "loss": 0.8937,
+      "step": 5887
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017987576120037053,
+      "loss": 0.8388,
+      "step": 5888
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017986914162293865,
+      "loss": 0.921,
+      "step": 5889
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017986252107882392,
+      "loss": 0.7911,
+      "step": 5890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017985589956810648,
+      "loss": 0.9521,
+      "step": 5891
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001798492770908664,
+      "loss": 0.8143,
+      "step": 5892
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017984265364718395,
+      "loss": 0.9195,
+      "step": 5893
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017983602923713922,
+      "loss": 0.8154,
+      "step": 5894
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017982940386081242,
+      "loss": 0.7816,
+      "step": 5895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017982277751828373,
+      "loss": 0.8646,
+      "step": 5896
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017981615020963335,
+      "loss": 0.792,
+      "step": 5897
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017980952193494146,
+      "loss": 0.9923,
+      "step": 5898
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017980289269428835,
+      "loss": 0.8939,
+      "step": 5899
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017979626248775419,
+      "loss": 0.8267,
+      "step": 5900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017978963131541926,
+      "loss": 0.9352,
+      "step": 5901
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001797829991773638,
+      "loss": 0.8708,
+      "step": 5902
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017977636607366813,
+      "loss": 0.8438,
+      "step": 5903
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001797697320044125,
+      "loss": 0.9006,
+      "step": 5904
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017976309696967718,
+      "loss": 1.0418,
+      "step": 5905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017975646096954253,
+      "loss": 1.0111,
+      "step": 5906
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017974982400408883,
+      "loss": 0.9413,
+      "step": 5907
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017974318607339639,
+      "loss": 0.7846,
+      "step": 5908
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001797365471775456,
+      "loss": 0.9508,
+      "step": 5909
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017972990731661678,
+      "loss": 0.8943,
+      "step": 5910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017972326649069031,
+      "loss": 0.8659,
+      "step": 5911
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017971662469984658,
+      "loss": 0.9606,
+      "step": 5912
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017970998194416596,
+      "loss": 0.8509,
+      "step": 5913
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017970333822372883,
+      "loss": 0.9884,
+      "step": 5914
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001796966935386156,
+      "loss": 0.8447,
+      "step": 5915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017969004788890674,
+      "loss": 0.9098,
+      "step": 5916
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017968340127468268,
+      "loss": 0.8506,
+      "step": 5917
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001796767536960238,
+      "loss": 0.9689,
+      "step": 5918
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017967010515301064,
+      "loss": 0.9706,
+      "step": 5919
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001796634556457236,
+      "loss": 0.8054,
+      "step": 5920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017965680517424324,
+      "loss": 0.7993,
+      "step": 5921
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017965015373864996,
+      "loss": 0.8878,
+      "step": 5922
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017964350133902433,
+      "loss": 0.7002,
+      "step": 5923
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017963684797544683,
+      "loss": 0.7341,
+      "step": 5924
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017963019364799801,
+      "loss": 0.9383,
+      "step": 5925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017962353835675844,
+      "loss": 0.9425,
+      "step": 5926
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017961688210180857,
+      "loss": 0.8517,
+      "step": 5927
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017961022488322906,
+      "loss": 0.8506,
+      "step": 5928
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017960356670110046,
+      "loss": 0.8702,
+      "step": 5929
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017959690755550334,
+      "loss": 0.9333,
+      "step": 5930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001795902474465183,
+      "loss": 0.7998,
+      "step": 5931
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017958358637422595,
+      "loss": 0.7286,
+      "step": 5932
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017957692433870695,
+      "loss": 0.9091,
+      "step": 5933
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017957026134004186,
+      "loss": 0.9253,
+      "step": 5934
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017956359737831138,
+      "loss": 0.9644,
+      "step": 5935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017955693245359617,
+      "loss": 0.9281,
+      "step": 5936
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017955026656597687,
+      "loss": 0.9025,
+      "step": 5937
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017954359971553415,
+      "loss": 0.8244,
+      "step": 5938
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017953693190234872,
+      "loss": 0.876,
+      "step": 5939
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017953026312650128,
+      "loss": 0.8361,
+      "step": 5940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017952359338807257,
+      "loss": 0.8675,
+      "step": 5941
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017951692268714328,
+      "loss": 0.9097,
+      "step": 5942
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017951025102379414,
+      "loss": 0.9399,
+      "step": 5943
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017950357839810593,
+      "loss": 0.837,
+      "step": 5944
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017949690481015943,
+      "loss": 0.8929,
+      "step": 5945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017949023026003538,
+      "loss": 0.8709,
+      "step": 5946
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001794835547478145,
+      "loss": 0.8134,
+      "step": 5947
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001794768782735777,
+      "loss": 0.8505,
+      "step": 5948
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017947020083740575,
+      "loss": 0.8518,
+      "step": 5949
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017946352243937948,
+      "loss": 0.9469,
+      "step": 5950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017945684307957966,
+      "loss": 0.8294,
+      "step": 5951
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017945016275808717,
+      "loss": 0.9454,
+      "step": 5952
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001794434814749829,
+      "loss": 0.9073,
+      "step": 5953
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017943679923034767,
+      "loss": 1.017,
+      "step": 5954
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017943011602426236,
+      "loss": 0.9292,
+      "step": 5955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017942343185680786,
+      "loss": 0.9336,
+      "step": 5956
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001794167467280651,
+      "loss": 0.8034,
+      "step": 5957
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017941006063811496,
+      "loss": 0.9044,
+      "step": 5958
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001794033735870384,
+      "loss": 0.9657,
+      "step": 5959
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017939668557491632,
+      "loss": 0.9293,
+      "step": 5960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017938999660182967,
+      "loss": 0.8057,
+      "step": 5961
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017938330666785943,
+      "loss": 0.7011,
+      "step": 5962
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017937661577308654,
+      "loss": 0.917,
+      "step": 5963
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017936992391759203,
+      "loss": 1.0092,
+      "step": 5964
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017936323110145686,
+      "loss": 0.8167,
+      "step": 5965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017935653732476206,
+      "loss": 0.8977,
+      "step": 5966
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017934984258758858,
+      "loss": 0.9223,
+      "step": 5967
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017934314689001753,
+      "loss": 0.8256,
+      "step": 5968
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001793364502321299,
+      "loss": 0.7656,
+      "step": 5969
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017932975261400675,
+      "loss": 0.8457,
+      "step": 5970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017932305403572916,
+      "loss": 0.8514,
+      "step": 5971
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001793163544973782,
+      "loss": 0.8626,
+      "step": 5972
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017930965399903497,
+      "loss": 0.8554,
+      "step": 5973
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017930295254078052,
+      "loss": 0.8352,
+      "step": 5974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000179296250122696,
+      "loss": 0.8623,
+      "step": 5975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017928954674486253,
+      "loss": 0.9092,
+      "step": 5976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017928284240736124,
+      "loss": 0.9246,
+      "step": 5977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017927613711027327,
+      "loss": 0.9481,
+      "step": 5978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017926943085367974,
+      "loss": 0.8145,
+      "step": 5979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001792627236376619,
+      "loss": 0.9042,
+      "step": 5980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017925601546230088,
+      "loss": 0.9146,
+      "step": 5981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017924930632767788,
+      "loss": 0.8635,
+      "step": 5982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001792425962338741,
+      "loss": 0.9056,
+      "step": 5983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017923588518097073,
+      "loss": 0.9457,
+      "step": 5984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017922917316904905,
+      "loss": 0.7599,
+      "step": 5985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017922246019819028,
+      "loss": 0.8563,
+      "step": 5986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001792157462684756,
+      "loss": 0.7591,
+      "step": 5987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017920903137998636,
+      "loss": 0.8031,
+      "step": 5988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017920231553280382,
+      "loss": 0.9576,
+      "step": 5989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001791955987270092,
+      "loss": 0.748,
+      "step": 5990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017918888096268393,
+      "loss": 1.0058,
+      "step": 5991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017918216223990912,
+      "loss": 0.9288,
+      "step": 5992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017917544255876625,
+      "loss": 0.7554,
+      "step": 5993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001791687219193366,
+      "loss": 0.7388,
+      "step": 5994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001791620003217015,
+      "loss": 0.8082,
+      "step": 5995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017915527776594232,
+      "loss": 0.8209,
+      "step": 5996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001791485542521404,
+      "loss": 0.7383,
+      "step": 5997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017914182978037718,
+      "loss": 0.9779,
+      "step": 5998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017913510435073397,
+      "loss": 0.8999,
+      "step": 5999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017912837796329217,
+      "loss": 0.8789,
+      "step": 6000
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 0.8621419072151184,
+      "eval_runtime": 852.8418,
+      "eval_samples_per_second": 28.212,
+      "eval_steps_per_second": 14.106,
+      "step": 6000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017912165061813328,
+      "loss": 0.8336,
+      "step": 6001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017911492231533868,
+      "loss": 0.8948,
+      "step": 6002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017910819305498975,
+      "loss": 0.867,
+      "step": 6003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000179101462837168,
+      "loss": 0.8485,
+      "step": 6004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001790947316619549,
+      "loss": 0.8979,
+      "step": 6005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017908799952943185,
+      "loss": 0.8826,
+      "step": 6006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017908126643968037,
+      "loss": 0.9747,
+      "step": 6007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017907453239278195,
+      "loss": 0.9112,
+      "step": 6008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001790677973888181,
+      "loss": 0.876,
+      "step": 6009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017906106142787036,
+      "loss": 0.7474,
+      "step": 6010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001790543245100202,
+      "loss": 0.8556,
+      "step": 6011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001790475866353492,
+      "loss": 0.7963,
+      "step": 6012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001790408478039389,
+      "loss": 0.9439,
+      "step": 6013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017903410801587086,
+      "loss": 0.8595,
+      "step": 6014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017902736727122667,
+      "loss": 0.9642,
+      "step": 6015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017902062557008787,
+      "loss": 0.7173,
+      "step": 6016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001790138829125361,
+      "loss": 1.036,
+      "step": 6017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017900713929865298,
+      "loss": 1.0617,
+      "step": 6018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001790003947285201,
+      "loss": 0.9068,
+      "step": 6019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001789936492022191,
+      "loss": 0.8233,
+      "step": 6020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017898690271983162,
+      "loss": 0.8566,
+      "step": 6021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017898015528143932,
+      "loss": 0.8795,
+      "step": 6022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017897340688712387,
+      "loss": 0.8071,
+      "step": 6023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017896665753696693,
+      "loss": 0.8582,
+      "step": 6024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017895990723105023,
+      "loss": 0.8226,
+      "step": 6025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001789531559694554,
+      "loss": 0.862,
+      "step": 6026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017894640375226424,
+      "loss": 0.953,
+      "step": 6027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001789396505795584,
+      "loss": 0.8714,
+      "step": 6028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017893289645141966,
+      "loss": 0.8438,
+      "step": 6029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017892614136792975,
+      "loss": 0.7766,
+      "step": 6030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017891938532917044,
+      "loss": 0.8856,
+      "step": 6031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001789126283352235,
+      "loss": 0.8231,
+      "step": 6032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017890587038617067,
+      "loss": 0.7567,
+      "step": 6033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017889911148209382,
+      "loss": 0.8275,
+      "step": 6034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001788923516230747,
+      "loss": 0.8716,
+      "step": 6035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017888559080919512,
+      "loss": 0.81,
+      "step": 6036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017887882904053698,
+      "loss": 0.9748,
+      "step": 6037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017887206631718203,
+      "loss": 0.8232,
+      "step": 6038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017886530263921214,
+      "loss": 0.8716,
+      "step": 6039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017885853800670925,
+      "loss": 0.7995,
+      "step": 6040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017885177241975514,
+      "loss": 0.8709,
+      "step": 6041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017884500587843176,
+      "loss": 0.8898,
+      "step": 6042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017883823838282095,
+      "loss": 0.8057,
+      "step": 6043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001788314699330047,
+      "loss": 0.9869,
+      "step": 6044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017882470052906486,
+      "loss": 0.8059,
+      "step": 6045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017881793017108337,
+      "loss": 0.9689,
+      "step": 6046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017881115885914222,
+      "loss": 0.9222,
+      "step": 6047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017880438659332332,
+      "loss": 0.9298,
+      "step": 6048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017879761337370864,
+      "loss": 0.8782,
+      "step": 6049
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017879083920038018,
+      "loss": 0.8683,
+      "step": 6050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001787840640734199,
+      "loss": 0.8407,
+      "step": 6051
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017877728799290985,
+      "loss": 0.8239,
+      "step": 6052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017877051095893204,
+      "loss": 0.7933,
+      "step": 6053
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017876373297156842,
+      "loss": 0.8532,
+      "step": 6054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001787569540309011,
+      "loss": 0.9691,
+      "step": 6055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017875017413701208,
+      "loss": 0.9708,
+      "step": 6056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017874339328998346,
+      "loss": 0.8451,
+      "step": 6057
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017873661148989732,
+      "loss": 0.8492,
+      "step": 6058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017872982873683567,
+      "loss": 0.8188,
+      "step": 6059
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017872304503088068,
+      "loss": 0.8748,
+      "step": 6060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017871626037211443,
+      "loss": 0.8245,
+      "step": 6061
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017870947476061901,
+      "loss": 0.8711,
+      "step": 6062
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017870268819647657,
+      "loss": 0.904,
+      "step": 6063
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017869590067976927,
+      "loss": 0.8792,
+      "step": 6064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017868911221057926,
+      "loss": 0.9017,
+      "step": 6065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017868232278898862,
+      "loss": 0.9607,
+      "step": 6066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017867553241507963,
+      "loss": 0.9237,
+      "step": 6067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017866874108893446,
+      "loss": 0.802,
+      "step": 6068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017866194881063525,
+      "loss": 0.7949,
+      "step": 6069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017865515558026428,
+      "loss": 0.8022,
+      "step": 6070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001786483613979037,
+      "loss": 0.9654,
+      "step": 6071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017864156626363578,
+      "loss": 0.7766,
+      "step": 6072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017863477017754278,
+      "loss": 0.8157,
+      "step": 6073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017862797313970691,
+      "loss": 0.9555,
+      "step": 6074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017862117515021048,
+      "loss": 0.9344,
+      "step": 6075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017861437620913574,
+      "loss": 0.8898,
+      "step": 6076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000178607576316565,
+      "loss": 0.871,
+      "step": 6077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017860077547258055,
+      "loss": 0.8347,
+      "step": 6078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001785939736772647,
+      "loss": 0.8482,
+      "step": 6079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001785871709306998,
+      "loss": 0.9283,
+      "step": 6080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017858036723296814,
+      "loss": 0.8609,
+      "step": 6081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017857356258415213,
+      "loss": 0.8809,
+      "step": 6082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017856675698433406,
+      "loss": 0.8991,
+      "step": 6083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017855995043359634,
+      "loss": 0.7443,
+      "step": 6084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017855314293202133,
+      "loss": 0.8625,
+      "step": 6085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017854633447969142,
+      "loss": 0.9104,
+      "step": 6086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017853952507668907,
+      "loss": 0.8869,
+      "step": 6087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017853271472309667,
+      "loss": 0.8356,
+      "step": 6088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001785259034189966,
+      "loss": 0.9582,
+      "step": 6089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017851909116447137,
+      "loss": 0.7769,
+      "step": 6090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017851227795960336,
+      "loss": 0.8349,
+      "step": 6091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001785054638044751,
+      "loss": 0.9252,
+      "step": 6092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017849864869916903,
+      "loss": 0.8018,
+      "step": 6093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017849183264376762,
+      "loss": 0.8535,
+      "step": 6094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017848501563835341,
+      "loss": 0.871,
+      "step": 6095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017847819768300889,
+      "loss": 0.8602,
+      "step": 6096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017847137877781655,
+      "loss": 0.7925,
+      "step": 6097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017846455892285895,
+      "loss": 0.8321,
+      "step": 6098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017845773811821865,
+      "loss": 0.7976,
+      "step": 6099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017845091636397816,
+      "loss": 0.9619,
+      "step": 6100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017844409366022008,
+      "loss": 0.8768,
+      "step": 6101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000178437270007027,
+      "loss": 0.8387,
+      "step": 6102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017843044540448143,
+      "loss": 0.8813,
+      "step": 6103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017842361985266608,
+      "loss": 0.7451,
+      "step": 6104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017841679335166348,
+      "loss": 0.8198,
+      "step": 6105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001784099659015563,
+      "loss": 0.8501,
+      "step": 6106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017840313750242713,
+      "loss": 0.8402,
+      "step": 6107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017839630815435868,
+      "loss": 0.9306,
+      "step": 6108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017838947785743354,
+      "loss": 0.9595,
+      "step": 6109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017838264661173444,
+      "loss": 0.8905,
+      "step": 6110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017837581441734401,
+      "loss": 0.8427,
+      "step": 6111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017836898127434497,
+      "loss": 0.8692,
+      "step": 6112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017836214718282,
+      "loss": 0.9423,
+      "step": 6113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017835531214285187,
+      "loss": 0.8992,
+      "step": 6114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017834847615452325,
+      "loss": 0.7407,
+      "step": 6115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017834163921791687,
+      "loss": 0.9648,
+      "step": 6116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017833480133311555,
+      "loss": 0.8618,
+      "step": 6117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017832796250020198,
+      "loss": 0.97,
+      "step": 6118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000178321122719259,
+      "loss": 0.9726,
+      "step": 6119
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017831428199036932,
+      "loss": 0.8618,
+      "step": 6120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017830744031361578,
+      "loss": 0.8343,
+      "step": 6121
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017830059768908117,
+      "loss": 0.9253,
+      "step": 6122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001782937541168483,
+      "loss": 0.8534,
+      "step": 6123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017828690959700006,
+      "loss": 0.9572,
+      "step": 6124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001782800641296192,
+      "loss": 0.8911,
+      "step": 6125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001782732177147887,
+      "loss": 0.8575,
+      "step": 6126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017826637035259126,
+      "loss": 0.8732,
+      "step": 6127
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001782595220431099,
+      "loss": 0.762,
+      "step": 6128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001782526727864274,
+      "loss": 0.9598,
+      "step": 6129
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017824582258262677,
+      "loss": 0.9585,
+      "step": 6130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017823897143179083,
+      "loss": 0.7021,
+      "step": 6131
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001782321193340025,
+      "loss": 0.7914,
+      "step": 6132
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001782252662893448,
+      "loss": 0.9369,
+      "step": 6133
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017821841229790058,
+      "loss": 0.8862,
+      "step": 6134
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017821155735975284,
+      "loss": 0.9771,
+      "step": 6135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017820470147498455,
+      "loss": 1.0174,
+      "step": 6136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017819784464367862,
+      "loss": 0.8816,
+      "step": 6137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017819098686591817,
+      "loss": 0.794,
+      "step": 6138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017818412814178612,
+      "loss": 0.7428,
+      "step": 6139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017817726847136549,
+      "loss": 0.8298,
+      "step": 6140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001781704078547393,
+      "loss": 0.8562,
+      "step": 6141
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017816354629199063,
+      "loss": 0.8934,
+      "step": 6142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00017815668378320248,
+      "loss": 0.8625,
+      "step": 6143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017814982032845792,
+      "loss": 0.8721,
+      "step": 6144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017814295592784003,
+      "loss": 0.8091,
+      "step": 6145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017813609058143189,
+      "loss": 0.7895,
+      "step": 6146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017812922428931657,
+      "loss": 0.8249,
+      "step": 6147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017812235705157723,
+      "loss": 0.8169,
+      "step": 6148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017811548886829693,
+      "loss": 0.9071,
+      "step": 6149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001781086197395588,
+      "loss": 0.8947,
+      "step": 6150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017810174966544605,
+      "loss": 0.774,
+      "step": 6151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017809487864604178,
+      "loss": 0.8124,
+      "step": 6152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001780880066814291,
+      "loss": 0.8405,
+      "step": 6153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017808113377169128,
+      "loss": 0.9093,
+      "step": 6154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017807425991691144,
+      "loss": 0.8109,
+      "step": 6155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001780673851171728,
+      "loss": 0.8039,
+      "step": 6156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001780605093725586,
+      "loss": 0.937,
+      "step": 6157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017805363268315197,
+      "loss": 0.8209,
+      "step": 6158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017804675504903624,
+      "loss": 0.8879,
+      "step": 6159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001780398764702946,
+      "loss": 0.7669,
+      "step": 6160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001780329969470103,
+      "loss": 0.8375,
+      "step": 6161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017802611647926663,
+      "loss": 0.9026,
+      "step": 6162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017801923506714684,
+      "loss": 0.845,
+      "step": 6163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017801235271073425,
+      "loss": 0.8869,
+      "step": 6164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017800546941011214,
+      "loss": 0.9292,
+      "step": 6165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017799858516536382,
+      "loss": 0.9276,
+      "step": 6166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017799169997657262,
+      "loss": 0.9855,
+      "step": 6167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017798481384382185,
+      "loss": 0.9476,
+      "step": 6168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017797792676719487,
+      "loss": 0.8415,
+      "step": 6169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017797103874677504,
+      "loss": 0.8865,
+      "step": 6170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017796414978264578,
+      "loss": 0.9281,
+      "step": 6171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017795725987489037,
+      "loss": 0.9527,
+      "step": 6172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017795036902359227,
+      "loss": 0.9307,
+      "step": 6173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017794347722883483,
+      "loss": 0.8352,
+      "step": 6174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017793658449070154,
+      "loss": 0.8816,
+      "step": 6175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017792969080927576,
+      "loss": 0.8464,
+      "step": 6176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017792279618464093,
+      "loss": 0.7881,
+      "step": 6177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017791590061688053,
+      "loss": 0.867,
+      "step": 6178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000177909004106078,
+      "loss": 0.8148,
+      "step": 6179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017790210665231683,
+      "loss": 0.7683,
+      "step": 6180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017789520825568046,
+      "loss": 0.8585,
+      "step": 6181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001778883089162524,
+      "loss": 0.889,
+      "step": 6182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001778814086341162,
+      "loss": 0.8375,
+      "step": 6183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017787450740935535,
+      "loss": 0.842,
+      "step": 6184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017786760524205334,
+      "loss": 0.925,
+      "step": 6185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017786070213229375,
+      "loss": 0.8114,
+      "step": 6186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017785379808016012,
+      "loss": 0.7007,
+      "step": 6187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000177846893085736,
+      "loss": 0.8434,
+      "step": 6188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000177839987149105,
+      "loss": 0.8595,
+      "step": 6189
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017783308027035066,
+      "loss": 0.7491,
+      "step": 6190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001778261724495566,
+      "loss": 0.7675,
+      "step": 6191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017781926368680643,
+      "loss": 0.7962,
+      "step": 6192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017781235398218375,
+      "loss": 0.8382,
+      "step": 6193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017780544333577224,
+      "loss": 0.9196,
+      "step": 6194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017779853174765546,
+      "loss": 0.7788,
+      "step": 6195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017779161921791713,
+      "loss": 0.8073,
+      "step": 6196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017778470574664085,
+      "loss": 0.9261,
+      "step": 6197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001777777913339104,
+      "loss": 0.8102,
+      "step": 6198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017777087597980936,
+      "loss": 0.7572,
+      "step": 6199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017776395968442153,
+      "loss": 0.8095,
+      "step": 6200
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 0.8491535782814026,
+      "eval_runtime": 853.0163,
+      "eval_samples_per_second": 28.206,
+      "eval_steps_per_second": 14.103,
+      "step": 6200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017775704244783053,
+      "loss": 0.939,
+      "step": 6201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001777501242701201,
+      "loss": 0.8421,
+      "step": 6202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000177743205151374,
+      "loss": 0.9808,
+      "step": 6203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000177736285091676,
+      "loss": 0.9612,
+      "step": 6204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017772936409110978,
+      "loss": 0.9483,
+      "step": 6205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017772244214975916,
+      "loss": 0.7909,
+      "step": 6206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017771551926770794,
+      "loss": 0.8527,
+      "step": 6207
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017770859544503985,
+      "loss": 0.9602,
+      "step": 6208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001777016706818387,
+      "loss": 1.0137,
+      "step": 6209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017769474497818832,
+      "loss": 0.7768,
+      "step": 6210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017768781833417253,
+      "loss": 0.7949,
+      "step": 6211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017768089074987522,
+      "loss": 0.8924,
+      "step": 6212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017767396222538015,
+      "loss": 0.8798,
+      "step": 6213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017766703276077118,
+      "loss": 0.8796,
+      "step": 6214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017766010235613225,
+      "loss": 0.9557,
+      "step": 6215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017765317101154724,
+      "loss": 0.8282,
+      "step": 6216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001776462387271,
+      "loss": 0.8248,
+      "step": 6217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001776393055028744,
+      "loss": 0.9398,
+      "step": 6218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001776323713389544,
+      "loss": 0.8649,
+      "step": 6219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017762543623542394,
+      "loss": 0.7343,
+      "step": 6220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017761850019236696,
+      "loss": 0.8642,
+      "step": 6221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017761156320986732,
+      "loss": 0.8493,
+      "step": 6222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017760462528800913,
+      "loss": 0.7645,
+      "step": 6223
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017759768642687624,
+      "loss": 0.8142,
+      "step": 6224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001775907466265527,
+      "loss": 0.8069,
+      "step": 6225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017758380588712248,
+      "loss": 0.8104,
+      "step": 6226
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017757686420866956,
+      "loss": 0.8839,
+      "step": 6227
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017756992159127802,
+      "loss": 0.9401,
+      "step": 6228
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017756297803503182,
+      "loss": 0.7293,
+      "step": 6229
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017755603354001505,
+      "loss": 0.832,
+      "step": 6230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017754908810631173,
+      "loss": 0.8216,
+      "step": 6231
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017754214173400595,
+      "loss": 0.837,
+      "step": 6232
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017753519442318178,
+      "loss": 0.8233,
+      "step": 6233
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001775282461739233,
+      "loss": 0.7978,
+      "step": 6234
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001775212969863146,
+      "loss": 0.9378,
+      "step": 6235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001775143468604398,
+      "loss": 0.9089,
+      "step": 6236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017750739579638298,
+      "loss": 0.7817,
+      "step": 6237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001775004437942283,
+      "loss": 0.8708,
+      "step": 6238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017749349085405998,
+      "loss": 0.8857,
+      "step": 6239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017748653697596201,
+      "loss": 0.8149,
+      "step": 6240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001774795821600187,
+      "loss": 0.8556,
+      "step": 6241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017747262640631416,
+      "loss": 0.7407,
+      "step": 6242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017746566971493257,
+      "loss": 0.8291,
+      "step": 6243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001774587120859582,
+      "loss": 0.8851,
+      "step": 6244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017745175351947516,
+      "loss": 0.9453,
+      "step": 6245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017744479401556772,
+      "loss": 0.8847,
+      "step": 6246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017743783357432013,
+      "loss": 0.8666,
+      "step": 6247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017743087219581661,
+      "loss": 0.8912,
+      "step": 6248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017742390988014146,
+      "loss": 0.9372,
+      "step": 6249
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017741694662737887,
+      "loss": 0.7743,
+      "step": 6250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001774099824376132,
+      "loss": 0.8448,
+      "step": 6251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017740301731092865,
+      "loss": 0.9899,
+      "step": 6252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001773960512474096,
+      "loss": 0.9045,
+      "step": 6253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017738908424714034,
+      "loss": 0.9008,
+      "step": 6254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001773821163102052,
+      "loss": 0.7259,
+      "step": 6255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017737514743668844,
+      "loss": 0.9194,
+      "step": 6256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017736817762667453,
+      "loss": 0.84,
+      "step": 6257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017736120688024775,
+      "loss": 0.9084,
+      "step": 6258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001773542351974925,
+      "loss": 0.9158,
+      "step": 6259
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001773472625784931,
+      "loss": 0.8996,
+      "step": 6260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017734028902333403,
+      "loss": 0.8645,
+      "step": 6261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017733331453209967,
+      "loss": 0.8921,
+      "step": 6262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017732633910487436,
+      "loss": 0.7879,
+      "step": 6263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017731936274174262,
+      "loss": 0.7822,
+      "step": 6264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017731238544278885,
+      "loss": 0.9127,
+      "step": 6265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001773054072080975,
+      "loss": 0.8674,
+      "step": 6266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017729842803775305,
+      "loss": 0.7174,
+      "step": 6267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017729144793183992,
+      "loss": 0.8385,
+      "step": 6268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001772844668904426,
+      "loss": 0.8137,
+      "step": 6269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017727748491364566,
+      "loss": 0.7822,
+      "step": 6270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017727050200153354,
+      "loss": 0.8189,
+      "step": 6271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017726351815419076,
+      "loss": 0.8822,
+      "step": 6272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017725653337170182,
+      "loss": 0.9396,
+      "step": 6273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017724954765415137,
+      "loss": 0.9578,
+      "step": 6274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017724256100162383,
+      "loss": 0.7554,
+      "step": 6275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017723557341420382,
+      "loss": 0.8797,
+      "step": 6276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017722858489197592,
+      "loss": 0.9342,
+      "step": 6277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001772215954350247,
+      "loss": 0.8658,
+      "step": 6278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017721460504343476,
+      "loss": 0.8604,
+      "step": 6279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017720761371729075,
+      "loss": 0.7218,
+      "step": 6280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001772006214566772,
+      "loss": 0.846,
+      "step": 6281
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001771936282616788,
+      "loss": 0.9438,
+      "step": 6282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017718663413238017,
+      "loss": 0.8905,
+      "step": 6283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000177179639068866,
+      "loss": 0.9057,
+      "step": 6284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017717264307122088,
+      "loss": 0.9027,
+      "step": 6285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017716564613952954,
+      "loss": 0.9681,
+      "step": 6286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017715864827387665,
+      "loss": 0.8387,
+      "step": 6287
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017715164947434693,
+      "loss": 0.9445,
+      "step": 6288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017714464974102507,
+      "loss": 0.9369,
+      "step": 6289
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017713764907399576,
+      "loss": 0.8556,
+      "step": 6290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001771306474733438,
+      "loss": 0.8723,
+      "step": 6291
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017712364493915387,
+      "loss": 0.7623,
+      "step": 6292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017711664147151076,
+      "loss": 0.7248,
+      "step": 6293
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017710963707049922,
+      "loss": 0.7961,
+      "step": 6294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017710263173620404,
+      "loss": 0.8911,
+      "step": 6295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017709562546870997,
+      "loss": 0.9159,
+      "step": 6296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017708861826810186,
+      "loss": 0.8487,
+      "step": 6297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017708161013446448,
+      "loss": 0.9868,
+      "step": 6298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001770746010678827,
+      "loss": 0.9768,
+      "step": 6299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001770675910684413,
+      "loss": 0.9138,
+      "step": 6300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017706058013622515,
+      "loss": 0.8629,
+      "step": 6301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017705356827131912,
+      "loss": 0.9061,
+      "step": 6302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017704655547380805,
+      "loss": 0.8465,
+      "step": 6303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017703954174377683,
+      "loss": 0.7297,
+      "step": 6304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017703252708131033,
+      "loss": 0.783,
+      "step": 6305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017702551148649347,
+      "loss": 0.8551,
+      "step": 6306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001770184949594112,
+      "loss": 0.9252,
+      "step": 6307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017701147750014834,
+      "loss": 0.7957,
+      "step": 6308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017700445910878994,
+      "loss": 0.8391,
+      "step": 6309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017699743978542086,
+      "loss": 0.8749,
+      "step": 6310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001769904195301261,
+      "loss": 0.7783,
+      "step": 6311
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017698339834299061,
+      "loss": 0.8092,
+      "step": 6312
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017697637622409942,
+      "loss": 0.8984,
+      "step": 6313
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017696935317353747,
+      "loss": 0.8179,
+      "step": 6314
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017696232919138974,
+      "loss": 1.0347,
+      "step": 6315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017695530427774128,
+      "loss": 0.8425,
+      "step": 6316
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017694827843267716,
+      "loss": 0.7734,
+      "step": 6317
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017694125165628233,
+      "loss": 0.9031,
+      "step": 6318
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017693422394864187,
+      "loss": 0.991,
+      "step": 6319
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017692719530984085,
+      "loss": 0.8024,
+      "step": 6320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017692016573996435,
+      "loss": 0.8807,
+      "step": 6321
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017691313523909746,
+      "loss": 0.9529,
+      "step": 6322
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001769061038073252,
+      "loss": 0.9634,
+      "step": 6323
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017689907144473273,
+      "loss": 0.9372,
+      "step": 6324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017689203815140516,
+      "loss": 0.8944,
+      "step": 6325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017688500392742758,
+      "loss": 0.8147,
+      "step": 6326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017687796877288522,
+      "loss": 0.8747,
+      "step": 6327
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017687093268786312,
+      "loss": 0.8835,
+      "step": 6328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017686389567244653,
+      "loss": 0.8729,
+      "step": 6329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017685685772672057,
+      "loss": 0.8136,
+      "step": 6330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017684981885077042,
+      "loss": 0.6872,
+      "step": 6331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001768427790446813,
+      "loss": 0.8837,
+      "step": 6332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017683573830853838,
+      "loss": 0.8325,
+      "step": 6333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001768286966424269,
+      "loss": 0.9525,
+      "step": 6334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001768216540464321,
+      "loss": 0.7924,
+      "step": 6335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017681461052063922,
+      "loss": 0.9635,
+      "step": 6336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017680756606513348,
+      "loss": 0.8544,
+      "step": 6337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017680052068000013,
+      "loss": 0.8132,
+      "step": 6338
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017679347436532449,
+      "loss": 0.8078,
+      "step": 6339
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017678642712119183,
+      "loss": 0.8219,
+      "step": 6340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017677937894768745,
+      "loss": 0.842,
+      "step": 6341
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017677232984489663,
+      "loss": 0.8578,
+      "step": 6342
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001767652798129047,
+      "loss": 0.9518,
+      "step": 6343
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017675822885179697,
+      "loss": 0.9367,
+      "step": 6344
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017675117696165884,
+      "loss": 0.8448,
+      "step": 6345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001767441241425756,
+      "loss": 0.7525,
+      "step": 6346
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017673707039463264,
+      "loss": 0.8136,
+      "step": 6347
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017673001571791533,
+      "loss": 0.7931,
+      "step": 6348
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017672296011250906,
+      "loss": 0.7877,
+      "step": 6349
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017671590357849922,
+      "loss": 0.7361,
+      "step": 6350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001767088461159712,
+      "loss": 0.9071,
+      "step": 6351
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017670178772501044,
+      "loss": 0.7791,
+      "step": 6352
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017669472840570237,
+      "loss": 0.7796,
+      "step": 6353
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017668766815813244,
+      "loss": 0.8028,
+      "step": 6354
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001766806069823861,
+      "loss": 0.7739,
+      "step": 6355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017667354487854877,
+      "loss": 0.753,
+      "step": 6356
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000176666481846706,
+      "loss": 0.7568,
+      "step": 6357
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001766594178869432,
+      "loss": 0.7599,
+      "step": 6358
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017665235299934592,
+      "loss": 0.781,
+      "step": 6359
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017664528718399963,
+      "loss": 0.8901,
+      "step": 6360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001766382204409899,
+      "loss": 0.7722,
+      "step": 6361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017663115277040224,
+      "loss": 0.8397,
+      "step": 6362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017662408417232217,
+      "loss": 0.8751,
+      "step": 6363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017661701464683524,
+      "loss": 0.9059,
+      "step": 6364
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017660994419402707,
+      "loss": 0.913,
+      "step": 6365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017660287281398318,
+      "loss": 0.8114,
+      "step": 6366
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017659580050678918,
+      "loss": 0.7746,
+      "step": 6367
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017658872727253067,
+      "loss": 0.8131,
+      "step": 6368
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017658165311129325,
+      "loss": 0.8606,
+      "step": 6369
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017657457802316254,
+      "loss": 0.7693,
+      "step": 6370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001765675020082242,
+      "loss": 0.871,
+      "step": 6371
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017656042506656382,
+      "loss": 0.8775,
+      "step": 6372
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017655334719826713,
+      "loss": 0.9732,
+      "step": 6373
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017654626840341972,
+      "loss": 0.826,
+      "step": 6374
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001765391886821073,
+      "loss": 0.7714,
+      "step": 6375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017653210803441556,
+      "loss": 0.9355,
+      "step": 6376
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017652502646043022,
+      "loss": 0.8789,
+      "step": 6377
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017651794396023694,
+      "loss": 0.8951,
+      "step": 6378
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017651086053392144,
+      "loss": 0.8755,
+      "step": 6379
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017650377618156954,
+      "loss": 0.8284,
+      "step": 6380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017649669090326692,
+      "loss": 0.8038,
+      "step": 6381
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017648960469909933,
+      "loss": 0.7654,
+      "step": 6382
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017648251756915252,
+      "loss": 0.8094,
+      "step": 6383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017647542951351234,
+      "loss": 0.8564,
+      "step": 6384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017646834053226451,
+      "loss": 0.9065,
+      "step": 6385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017646125062549488,
+      "loss": 0.8052,
+      "step": 6386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001764541597932892,
+      "loss": 0.8265,
+      "step": 6387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017644706803573336,
+      "loss": 0.9092,
+      "step": 6388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017643997535291318,
+      "loss": 0.7564,
+      "step": 6389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017643288174491444,
+      "loss": 0.7907,
+      "step": 6390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001764257872118231,
+      "loss": 0.74,
+      "step": 6391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017641869175372493,
+      "loss": 0.8479,
+      "step": 6392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017641159537070585,
+      "loss": 0.8305,
+      "step": 6393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017640449806285177,
+      "loss": 0.7744,
+      "step": 6394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001763973998302486,
+      "loss": 0.7181,
+      "step": 6395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017639030067298215,
+      "loss": 0.8193,
+      "step": 6396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017638320059113852,
+      "loss": 0.8864,
+      "step": 6397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001763760995848035,
+      "loss": 0.8354,
+      "step": 6398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017636899765406306,
+      "loss": 0.8577,
+      "step": 6399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017636189479900316,
+      "loss": 0.8446,
+      "step": 6400
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 0.8379415273666382,
+      "eval_runtime": 850.4174,
+      "eval_samples_per_second": 28.292,
+      "eval_steps_per_second": 14.146,
+      "step": 6400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017635479101970982,
+      "loss": 0.8868,
+      "step": 6401
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017634768631626896,
+      "loss": 0.7709,
+      "step": 6402
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001763405806887666,
+      "loss": 0.8057,
+      "step": 6403
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017633347413728876,
+      "loss": 0.8968,
+      "step": 6404
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001763263666619214,
+      "loss": 0.8058,
+      "step": 6405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001763192582627506,
+      "loss": 0.7786,
+      "step": 6406
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017631214893986234,
+      "loss": 0.8564,
+      "step": 6407
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017630503869334272,
+      "loss": 0.8643,
+      "step": 6408
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017629792752327776,
+      "loss": 0.8124,
+      "step": 6409
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017629081542975354,
+      "loss": 0.8982,
+      "step": 6410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017628370241285616,
+      "loss": 0.9033,
+      "step": 6411
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017627658847267167,
+      "loss": 0.9086,
+      "step": 6412
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001762694736092862,
+      "loss": 0.8585,
+      "step": 6413
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017626235782278586,
+      "loss": 0.7944,
+      "step": 6414
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017625524111325677,
+      "loss": 0.8998,
+      "step": 6415
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017624812348078508,
+      "loss": 0.9056,
+      "step": 6416
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017624100492545693,
+      "loss": 0.8402,
+      "step": 6417
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017623388544735845,
+      "loss": 0.8304,
+      "step": 6418
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017622676504657587,
+      "loss": 0.9,
+      "step": 6419
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001762196437231953,
+      "loss": 0.8521,
+      "step": 6420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000176212521477303,
+      "loss": 0.8026,
+      "step": 6421
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001762053983089851,
+      "loss": 0.8626,
+      "step": 6422
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001761982742183279,
+      "loss": 0.8057,
+      "step": 6423
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017619114920541752,
+      "loss": 0.8373,
+      "step": 6424
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017618402327034032,
+      "loss": 0.7684,
+      "step": 6425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017617689641318245,
+      "loss": 0.9065,
+      "step": 6426
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017616976863403018,
+      "loss": 0.7372,
+      "step": 6427
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017616263993296984,
+      "loss": 0.7835,
+      "step": 6428
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00017615551031008767,
+      "loss": 0.8436,
+      "step": 6429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017614837976546993,
+      "loss": 0.8094,
+      "step": 6430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000176141248299203,
+      "loss": 0.8447,
+      "step": 6431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017613411591137316,
+      "loss": 0.8515,
+      "step": 6432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017612698260206666,
+      "loss": 0.7602,
+      "step": 6433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017611984837136998,
+      "loss": 0.9085,
+      "step": 6434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017611271321936936,
+      "loss": 0.7055,
+      "step": 6435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017610557714615118,
+      "loss": 0.74,
+      "step": 6436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017609844015180183,
+      "loss": 0.8982,
+      "step": 6437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017609130223640769,
+      "loss": 0.8519,
+      "step": 6438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017608416340005514,
+      "loss": 0.6622,
+      "step": 6439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017607702364283057,
+      "loss": 0.9284,
+      "step": 6440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017606988296482045,
+      "loss": 0.6781,
+      "step": 6441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017606274136611116,
+      "loss": 0.7754,
+      "step": 6442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017605559884678918,
+      "loss": 0.7622,
+      "step": 6443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017604845540694086,
+      "loss": 0.8176,
+      "step": 6444
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017604131104665278,
+      "loss": 0.92,
+      "step": 6445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017603416576601134,
+      "loss": 0.8903,
+      "step": 6446
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017602701956510302,
+      "loss": 0.8033,
+      "step": 6447
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017601987244401435,
+      "loss": 0.9265,
+      "step": 6448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001760127244028318,
+      "loss": 0.8479,
+      "step": 6449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017600557544164195,
+      "loss": 0.7622,
+      "step": 6450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017599842556053122,
+      "loss": 0.8077,
+      "step": 6451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017599127475958622,
+      "loss": 0.8355,
+      "step": 6452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001759841230388935,
+      "loss": 1.0408,
+      "step": 6453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017597697039853957,
+      "loss": 1.0032,
+      "step": 6454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017596981683861104,
+      "loss": 0.7273,
+      "step": 6455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001759626623591945,
+      "loss": 0.7824,
+      "step": 6456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017595550696037653,
+      "loss": 0.8386,
+      "step": 6457
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017594835064224372,
+      "loss": 0.8881,
+      "step": 6458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001759411934048827,
+      "loss": 0.8377,
+      "step": 6459
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017593403524838007,
+      "loss": 0.9841,
+      "step": 6460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017592687617282252,
+      "loss": 0.8062,
+      "step": 6461
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017591971617829665,
+      "loss": 0.8937,
+      "step": 6462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017591255526488915,
+      "loss": 0.9098,
+      "step": 6463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017590539343268673,
+      "loss": 0.8815,
+      "step": 6464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017589823068177594,
+      "loss": 0.8546,
+      "step": 6465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017589106701224361,
+      "loss": 0.9261,
+      "step": 6466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017588390242417637,
+      "loss": 0.7519,
+      "step": 6467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017587673691766095,
+      "loss": 0.8765,
+      "step": 6468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001758695704927841,
+      "loss": 0.8628,
+      "step": 6469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017586240314963254,
+      "loss": 0.8358,
+      "step": 6470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000175855234888293,
+      "loss": 0.8237,
+      "step": 6471
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017584806570885227,
+      "loss": 0.8509,
+      "step": 6472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001758408956113971,
+      "loss": 0.9517,
+      "step": 6473
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017583372459601432,
+      "loss": 0.8873,
+      "step": 6474
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017582655266279063,
+      "loss": 0.8102,
+      "step": 6475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001758193798118129,
+      "loss": 0.7526,
+      "step": 6476
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017581220604316795,
+      "loss": 0.8271,
+      "step": 6477
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017580503135694258,
+      "loss": 0.8252,
+      "step": 6478
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017579785575322365,
+      "loss": 0.8085,
+      "step": 6479
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000175790679232098,
+      "loss": 0.7596,
+      "step": 6480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001757835017936525,
+      "loss": 0.8923,
+      "step": 6481
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017577632343797396,
+      "loss": 0.7999,
+      "step": 6482
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017576914416514935,
+      "loss": 0.7292,
+      "step": 6483
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001757619639752655,
+      "loss": 0.8459,
+      "step": 6484
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017575478286840934,
+      "loss": 0.7956,
+      "step": 6485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001757476008446678,
+      "loss": 0.7697,
+      "step": 6486
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017574041790412777,
+      "loss": 0.8365,
+      "step": 6487
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001757332340468762,
+      "loss": 0.9888,
+      "step": 6488
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017572604927300006,
+      "loss": 0.8231,
+      "step": 6489
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001757188635825863,
+      "loss": 0.8361,
+      "step": 6490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017571167697572187,
+      "loss": 0.8242,
+      "step": 6491
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017570448945249378,
+      "loss": 0.8929,
+      "step": 6492
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017569730101298903,
+      "loss": 0.7221,
+      "step": 6493
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017569011165729457,
+      "loss": 0.818,
+      "step": 6494
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017568292138549744,
+      "loss": 0.7027,
+      "step": 6495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001756757301976847,
+      "loss": 0.8561,
+      "step": 6496
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001756685380939434,
+      "loss": 0.7079,
+      "step": 6497
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001756613450743605,
+      "loss": 0.7453,
+      "step": 6498
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001756541511390231,
+      "loss": 0.8381,
+      "step": 6499
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017564695628801831,
+      "loss": 0.8549,
+      "step": 6500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001756397605214332,
+      "loss": 0.8686,
+      "step": 6501
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017563256383935482,
+      "loss": 0.7493,
+      "step": 6502
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017562536624187033,
+      "loss": 0.8752,
+      "step": 6503
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017561816772906677,
+      "loss": 0.9124,
+      "step": 6504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017561096830103132,
+      "loss": 0.8565,
+      "step": 6505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017560376795785115,
+      "loss": 0.9197,
+      "step": 6506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017559656669961332,
+      "loss": 0.9588,
+      "step": 6507
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017558936452640505,
+      "loss": 0.8376,
+      "step": 6508
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001755821614383135,
+      "loss": 0.7974,
+      "step": 6509
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017557495743542585,
+      "loss": 0.8652,
+      "step": 6510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017556775251782928,
+      "loss": 0.8826,
+      "step": 6511
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017556054668561098,
+      "loss": 0.9917,
+      "step": 6512
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017555333993885826,
+      "loss": 0.8165,
+      "step": 6513
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001755461322776582,
+      "loss": 0.9001,
+      "step": 6514
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017553892370209814,
+      "loss": 0.9821,
+      "step": 6515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001755317142122653,
+      "loss": 0.8912,
+      "step": 6516
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017552450380824695,
+      "loss": 0.8759,
+      "step": 6517
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017551729249013032,
+      "loss": 0.7507,
+      "step": 6518
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017551008025800272,
+      "loss": 0.8555,
+      "step": 6519
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017550286711195146,
+      "loss": 0.8239,
+      "step": 6520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001754956530520638,
+      "loss": 0.9247,
+      "step": 6521
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001754884380784271,
+      "loss": 0.8584,
+      "step": 6522
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017548122219112865,
+      "loss": 0.812,
+      "step": 6523
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001754740053902558,
+      "loss": 0.8778,
+      "step": 6524
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017546678767589588,
+      "loss": 0.8403,
+      "step": 6525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001754595690481363,
+      "loss": 0.7872,
+      "step": 6526
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017545234950706436,
+      "loss": 0.8932,
+      "step": 6527
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017544512905276748,
+      "loss": 0.8893,
+      "step": 6528
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017543790768533306,
+      "loss": 0.9185,
+      "step": 6529
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001754306854048485,
+      "loss": 0.8363,
+      "step": 6530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017542346221140117,
+      "loss": 0.6956,
+      "step": 6531
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017541623810507856,
+      "loss": 0.881,
+      "step": 6532
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017540901308596803,
+      "loss": 0.8714,
+      "step": 6533
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017540178715415712,
+      "loss": 0.8349,
+      "step": 6534
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001753945603097332,
+      "loss": 0.8531,
+      "step": 6535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017538733255278378,
+      "loss": 1.0023,
+      "step": 6536
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017538010388339634,
+      "loss": 0.8833,
+      "step": 6537
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001753728743016584,
+      "loss": 0.9002,
+      "step": 6538
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017536564380765739,
+      "loss": 0.8727,
+      "step": 6539
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017535841240148088,
+      "loss": 0.8515,
+      "step": 6540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017535118008321635,
+      "loss": 0.9949,
+      "step": 6541
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017534394685295138,
+      "loss": 0.7549,
+      "step": 6542
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017533671271077349,
+      "loss": 0.7667,
+      "step": 6543
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017532947765677026,
+      "loss": 0.7363,
+      "step": 6544
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017532224169102925,
+      "loss": 0.927,
+      "step": 6545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000175315004813638,
+      "loss": 0.752,
+      "step": 6546
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017530776702468417,
+      "loss": 0.8209,
+      "step": 6547
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001753005283242553,
+      "loss": 0.8156,
+      "step": 6548
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017529328871243905,
+      "loss": 0.8523,
+      "step": 6549
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000175286048189323,
+      "loss": 0.7847,
+      "step": 6550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017527880675499482,
+      "loss": 0.8723,
+      "step": 6551
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017527156440954212,
+      "loss": 0.7289,
+      "step": 6552
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017526432115305256,
+      "loss": 0.8114,
+      "step": 6553
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017525707698561385,
+      "loss": 0.7896,
+      "step": 6554
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017524983190731367,
+      "loss": 0.7855,
+      "step": 6555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017524258591823967,
+      "loss": 1.0148,
+      "step": 6556
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001752353390184795,
+      "loss": 0.8579,
+      "step": 6557
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000175228091208121,
+      "loss": 0.9363,
+      "step": 6558
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001752208424872518,
+      "loss": 0.8425,
+      "step": 6559
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017521359285595968,
+      "loss": 0.798,
+      "step": 6560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017520634231433236,
+      "loss": 0.9158,
+      "step": 6561
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017519909086245757,
+      "loss": 0.7529,
+      "step": 6562
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017519183850042314,
+      "loss": 0.7562,
+      "step": 6563
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001751845852283168,
+      "loss": 0.8211,
+      "step": 6564
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001751773310462264,
+      "loss": 0.7562,
+      "step": 6565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017517007595423966,
+      "loss": 0.8773,
+      "step": 6566
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001751628199524444,
+      "loss": 0.8407,
+      "step": 6567
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001751555630409285,
+      "loss": 0.756,
+      "step": 6568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017514830521977976,
+      "loss": 0.94,
+      "step": 6569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017514104648908601,
+      "loss": 0.8319,
+      "step": 6570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017513378684893516,
+      "loss": 0.7659,
+      "step": 6571
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017512652629941497,
+      "loss": 0.8038,
+      "step": 6572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017511926484061342,
+      "loss": 0.8342,
+      "step": 6573
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017511200247261833,
+      "loss": 0.8388,
+      "step": 6574
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017510473919551766,
+      "loss": 0.8588,
+      "step": 6575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017509747500939928,
+      "loss": 0.7625,
+      "step": 6576
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017509020991435113,
+      "loss": 0.7572,
+      "step": 6577
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001750829439104611,
+      "loss": 0.8097,
+      "step": 6578
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001750756769978172,
+      "loss": 0.8768,
+      "step": 6579
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001750684091765073,
+      "loss": 0.8169,
+      "step": 6580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017506114044661946,
+      "loss": 0.6668,
+      "step": 6581
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017505387080824156,
+      "loss": 0.7706,
+      "step": 6582
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017504660026146167,
+      "loss": 0.7558,
+      "step": 6583
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017503932880636773,
+      "loss": 0.7031,
+      "step": 6584
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001750320564430478,
+      "loss": 0.84,
+      "step": 6585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017502478317158986,
+      "loss": 0.7966,
+      "step": 6586
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017501750899208193,
+      "loss": 0.7573,
+      "step": 6587
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017501023390461212,
+      "loss": 0.6979,
+      "step": 6588
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001750029579092684,
+      "loss": 0.7113,
+      "step": 6589
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017499568100613888,
+      "loss": 0.7232,
+      "step": 6590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017498840319531164,
+      "loss": 0.73,
+      "step": 6591
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017498112447687474,
+      "loss": 0.7107,
+      "step": 6592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017497384485091626,
+      "loss": 0.824,
+      "step": 6593
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017496656431752436,
+      "loss": 0.8346,
+      "step": 6594
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017495928287678716,
+      "loss": 0.854,
+      "step": 6595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001749520005287927,
+      "loss": 0.711,
+      "step": 6596
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017494471727362926,
+      "loss": 0.803,
+      "step": 6597
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017493743311138487,
+      "loss": 0.7881,
+      "step": 6598
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017493014804214772,
+      "loss": 0.7954,
+      "step": 6599
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017492286206600605,
+      "loss": 0.8704,
+      "step": 6600
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 0.826124370098114,
+      "eval_runtime": 852.9701,
+      "eval_samples_per_second": 28.207,
+      "eval_steps_per_second": 14.104,
+      "step": 6600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000174915575183048,
+      "loss": 0.9107,
+      "step": 6601
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017490828739336173,
+      "loss": 0.8963,
+      "step": 6602
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001749009986970355,
+      "loss": 0.8902,
+      "step": 6603
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001748937090941575,
+      "loss": 0.7695,
+      "step": 6604
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017488641858481597,
+      "loss": 0.9326,
+      "step": 6605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017487912716909913,
+      "loss": 0.7716,
+      "step": 6606
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001748718348470953,
+      "loss": 0.7718,
+      "step": 6607
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001748645416188926,
+      "loss": 0.7549,
+      "step": 6608
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017485724748457943,
+      "loss": 0.8148,
+      "step": 6609
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017484995244424406,
+      "loss": 0.7539,
+      "step": 6610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017484265649797472,
+      "loss": 0.8065,
+      "step": 6611
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017483535964585975,
+      "loss": 0.6964,
+      "step": 6612
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001748280618879875,
+      "loss": 0.8968,
+      "step": 6613
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017482076322444625,
+      "loss": 0.9198,
+      "step": 6614
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017481346365532435,
+      "loss": 0.8057,
+      "step": 6615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017480616318071013,
+      "loss": 0.8991,
+      "step": 6616
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017479886180069202,
+      "loss": 0.8094,
+      "step": 6617
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001747915595153583,
+      "loss": 0.726,
+      "step": 6618
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017478425632479742,
+      "loss": 0.8703,
+      "step": 6619
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017477695222909776,
+      "loss": 0.8213,
+      "step": 6620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017476964722834768,
+      "loss": 0.7589,
+      "step": 6621
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017476234132263567,
+      "loss": 0.7968,
+      "step": 6622
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017475503451205005,
+      "loss": 0.769,
+      "step": 6623
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017474772679667937,
+      "loss": 0.7122,
+      "step": 6624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017474041817661204,
+      "loss": 0.8922,
+      "step": 6625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017473310865193648,
+      "loss": 0.7684,
+      "step": 6626
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017472579822274117,
+      "loss": 0.8242,
+      "step": 6627
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017471848688911464,
+      "loss": 0.8025,
+      "step": 6628
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017471117465114535,
+      "loss": 0.8759,
+      "step": 6629
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017470386150892178,
+      "loss": 0.8143,
+      "step": 6630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017469654746253248,
+      "loss": 0.8445,
+      "step": 6631
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017468923251206592,
+      "loss": 0.8879,
+      "step": 6632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001746819166576107,
+      "loss": 0.7957,
+      "step": 6633
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017467459989925537,
+      "loss": 0.9195,
+      "step": 6634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017466728223708842,
+      "loss": 0.7573,
+      "step": 6635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017465996367119847,
+      "loss": 0.9317,
+      "step": 6636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001746526442016741,
+      "loss": 0.9207,
+      "step": 6637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017464532382860384,
+      "loss": 0.8591,
+      "step": 6638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017463800255207635,
+      "loss": 0.8997,
+      "step": 6639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017463068037218025,
+      "loss": 0.858,
+      "step": 6640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017462335728900415,
+      "loss": 0.8413,
+      "step": 6641
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017461603330263663,
+      "loss": 0.7863,
+      "step": 6642
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001746087084131664,
+      "loss": 0.8493,
+      "step": 6643
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017460138262068208,
+      "loss": 0.7263,
+      "step": 6644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001745940559252724,
+      "loss": 0.7337,
+      "step": 6645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017458672832702593,
+      "loss": 0.7529,
+      "step": 6646
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017457939982603145,
+      "loss": 0.8267,
+      "step": 6647
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001745720704223776,
+      "loss": 0.7677,
+      "step": 6648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017456474011615312,
+      "loss": 0.8544,
+      "step": 6649
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017455740890744674,
+      "loss": 0.8918,
+      "step": 6650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017455007679634716,
+      "loss": 0.9766,
+      "step": 6651
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017454274378294316,
+      "loss": 0.766,
+      "step": 6652
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017453540986732348,
+      "loss": 0.8666,
+      "step": 6653
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017452807504957684,
+      "loss": 0.8818,
+      "step": 6654
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001745207393297921,
+      "loss": 0.8547,
+      "step": 6655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017451340270805798,
+      "loss": 0.8957,
+      "step": 6656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017450606518446332,
+      "loss": 0.7057,
+      "step": 6657
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017449872675909692,
+      "loss": 0.7728,
+      "step": 6658
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017449138743204755,
+      "loss": 0.7954,
+      "step": 6659
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001744840472034041,
+      "loss": 0.7135,
+      "step": 6660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017447670607325538,
+      "loss": 0.8003,
+      "step": 6661
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017446936404169025,
+      "loss": 0.8688,
+      "step": 6662
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017446202110879758,
+      "loss": 0.7643,
+      "step": 6663
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017445467727466628,
+      "loss": 0.8016,
+      "step": 6664
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017444733253938513,
+      "loss": 0.8477,
+      "step": 6665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017443998690304312,
+      "loss": 0.8404,
+      "step": 6666
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017443264036572914,
+      "loss": 0.7436,
+      "step": 6667
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017442529292753208,
+      "loss": 1.074,
+      "step": 6668
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017441794458854086,
+      "loss": 0.8354,
+      "step": 6669
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017441059534884447,
+      "loss": 0.8381,
+      "step": 6670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017440324520853184,
+      "loss": 0.9189,
+      "step": 6671
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001743958941676919,
+      "loss": 0.7586,
+      "step": 6672
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017438854222641366,
+      "loss": 0.7509,
+      "step": 6673
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017438118938478607,
+      "loss": 0.7505,
+      "step": 6674
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017437383564289816,
+      "loss": 0.9469,
+      "step": 6675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017436648100083892,
+      "loss": 0.9062,
+      "step": 6676
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017435912545869735,
+      "loss": 0.8094,
+      "step": 6677
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001743517690165625,
+      "loss": 0.7876,
+      "step": 6678
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017434441167452338,
+      "loss": 0.8392,
+      "step": 6679
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017433705343266906,
+      "loss": 0.88,
+      "step": 6680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001743296942910886,
+      "loss": 0.8116,
+      "step": 6681
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017432233424987108,
+      "loss": 0.8278,
+      "step": 6682
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017431497330910552,
+      "loss": 0.6856,
+      "step": 6683
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017430761146888113,
+      "loss": 0.7705,
+      "step": 6684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001743002487292869,
+      "loss": 0.8374,
+      "step": 6685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017429288509041198,
+      "loss": 0.8038,
+      "step": 6686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001742855205523455,
+      "loss": 0.7972,
+      "step": 6687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001742781551151766,
+      "loss": 0.8022,
+      "step": 6688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001742707887789944,
+      "loss": 0.7471,
+      "step": 6689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017426342154388812,
+      "loss": 0.8766,
+      "step": 6690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017425605340994684,
+      "loss": 0.8018,
+      "step": 6691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001742486843772598,
+      "loss": 0.8109,
+      "step": 6692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001742413144459162,
+      "loss": 0.9807,
+      "step": 6693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001742339436160052,
+      "loss": 0.7877,
+      "step": 6694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017422657188761602,
+      "loss": 0.8713,
+      "step": 6695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001742191992608379,
+      "loss": 0.7086,
+      "step": 6696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017421182573576006,
+      "loss": 0.8302,
+      "step": 6697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017420445131247173,
+      "loss": 0.8448,
+      "step": 6698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001741970759910622,
+      "loss": 0.7428,
+      "step": 6699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017418969977162072,
+      "loss": 0.9075,
+      "step": 6700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017418232265423655,
+      "loss": 0.7786,
+      "step": 6701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000174174944638999,
+      "loss": 0.6772,
+      "step": 6702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017416756572599738,
+      "loss": 0.7616,
+      "step": 6703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017416018591532095,
+      "loss": 0.6989,
+      "step": 6704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017415280520705908,
+      "loss": 0.9485,
+      "step": 6705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017414542360130107,
+      "loss": 0.8607,
+      "step": 6706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001741380410981363,
+      "loss": 1.0408,
+      "step": 6707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017413065769765406,
+      "loss": 0.8151,
+      "step": 6708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017412327339994378,
+      "loss": 0.8309,
+      "step": 6709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017411588820509478,
+      "loss": 0.9049,
+      "step": 6710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001741085021131965,
+      "loss": 0.896,
+      "step": 6711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001741011151243383,
+      "loss": 0.9205,
+      "step": 6712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017409372723860958,
+      "loss": 0.7642,
+      "step": 6713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001740863384560998,
+      "loss": 0.737,
+      "step": 6714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017407894877689834,
+      "loss": 0.736,
+      "step": 6715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017407155820109468,
+      "loss": 0.8546,
+      "step": 6716
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017406416672877822,
+      "loss": 0.8545,
+      "step": 6717
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017405677436003848,
+      "loss": 0.8223,
+      "step": 6718
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001740493810949649,
+      "loss": 0.8759,
+      "step": 6719
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017404198693364694,
+      "loss": 0.9648,
+      "step": 6720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017403459187617418,
+      "loss": 0.8773,
+      "step": 6721
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017402719592263602,
+      "loss": 0.9211,
+      "step": 6722
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017401979907312204,
+      "loss": 0.9105,
+      "step": 6723
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017401240132772173,
+      "loss": 0.8127,
+      "step": 6724
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017400500268652467,
+      "loss": 0.8855,
+      "step": 6725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001739976031496204,
+      "loss": 0.7939,
+      "step": 6726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017399020271709846,
+      "loss": 0.801,
+      "step": 6727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001739828013890484,
+      "loss": 0.6792,
+      "step": 6728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017397539916555983,
+      "loss": 0.8508,
+      "step": 6729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017396799604672234,
+      "loss": 0.9351,
+      "step": 6730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017396059203262553,
+      "loss": 0.767,
+      "step": 6731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000173953187123359,
+      "loss": 0.8061,
+      "step": 6732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001739457813190124,
+      "loss": 0.8498,
+      "step": 6733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017393837461967537,
+      "loss": 0.778,
+      "step": 6734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017393096702543752,
+      "loss": 0.9172,
+      "step": 6735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017392355853638853,
+      "loss": 0.838,
+      "step": 6736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017391614915261805,
+      "loss": 0.8555,
+      "step": 6737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017390873887421576,
+      "loss": 0.8131,
+      "step": 6738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017390132770127138,
+      "loss": 0.8409,
+      "step": 6739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017389391563387457,
+      "loss": 0.6851,
+      "step": 6740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017388650267211508,
+      "loss": 0.8599,
+      "step": 6741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001738790888160826,
+      "loss": 0.8053,
+      "step": 6742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017387167406586687,
+      "loss": 0.7697,
+      "step": 6743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017386425842155765,
+      "loss": 0.9367,
+      "step": 6744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001738568418832447,
+      "loss": 0.7501,
+      "step": 6745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017384942445101772,
+      "loss": 0.7951,
+      "step": 6746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017384200612496656,
+      "loss": 0.8244,
+      "step": 6747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017383458690518094,
+      "loss": 0.8265,
+      "step": 6748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017382716679175073,
+      "loss": 0.7551,
+      "step": 6749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017381974578476567,
+      "loss": 0.9778,
+      "step": 6750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017381232388431563,
+      "loss": 0.7899,
+      "step": 6751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017380490109049042,
+      "loss": 0.7757,
+      "step": 6752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017379747740337993,
+      "loss": 0.7602,
+      "step": 6753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001737900528230739,
+      "loss": 0.8843,
+      "step": 6754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017378262734966227,
+      "loss": 0.9267,
+      "step": 6755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001737752009832349,
+      "loss": 0.96,
+      "step": 6756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001737677737238817,
+      "loss": 0.8142,
+      "step": 6757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001737603455716925,
+      "loss": 0.6974,
+      "step": 6758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017375291652675724,
+      "loss": 0.8806,
+      "step": 6759
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017374548658916585,
+      "loss": 0.7874,
+      "step": 6760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017373805575900825,
+      "loss": 0.7934,
+      "step": 6761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001737306240363744,
+      "loss": 0.7917,
+      "step": 6762
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001737231914213542,
+      "loss": 0.9289,
+      "step": 6763
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017371575791403761,
+      "loss": 0.6955,
+      "step": 6764
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017370832351451461,
+      "loss": 0.727,
+      "step": 6765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017370088822287524,
+      "loss": 0.9229,
+      "step": 6766
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017369345203920941,
+      "loss": 0.8285,
+      "step": 6767
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017368601496360717,
+      "loss": 0.8389,
+      "step": 6768
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001736785769961585,
+      "loss": 0.8062,
+      "step": 6769
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017367113813695348,
+      "loss": 0.8559,
+      "step": 6770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001736636983860821,
+      "loss": 0.6993,
+      "step": 6771
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001736562577436344,
+      "loss": 0.8236,
+      "step": 6772
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017364881620970048,
+      "loss": 0.9221,
+      "step": 6773
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017364137378437033,
+      "loss": 0.9098,
+      "step": 6774
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017363393046773413,
+      "loss": 0.8043,
+      "step": 6775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017362648625988188,
+      "loss": 0.9035,
+      "step": 6776
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017361904116090373,
+      "loss": 0.8577,
+      "step": 6777
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001736115951708898,
+      "loss": 0.8406,
+      "step": 6778
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017360414828993014,
+      "loss": 0.7206,
+      "step": 6779
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017359670051811496,
+      "loss": 0.8547,
+      "step": 6780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017358925185553437,
+      "loss": 0.8403,
+      "step": 6781
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017358180230227853,
+      "loss": 0.7439,
+      "step": 6782
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017357435185843758,
+      "loss": 0.8406,
+      "step": 6783
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017356690052410174,
+      "loss": 0.8952,
+      "step": 6784
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017355944829936117,
+      "loss": 0.7928,
+      "step": 6785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017355199518430604,
+      "loss": 0.905,
+      "step": 6786
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017354454117902664,
+      "loss": 0.8222,
+      "step": 6787
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017353708628361307,
+      "loss": 0.9432,
+      "step": 6788
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017352963049815565,
+      "loss": 0.9816,
+      "step": 6789
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001735221738227446,
+      "loss": 0.6917,
+      "step": 6790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017351471625747018,
+      "loss": 0.787,
+      "step": 6791
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001735072578024226,
+      "loss": 0.8344,
+      "step": 6792
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001734997984576922,
+      "loss": 0.9273,
+      "step": 6793
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001734923382233692,
+      "loss": 0.7136,
+      "step": 6794
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017348487709954396,
+      "loss": 0.844,
+      "step": 6795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017347741508630672,
+      "loss": 0.8438,
+      "step": 6796
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017346995218374785,
+      "loss": 0.6584,
+      "step": 6797
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017346248839195764,
+      "loss": 0.8392,
+      "step": 6798
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017345502371102645,
+      "loss": 0.8022,
+      "step": 6799
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017344755814104458,
+      "loss": 0.8175,
+      "step": 6800
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 0.8164509534835815,
+      "eval_runtime": 849.136,
+      "eval_samples_per_second": 28.335,
+      "eval_steps_per_second": 14.167,
+      "step": 6800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017344009168210244,
+      "loss": 0.8752,
+      "step": 6801
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017343262433429038,
+      "loss": 0.8497,
+      "step": 6802
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001734251560976988,
+      "loss": 0.7803,
+      "step": 6803
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017341768697241803,
+      "loss": 0.8577,
+      "step": 6804
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017341021695853855,
+      "loss": 0.7277,
+      "step": 6805
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017340274605615071,
+      "loss": 0.7958,
+      "step": 6806
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000173395274265345,
+      "loss": 0.8015,
+      "step": 6807
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017338780158621177,
+      "loss": 0.7636,
+      "step": 6808
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017338032801884152,
+      "loss": 0.7798,
+      "step": 6809
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017337285356332468,
+      "loss": 0.8244,
+      "step": 6810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017336537821975175,
+      "loss": 0.904,
+      "step": 6811
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017335790198821317,
+      "loss": 0.7731,
+      "step": 6812
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017335042486879947,
+      "loss": 0.7706,
+      "step": 6813
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017334294686160106,
+      "loss": 0.764,
+      "step": 6814
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017333546796670856,
+      "loss": 0.8137,
+      "step": 6815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001733279881842124,
+      "loss": 0.8353,
+      "step": 6816
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017332050751420316,
+      "loss": 0.8381,
+      "step": 6817
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017331302595677143,
+      "loss": 0.8434,
+      "step": 6818
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017330554351200763,
+      "loss": 0.8586,
+      "step": 6819
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017329806018000237,
+      "loss": 0.7659,
+      "step": 6820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001732905759608463,
+      "loss": 0.763,
+      "step": 6821
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017328309085462996,
+      "loss": 0.8308,
+      "step": 6822
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001732756048614439,
+      "loss": 0.7754,
+      "step": 6823
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017326811798137876,
+      "loss": 0.8439,
+      "step": 6824
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017326063021452516,
+      "loss": 0.9119,
+      "step": 6825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017325314156097375,
+      "loss": 0.8095,
+      "step": 6826
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017324565202081512,
+      "loss": 0.846,
+      "step": 6827
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017323816159413994,
+      "loss": 0.7677,
+      "step": 6828
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017323067028103886,
+      "loss": 0.7464,
+      "step": 6829
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017322317808160257,
+      "loss": 0.8285,
+      "step": 6830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017321568499592175,
+      "loss": 0.699,
+      "step": 6831
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017320819102408706,
+      "loss": 0.8391,
+      "step": 6832
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017320069616618927,
+      "loss": 0.8589,
+      "step": 6833
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000173193200422319,
+      "loss": 0.8516,
+      "step": 6834
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017318570379256702,
+      "loss": 0.9883,
+      "step": 6835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017317820627702407,
+      "loss": 0.7786,
+      "step": 6836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017317070787578088,
+      "loss": 0.7792,
+      "step": 6837
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017316320858892822,
+      "loss": 0.8877,
+      "step": 6838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001731557084165569,
+      "loss": 0.8103,
+      "step": 6839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017314820735875758,
+      "loss": 0.8532,
+      "step": 6840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017314070541562114,
+      "loss": 0.8307,
+      "step": 6841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017313320258723836,
+      "loss": 0.7777,
+      "step": 6842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001731256988737,
+      "loss": 0.7682,
+      "step": 6843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017311819427509698,
+      "loss": 0.7664,
+      "step": 6844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017311068879152005,
+      "loss": 0.8743,
+      "step": 6845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017310318242306008,
+      "loss": 0.8591,
+      "step": 6846
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001730956751698079,
+      "loss": 0.8069,
+      "step": 6847
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001730881670318544,
+      "loss": 0.8012,
+      "step": 6848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017308065800929044,
+      "loss": 0.904,
+      "step": 6849
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017307314810220692,
+      "loss": 0.8393,
+      "step": 6850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001730656373106947,
+      "loss": 0.7432,
+      "step": 6851
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017305812563484473,
+      "loss": 0.7856,
+      "step": 6852
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017305061307474787,
+      "loss": 0.8762,
+      "step": 6853
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017304309963049513,
+      "loss": 0.8395,
+      "step": 6854
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017303558530217737,
+      "loss": 0.8343,
+      "step": 6855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017302807008988557,
+      "loss": 0.6931,
+      "step": 6856
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017302055399371068,
+      "loss": 0.8066,
+      "step": 6857
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001730130370137437,
+      "loss": 0.7998,
+      "step": 6858
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017300551915007558,
+      "loss": 0.8516,
+      "step": 6859
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001729980004027973,
+      "loss": 0.7751,
+      "step": 6860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017299048077199988,
+      "loss": 0.9001,
+      "step": 6861
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017298296025777433,
+      "loss": 0.7336,
+      "step": 6862
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017297543886021172,
+      "loss": 0.9154,
+      "step": 6863
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000172967916579403,
+      "loss": 0.8101,
+      "step": 6864
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017296039341543925,
+      "loss": 0.7918,
+      "step": 6865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017295286936841158,
+      "loss": 0.8825,
+      "step": 6866
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017294534443841096,
+      "loss": 0.7957,
+      "step": 6867
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017293781862552852,
+      "loss": 0.8065,
+      "step": 6868
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017293029192985536,
+      "loss": 0.9699,
+      "step": 6869
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017292276435148252,
+      "loss": 0.9437,
+      "step": 6870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017291523589050116,
+      "loss": 0.8593,
+      "step": 6871
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017290770654700239,
+      "loss": 0.74,
+      "step": 6872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017290017632107737,
+      "loss": 0.7223,
+      "step": 6873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017289264521281716,
+      "loss": 0.8642,
+      "step": 6874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000172885113222313,
+      "loss": 0.7916,
+      "step": 6875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017287758034965596,
+      "loss": 0.8477,
+      "step": 6876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001728700465949373,
+      "loss": 0.7859,
+      "step": 6877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017286251195824815,
+      "loss": 0.8113,
+      "step": 6878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017285497643967972,
+      "loss": 0.8827,
+      "step": 6879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017284744003932324,
+      "loss": 0.8248,
+      "step": 6880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001728399027572699,
+      "loss": 0.8621,
+      "step": 6881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001728323645936109,
+      "loss": 0.9558,
+      "step": 6882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017282482554843752,
+      "loss": 0.7467,
+      "step": 6883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000172817285621841,
+      "loss": 0.7973,
+      "step": 6884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001728097448139126,
+      "loss": 0.7873,
+      "step": 6885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017280220312474356,
+      "loss": 0.824,
+      "step": 6886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001727946605544252,
+      "loss": 0.8653,
+      "step": 6887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017278711710304876,
+      "loss": 0.8328,
+      "step": 6888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001727795727707056,
+      "loss": 0.8465,
+      "step": 6889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000172772027557487,
+      "loss": 0.8096,
+      "step": 6890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017276448146348426,
+      "loss": 0.8096,
+      "step": 6891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017275693448878877,
+      "loss": 0.8379,
+      "step": 6892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001727493866334918,
+      "loss": 0.885,
+      "step": 6893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001727418378976848,
+      "loss": 0.738,
+      "step": 6894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017273428828145902,
+      "loss": 0.8708,
+      "step": 6895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017272673778490592,
+      "loss": 0.9241,
+      "step": 6896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017271918640811689,
+      "loss": 0.796,
+      "step": 6897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001727116341511833,
+      "loss": 0.7384,
+      "step": 6898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001727040810141965,
+      "loss": 0.9413,
+      "step": 6899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000172696526997248,
+      "loss": 0.6889,
+      "step": 6900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017268897210042918,
+      "loss": 0.9435,
+      "step": 6901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001726814163238315,
+      "loss": 0.7786,
+      "step": 6902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017267385966754638,
+      "loss": 0.7649,
+      "step": 6903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017266630213166537,
+      "loss": 0.9152,
+      "step": 6904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017265874371627982,
+      "loss": 0.9413,
+      "step": 6905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017265118442148127,
+      "loss": 0.8126,
+      "step": 6906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017264362424736123,
+      "loss": 0.9434,
+      "step": 6907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017263606319401116,
+      "loss": 0.8001,
+      "step": 6908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001726285012615226,
+      "loss": 0.8758,
+      "step": 6909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001726209384499871,
+      "loss": 0.8398,
+      "step": 6910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017261337475949615,
+      "loss": 0.8126,
+      "step": 6911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001726058101901413,
+      "loss": 0.7612,
+      "step": 6912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001725982447420141,
+      "loss": 0.74,
+      "step": 6913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001725906784152062,
+      "loss": 0.7711,
+      "step": 6914
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017258311120980907,
+      "loss": 0.8679,
+      "step": 6915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017257554312591434,
+      "loss": 0.8005,
+      "step": 6916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017256797416361362,
+      "loss": 0.7584,
+      "step": 6917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017256040432299852,
+      "loss": 0.722,
+      "step": 6918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017255283360416063,
+      "loss": 0.8638,
+      "step": 6919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017254526200719163,
+      "loss": 0.75,
+      "step": 6920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017253768953218312,
+      "loss": 0.8538,
+      "step": 6921
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017253011617922675,
+      "loss": 0.7812,
+      "step": 6922
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017252254194841422,
+      "loss": 0.7651,
+      "step": 6923
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001725149668398372,
+      "loss": 0.8716,
+      "step": 6924
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017250739085358728,
+      "loss": 0.7705,
+      "step": 6925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017249981398975633,
+      "loss": 0.7583,
+      "step": 6926
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017249223624843588,
+      "loss": 0.8229,
+      "step": 6927
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017248465762971776,
+      "loss": 0.7951,
+      "step": 6928
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017247707813369364,
+      "loss": 0.796,
+      "step": 6929
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001724694977604553,
+      "loss": 0.7917,
+      "step": 6930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017246191651009444,
+      "loss": 0.8793,
+      "step": 6931
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017245433438270284,
+      "loss": 0.8816,
+      "step": 6932
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001724467513783723,
+      "loss": 0.7806,
+      "step": 6933
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017243916749719453,
+      "loss": 0.7327,
+      "step": 6934
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017243158273926138,
+      "loss": 0.8183,
+      "step": 6935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017242399710466465,
+      "loss": 0.9106,
+      "step": 6936
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017241641059349613,
+      "loss": 0.7354,
+      "step": 6937
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017240882320584763,
+      "loss": 0.7885,
+      "step": 6938
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000172401234941811,
+      "loss": 0.7684,
+      "step": 6939
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017239364580147806,
+      "loss": 0.8077,
+      "step": 6940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017238605578494073,
+      "loss": 0.7742,
+      "step": 6941
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001723784648922908,
+      "loss": 0.7818,
+      "step": 6942
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001723708731236202,
+      "loss": 0.7421,
+      "step": 6943
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017236328047902076,
+      "loss": 0.8204,
+      "step": 6944
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001723556869585844,
+      "loss": 0.8329,
+      "step": 6945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017234809256240307,
+      "loss": 0.7514,
+      "step": 6946
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017234049729056863,
+      "loss": 0.7954,
+      "step": 6947
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017233290114317305,
+      "loss": 0.7324,
+      "step": 6948
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017232530412030826,
+      "loss": 0.8576,
+      "step": 6949
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017231770622206613,
+      "loss": 0.723,
+      "step": 6950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017231010744853874,
+      "loss": 0.7776,
+      "step": 6951
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000172302507799818,
+      "loss": 0.9261,
+      "step": 6952
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001722949072759959,
+      "loss": 0.8975,
+      "step": 6953
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017228730587716442,
+      "loss": 0.8387,
+      "step": 6954
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017227970360341556,
+      "loss": 0.8128,
+      "step": 6955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017227210045484136,
+      "loss": 0.8127,
+      "step": 6956
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017226449643153386,
+      "loss": 0.7694,
+      "step": 6957
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017225689153358502,
+      "loss": 0.7617,
+      "step": 6958
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017224928576108694,
+      "loss": 0.738,
+      "step": 6959
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017224167911413166,
+      "loss": 0.8416,
+      "step": 6960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017223407159281127,
+      "loss": 0.7707,
+      "step": 6961
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001722264631972178,
+      "loss": 0.7967,
+      "step": 6962
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017221885392744337,
+      "loss": 0.8346,
+      "step": 6963
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017221124378358007,
+      "loss": 0.8443,
+      "step": 6964
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017220363276572003,
+      "loss": 0.8783,
+      "step": 6965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001721960208739553,
+      "loss": 0.8438,
+      "step": 6966
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017218840810837808,
+      "loss": 0.8016,
+      "step": 6967
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001721807944690805,
+      "loss": 0.7896,
+      "step": 6968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001721731799561547,
+      "loss": 0.8922,
+      "step": 6969
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017216556456969278,
+      "loss": 0.9415,
+      "step": 6970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017215794830978702,
+      "loss": 0.7448,
+      "step": 6971
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017215033117652954,
+      "loss": 0.8034,
+      "step": 6972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017214271317001254,
+      "loss": 0.7798,
+      "step": 6973
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017213509429032825,
+      "loss": 0.8319,
+      "step": 6974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001721274745375688,
+      "loss": 0.8264,
+      "step": 6975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017211985391182654,
+      "loss": 0.7542,
+      "step": 6976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001721122324131936,
+      "loss": 0.8495,
+      "step": 6977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001721046100417623,
+      "loss": 0.7906,
+      "step": 6978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017209698679762481,
+      "loss": 0.9877,
+      "step": 6979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001720893626808735,
+      "loss": 0.735,
+      "step": 6980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017208173769160058,
+      "loss": 0.7923,
+      "step": 6981
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017207411182989832,
+      "loss": 0.8158,
+      "step": 6982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001720664850958591,
+      "loss": 0.8282,
+      "step": 6983
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017205885748957514,
+      "loss": 0.9135,
+      "step": 6984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017205122901113883,
+      "loss": 0.7649,
+      "step": 6985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017204359966064244,
+      "loss": 0.6536,
+      "step": 6986
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017203596943817838,
+      "loss": 0.8254,
+      "step": 6987
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017202833834383894,
+      "loss": 0.8228,
+      "step": 6988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001720207063777165,
+      "loss": 0.7044,
+      "step": 6989
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017201307353990344,
+      "loss": 0.889,
+      "step": 6990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001720054398304921,
+      "loss": 0.7484,
+      "step": 6991
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017199780524957496,
+      "loss": 0.8798,
+      "step": 6992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017199016979724436,
+      "loss": 0.9185,
+      "step": 6993
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017198253347359271,
+      "loss": 0.8798,
+      "step": 6994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001719748962787125,
+      "loss": 0.8209,
+      "step": 6995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001719672582126961,
+      "loss": 0.7987,
+      "step": 6996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017195961927563596,
+      "loss": 0.7978,
+      "step": 6997
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017195197946762457,
+      "loss": 0.7385,
+      "step": 6998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001719443387887544,
+      "loss": 0.76,
+      "step": 6999
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017193669723911787,
+      "loss": 0.7169,
+      "step": 7000
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 0.8066107630729675,
+      "eval_runtime": 853.5766,
+      "eval_samples_per_second": 28.187,
+      "eval_steps_per_second": 14.094,
+      "step": 7000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017192905481880753,
+      "loss": 0.7387,
+      "step": 7001
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017192141152791584,
+      "loss": 0.8576,
+      "step": 7002
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017191376736653532,
+      "loss": 0.909,
+      "step": 7003
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017190612233475852,
+      "loss": 0.7323,
+      "step": 7004
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017189847643267792,
+      "loss": 0.8358,
+      "step": 7005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001718908296603861,
+      "loss": 0.7068,
+      "step": 7006
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017188318201797557,
+      "loss": 0.9094,
+      "step": 7007
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017187553350553894,
+      "loss": 0.7086,
+      "step": 7008
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017186788412316877,
+      "loss": 0.8237,
+      "step": 7009
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017186023387095762,
+      "loss": 0.7435,
+      "step": 7010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017185258274899813,
+      "loss": 0.7312,
+      "step": 7011
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001718449307573828,
+      "loss": 0.8339,
+      "step": 7012
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017183727789620438,
+      "loss": 0.8423,
+      "step": 7013
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001718296241655554,
+      "loss": 0.9338,
+      "step": 7014
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017182196956552854,
+      "loss": 0.7911,
+      "step": 7015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017181431409621644,
+      "loss": 0.7694,
+      "step": 7016
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017180665775771173,
+      "loss": 0.8588,
+      "step": 7017
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017179900055010712,
+      "loss": 0.7472,
+      "step": 7018
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017179134247349525,
+      "loss": 0.7692,
+      "step": 7019
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017178368352796882,
+      "loss": 0.8783,
+      "step": 7020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017177602371362056,
+      "loss": 0.8254,
+      "step": 7021
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001717683630305431,
+      "loss": 0.8545,
+      "step": 7022
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017176070147882926,
+      "loss": 0.8083,
+      "step": 7023
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001717530390585717,
+      "loss": 0.8239,
+      "step": 7024
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017174537576986317,
+      "loss": 0.73,
+      "step": 7025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017173771161279645,
+      "loss": 0.7928,
+      "step": 7026
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017173004658746428,
+      "loss": 0.8106,
+      "step": 7027
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017172238069395945,
+      "loss": 0.8875,
+      "step": 7028
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017171471393237473,
+      "loss": 0.7108,
+      "step": 7029
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001717070463028029,
+      "loss": 0.7751,
+      "step": 7030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017169937780533678,
+      "loss": 0.7476,
+      "step": 7031
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001716917084400692,
+      "loss": 0.7108,
+      "step": 7032
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017168403820709294,
+      "loss": 0.7109,
+      "step": 7033
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017167636710650088,
+      "loss": 0.7862,
+      "step": 7034
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017166869513838583,
+      "loss": 0.8964,
+      "step": 7035
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017166102230284068,
+      "loss": 0.8066,
+      "step": 7036
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001716533485999583,
+      "loss": 0.7645,
+      "step": 7037
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017164567402983152,
+      "loss": 0.8183,
+      "step": 7038
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017163799859255328,
+      "loss": 0.7759,
+      "step": 7039
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017163032228821644,
+      "loss": 0.8529,
+      "step": 7040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017162264511691393,
+      "loss": 0.8232,
+      "step": 7041
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017161496707873865,
+      "loss": 0.7431,
+      "step": 7042
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017160728817378355,
+      "loss": 0.7931,
+      "step": 7043
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017159960840214158,
+      "loss": 0.7886,
+      "step": 7044
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017159192776390568,
+      "loss": 0.9154,
+      "step": 7045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017158424625916875,
+      "loss": 0.7847,
+      "step": 7046
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017157656388802387,
+      "loss": 0.8057,
+      "step": 7047
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017156888065056396,
+      "loss": 0.7928,
+      "step": 7048
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017156119654688202,
+      "loss": 0.8104,
+      "step": 7049
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017155351157707106,
+      "loss": 0.75,
+      "step": 7050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017154582574122407,
+      "loss": 0.7772,
+      "step": 7051
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017153813903943412,
+      "loss": 0.7086,
+      "step": 7052
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017153045147179417,
+      "loss": 0.874,
+      "step": 7053
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017152276303839735,
+      "loss": 0.8546,
+      "step": 7054
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001715150737393367,
+      "loss": 0.8254,
+      "step": 7055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017150738357470525,
+      "loss": 0.8585,
+      "step": 7056
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001714996925445961,
+      "loss": 0.8256,
+      "step": 7057
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001714920006491023,
+      "loss": 0.7576,
+      "step": 7058
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000171484307888317,
+      "loss": 0.8075,
+      "step": 7059
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017147661426233327,
+      "loss": 0.7856,
+      "step": 7060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017146891977124428,
+      "loss": 0.7491,
+      "step": 7061
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001714612244151431,
+      "loss": 0.8037,
+      "step": 7062
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001714535281941229,
+      "loss": 0.6901,
+      "step": 7063
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001714458311082768,
+      "loss": 0.7877,
+      "step": 7064
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000171438133157698,
+      "loss": 0.8534,
+      "step": 7065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017143043434247963,
+      "loss": 0.8457,
+      "step": 7066
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001714227346627149,
+      "loss": 0.7429,
+      "step": 7067
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017141503411849702,
+      "loss": 0.8387,
+      "step": 7068
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017140733270991915,
+      "loss": 0.8086,
+      "step": 7069
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017139963043707454,
+      "loss": 0.7115,
+      "step": 7070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001713919273000564,
+      "loss": 0.8393,
+      "step": 7071
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017138422329895795,
+      "loss": 0.8303,
+      "step": 7072
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017137651843387238,
+      "loss": 0.7551,
+      "step": 7073
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017136881270489308,
+      "loss": 0.7701,
+      "step": 7074
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017136110611211324,
+      "loss": 0.9159,
+      "step": 7075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017135339865562613,
+      "loss": 0.8095,
+      "step": 7076
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000171345690335525,
+      "loss": 0.8717,
+      "step": 7077
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017133798115190324,
+      "loss": 0.76,
+      "step": 7078
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001713302711048541,
+      "loss": 0.7348,
+      "step": 7079
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017132256019447087,
+      "loss": 0.838,
+      "step": 7080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017131484842084694,
+      "loss": 0.8434,
+      "step": 7081
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017130713578407563,
+      "loss": 0.9242,
+      "step": 7082
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017129942228425024,
+      "loss": 0.8833,
+      "step": 7083
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017129170792146422,
+      "loss": 0.8214,
+      "step": 7084
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017128399269581083,
+      "loss": 0.8005,
+      "step": 7085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017127627660738355,
+      "loss": 0.9315,
+      "step": 7086
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001712685596562757,
+      "loss": 0.9651,
+      "step": 7087
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017126084184258074,
+      "loss": 0.7199,
+      "step": 7088
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017125312316639202,
+      "loss": 0.8863,
+      "step": 7089
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017124540362780302,
+      "loss": 0.7892,
+      "step": 7090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001712376832269071,
+      "loss": 0.8175,
+      "step": 7091
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017122996196379777,
+      "loss": 0.8226,
+      "step": 7092
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017122223983856847,
+      "loss": 0.8255,
+      "step": 7093
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017121451685131263,
+      "loss": 1.0177,
+      "step": 7094
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017120679300212374,
+      "loss": 0.7962,
+      "step": 7095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001711990682910953,
+      "loss": 0.8514,
+      "step": 7096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001711913427183208,
+      "loss": 0.7546,
+      "step": 7097
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017118361628389375,
+      "loss": 0.8856,
+      "step": 7098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017117588898790765,
+      "loss": 0.7801,
+      "step": 7099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017116816083045602,
+      "loss": 0.7635,
+      "step": 7100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017116043181163244,
+      "loss": 0.668,
+      "step": 7101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001711527019315304,
+      "loss": 0.8142,
+      "step": 7102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001711449711902435,
+      "loss": 0.835,
+      "step": 7103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017113723958786525,
+      "loss": 0.8001,
+      "step": 7104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001711295071244893,
+      "loss": 0.7708,
+      "step": 7105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001711217738002092,
+      "loss": 0.852,
+      "step": 7106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017111403961511855,
+      "loss": 0.7145,
+      "step": 7107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017110630456931096,
+      "loss": 0.8106,
+      "step": 7108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001710985686628801,
+      "loss": 0.8076,
+      "step": 7109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001710908318959195,
+      "loss": 0.7887,
+      "step": 7110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017108309426852288,
+      "loss": 0.7953,
+      "step": 7111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017107535578078387,
+      "loss": 0.86,
+      "step": 7112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017106761643279615,
+      "loss": 0.7498,
+      "step": 7113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017105987622465334,
+      "loss": 0.7957,
+      "step": 7114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017105213515644914,
+      "loss": 0.8328,
+      "step": 7115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001710443932282773,
+      "loss": 0.8935,
+      "step": 7116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017103665044023144,
+      "loss": 0.8466,
+      "step": 7117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017102890679240534,
+      "loss": 0.8222,
+      "step": 7118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017102116228489268,
+      "loss": 0.8372,
+      "step": 7119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001710134169177872,
+      "loss": 0.7505,
+      "step": 7120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017100567069118267,
+      "loss": 0.8396,
+      "step": 7121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017099792360517285,
+      "loss": 0.8166,
+      "step": 7122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017099017565985145,
+      "loss": 0.8259,
+      "step": 7123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017098242685531232,
+      "loss": 0.8737,
+      "step": 7124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001709746771916492,
+      "loss": 0.8615,
+      "step": 7125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001709669266689559,
+      "loss": 0.7569,
+      "step": 7126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017095917528732617,
+      "loss": 0.8431,
+      "step": 7127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017095142304685392,
+      "loss": 0.8223,
+      "step": 7128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017094366994763294,
+      "loss": 0.7952,
+      "step": 7129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017093591598975707,
+      "loss": 0.7524,
+      "step": 7130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017092816117332015,
+      "loss": 0.7984,
+      "step": 7131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000170920405498416,
+      "loss": 0.8189,
+      "step": 7132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001709126489651386,
+      "loss": 0.6775,
+      "step": 7133
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017090489157358177,
+      "loss": 0.8587,
+      "step": 7134
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017089713332383934,
+      "loss": 0.649,
+      "step": 7135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001708893742160053,
+      "loss": 0.7624,
+      "step": 7136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017088161425017352,
+      "loss": 0.7758,
+      "step": 7137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017087385342643788,
+      "loss": 0.6952,
+      "step": 7138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001708660917448924,
+      "loss": 0.8547,
+      "step": 7139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017085832920563099,
+      "loss": 0.8605,
+      "step": 7140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001708505658087476,
+      "loss": 0.7462,
+      "step": 7141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017084280155433616,
+      "loss": 0.7694,
+      "step": 7142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001708350364424907,
+      "loss": 0.8748,
+      "step": 7143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017082727047330515,
+      "loss": 0.8321,
+      "step": 7144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017081950364687353,
+      "loss": 0.8335,
+      "step": 7145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017081173596328986,
+      "loss": 0.7808,
+      "step": 7146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017080396742264812,
+      "loss": 0.826,
+      "step": 7147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017079619802504238,
+      "loss": 0.7018,
+      "step": 7148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017078842777056661,
+      "loss": 0.8703,
+      "step": 7149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001707806566593149,
+      "loss": 0.8555,
+      "step": 7150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001707728846913813,
+      "loss": 0.9124,
+      "step": 7151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017076511186685987,
+      "loss": 0.8646,
+      "step": 7152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017075733818584473,
+      "loss": 0.9346,
+      "step": 7153
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017074956364842995,
+      "loss": 0.7698,
+      "step": 7154
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017074178825470955,
+      "loss": 0.8239,
+      "step": 7155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017073401200477773,
+      "loss": 0.7831,
+      "step": 7156
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017072623489872857,
+      "loss": 0.6898,
+      "step": 7157
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001707184569366562,
+      "loss": 0.8475,
+      "step": 7158
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017071067811865476,
+      "loss": 0.7866,
+      "step": 7159
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001707028984448184,
+      "loss": 0.938,
+      "step": 7160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001706951179152413,
+      "loss": 0.8097,
+      "step": 7161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017068733653001763,
+      "loss": 0.9402,
+      "step": 7162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001706795542892415,
+      "loss": 0.8173,
+      "step": 7163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017067177119300722,
+      "loss": 0.7454,
+      "step": 7164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017066398724140892,
+      "loss": 0.8379,
+      "step": 7165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001706562024345408,
+      "loss": 0.799,
+      "step": 7166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001706484167724971,
+      "loss": 0.8174,
+      "step": 7167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017064063025537204,
+      "loss": 0.7734,
+      "step": 7168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001706328428832599,
+      "loss": 0.8946,
+      "step": 7169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017062505465625488,
+      "loss": 0.9525,
+      "step": 7170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001706172655744513,
+      "loss": 0.792,
+      "step": 7171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001706094756379434,
+      "loss": 0.8987,
+      "step": 7172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017060168484682548,
+      "loss": 0.7529,
+      "step": 7173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017059389320119183,
+      "loss": 0.8688,
+      "step": 7174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017058610070113677,
+      "loss": 0.76,
+      "step": 7175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017057830734675455,
+      "loss": 0.8064,
+      "step": 7176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017057051313813955,
+      "loss": 0.8615,
+      "step": 7177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001705627180753861,
+      "loss": 0.7769,
+      "step": 7178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017055492215858857,
+      "loss": 0.8566,
+      "step": 7179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017054712538784128,
+      "loss": 0.835,
+      "step": 7180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001705393277632386,
+      "loss": 0.7105,
+      "step": 7181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001705315292848749,
+      "loss": 0.836,
+      "step": 7182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001705237299528446,
+      "loss": 0.8668,
+      "step": 7183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017051592976724208,
+      "loss": 0.7474,
+      "step": 7184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017050812872816174,
+      "loss": 0.8603,
+      "step": 7185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017050032683569804,
+      "loss": 0.8461,
+      "step": 7186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017049252408994533,
+      "loss": 0.7453,
+      "step": 7187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017048472049099812,
+      "loss": 0.9199,
+      "step": 7188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017047691603895083,
+      "loss": 0.8585,
+      "step": 7189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017046911073389793,
+      "loss": 0.8488,
+      "step": 7190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017046130457593387,
+      "loss": 0.8571,
+      "step": 7191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017045349756515313,
+      "loss": 0.7368,
+      "step": 7192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017044568970165028,
+      "loss": 0.7968,
+      "step": 7193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001704378809855197,
+      "loss": 0.808,
+      "step": 7194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000170430071416856,
+      "loss": 0.6645,
+      "step": 7195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017042226099575362,
+      "loss": 0.7993,
+      "step": 7196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017041444972230716,
+      "loss": 0.7902,
+      "step": 7197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017040663759661112,
+      "loss": 0.7142,
+      "step": 7198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001703988246187601,
+      "loss": 0.7632,
+      "step": 7199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001703910107888486,
+      "loss": 0.6899,
+      "step": 7200
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 0.7952414155006409,
+      "eval_runtime": 852.6819,
+      "eval_samples_per_second": 28.217,
+      "eval_steps_per_second": 14.108,
+      "step": 7200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001703831961069712,
+      "loss": 0.7821,
+      "step": 7201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017037538057322257,
+      "loss": 0.7557,
+      "step": 7202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001703675641876972,
+      "loss": 0.7409,
+      "step": 7203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017035974695048974,
+      "loss": 0.8735,
+      "step": 7204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017035192886169482,
+      "loss": 0.7753,
+      "step": 7205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017034410992140703,
+      "loss": 0.7748,
+      "step": 7206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017033629012972102,
+      "loss": 0.8233,
+      "step": 7207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017032846948673145,
+      "loss": 0.7385,
+      "step": 7208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017032064799253296,
+      "loss": 0.7955,
+      "step": 7209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001703128256472202,
+      "loss": 0.7851,
+      "step": 7210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017030500245088787,
+      "loss": 0.706,
+      "step": 7211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017029717840363068,
+      "loss": 0.8445,
+      "step": 7212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001702893535055433,
+      "loss": 0.9195,
+      "step": 7213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017028152775672042,
+      "loss": 0.8525,
+      "step": 7214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017027370115725674,
+      "loss": 0.645,
+      "step": 7215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017026587370724708,
+      "loss": 0.8119,
+      "step": 7216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001702580454067861,
+      "loss": 0.6696,
+      "step": 7217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017025021625596853,
+      "loss": 0.841,
+      "step": 7218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001702423862548892,
+      "loss": 0.7306,
+      "step": 7219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017023455540364285,
+      "loss": 0.7915,
+      "step": 7220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017022672370232427,
+      "loss": 0.6615,
+      "step": 7221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017021889115102822,
+      "loss": 0.8118,
+      "step": 7222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001702110577498495,
+      "loss": 0.8334,
+      "step": 7223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017020322349888299,
+      "loss": 0.8095,
+      "step": 7224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017019538839822338,
+      "loss": 0.6549,
+      "step": 7225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017018755244796566,
+      "loss": 0.8695,
+      "step": 7226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017017971564820453,
+      "loss": 0.7737,
+      "step": 7227
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017017187799903488,
+      "loss": 0.7727,
+      "step": 7228
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017016403950055164,
+      "loss": 0.8099,
+      "step": 7229
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017015620015284962,
+      "loss": 0.9062,
+      "step": 7230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001701483599560237,
+      "loss": 0.8106,
+      "step": 7231
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017014051891016884,
+      "loss": 0.7718,
+      "step": 7232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017013267701537985,
+      "loss": 0.7874,
+      "step": 7233
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017012483427175168,
+      "loss": 0.8436,
+      "step": 7234
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017011699067937927,
+      "loss": 0.7371,
+      "step": 7235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017010914623835754,
+      "loss": 0.8047,
+      "step": 7236
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017010130094878142,
+      "loss": 0.7858,
+      "step": 7237
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017009345481074594,
+      "loss": 0.7948,
+      "step": 7238
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017008560782434592,
+      "loss": 0.7531,
+      "step": 7239
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017007775998967647,
+      "loss": 0.8326,
+      "step": 7240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017006991130683253,
+      "loss": 0.7869,
+      "step": 7241
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017006206177590905,
+      "loss": 0.7366,
+      "step": 7242
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017005421139700108,
+      "loss": 0.8813,
+      "step": 7243
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017004636017020364,
+      "loss": 0.7968,
+      "step": 7244
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017003850809561177,
+      "loss": 0.7626,
+      "step": 7245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017003065517332043,
+      "loss": 0.8162,
+      "step": 7246
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017002280140342477,
+      "loss": 0.8657,
+      "step": 7247
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001700149467860198,
+      "loss": 0.7973,
+      "step": 7248
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017000709132120053,
+      "loss": 0.8556,
+      "step": 7249
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016999923500906212,
+      "loss": 0.835,
+      "step": 7250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016999137784969964,
+      "loss": 0.767,
+      "step": 7251
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016998351984320813,
+      "loss": 0.8744,
+      "step": 7252
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001699756609896828,
+      "loss": 0.8071,
+      "step": 7253
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001699678012892187,
+      "loss": 0.6841,
+      "step": 7254
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016995994074191093,
+      "loss": 0.653,
+      "step": 7255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001699520793478547,
+      "loss": 0.7228,
+      "step": 7256
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001699442171071451,
+      "loss": 0.8409,
+      "step": 7257
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016993635401987739,
+      "loss": 0.9139,
+      "step": 7258
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001699284900861466,
+      "loss": 0.7895,
+      "step": 7259
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000169920625306048,
+      "loss": 0.7643,
+      "step": 7260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016991275967967676,
+      "loss": 0.833,
+      "step": 7261
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016990489320712804,
+      "loss": 0.8108,
+      "step": 7262
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016989702588849712,
+      "loss": 0.769,
+      "step": 7263
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016988915772387918,
+      "loss": 0.6435,
+      "step": 7264
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001698812887133695,
+      "loss": 0.7948,
+      "step": 7265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001698734188570632,
+      "loss": 0.7641,
+      "step": 7266
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001698655481550557,
+      "loss": 0.8182,
+      "step": 7267
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001698576766074421,
+      "loss": 0.8789,
+      "step": 7268
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016984980421431777,
+      "loss": 0.7876,
+      "step": 7269
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000169841930975778,
+      "loss": 0.9361,
+      "step": 7270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016983405689191802,
+      "loss": 0.9471,
+      "step": 7271
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016982618196283316,
+      "loss": 0.8111,
+      "step": 7272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016981830618861875,
+      "loss": 0.76,
+      "step": 7273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001698104295693701,
+      "loss": 0.8149,
+      "step": 7274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016980255210518252,
+      "loss": 0.9028,
+      "step": 7275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016979467379615138,
+      "loss": 0.7428,
+      "step": 7276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001697867946423721,
+      "loss": 0.7478,
+      "step": 7277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016977891464393986,
+      "loss": 0.7859,
+      "step": 7278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016977103380095022,
+      "loss": 0.8093,
+      "step": 7279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001697631521134985,
+      "loss": 0.7356,
+      "step": 7280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016975526958168003,
+      "loss": 0.9204,
+      "step": 7281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001697473862055903,
+      "loss": 0.7984,
+      "step": 7282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016973950198532472,
+      "loss": 0.879,
+      "step": 7283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001697316169209787,
+      "loss": 0.6689,
+      "step": 7284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016972373101264765,
+      "loss": 0.8115,
+      "step": 7285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016971584426042703,
+      "loss": 0.8332,
+      "step": 7286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016970795666441227,
+      "loss": 0.7822,
+      "step": 7287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016970006822469893,
+      "loss": 0.8281,
+      "step": 7288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001696921789413824,
+      "loss": 0.807,
+      "step": 7289
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016968428881455815,
+      "loss": 0.9184,
+      "step": 7290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016967639784432176,
+      "loss": 0.9638,
+      "step": 7291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016966850603076868,
+      "loss": 0.8274,
+      "step": 7292
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016966061337399444,
+      "loss": 0.6858,
+      "step": 7293
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016965271987409456,
+      "loss": 0.8413,
+      "step": 7294
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016964482553116458,
+      "loss": 0.8427,
+      "step": 7295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001696369303453001,
+      "loss": 0.7929,
+      "step": 7296
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016962903431659657,
+      "loss": 0.793,
+      "step": 7297
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016962113744514964,
+      "loss": 0.8075,
+      "step": 7298
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001696132397310549,
+      "loss": 0.8093,
+      "step": 7299
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001696053411744079,
+      "loss": 0.7715,
+      "step": 7300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001695974417753042,
+      "loss": 0.7443,
+      "step": 7301
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001695895415338395,
+      "loss": 0.8281,
+      "step": 7302
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016958164045010934,
+      "loss": 0.7539,
+      "step": 7303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001695737385242094,
+      "loss": 0.6985,
+      "step": 7304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016956583575623532,
+      "loss": 0.7944,
+      "step": 7305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016955793214628272,
+      "loss": 0.8576,
+      "step": 7306
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016955002769444727,
+      "loss": 0.8166,
+      "step": 7307
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016954212240082465,
+      "loss": 0.7483,
+      "step": 7308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001695342162655105,
+      "loss": 0.8008,
+      "step": 7309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001695263092886006,
+      "loss": 0.7049,
+      "step": 7310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016951840147019057,
+      "loss": 0.7973,
+      "step": 7311
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016951049281037613,
+      "loss": 0.7352,
+      "step": 7312
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016950258330925304,
+      "loss": 0.7222,
+      "step": 7313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016949467296691702,
+      "loss": 0.6898,
+      "step": 7314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016948676178346375,
+      "loss": 0.9012,
+      "step": 7315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016947884975898906,
+      "loss": 0.7468,
+      "step": 7316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001694709368935887,
+      "loss": 0.8363,
+      "step": 7317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001694630231873584,
+      "loss": 0.7715,
+      "step": 7318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000169455108640394,
+      "loss": 0.7165,
+      "step": 7319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016944719325279124,
+      "loss": 0.8042,
+      "step": 7320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016943927702464591,
+      "loss": 0.8787,
+      "step": 7321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016943135995605392,
+      "loss": 0.8316,
+      "step": 7322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016942344204711097,
+      "loss": 0.7241,
+      "step": 7323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016941552329791302,
+      "loss": 0.6197,
+      "step": 7324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016940760370855577,
+      "loss": 0.8205,
+      "step": 7325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001693996832791352,
+      "loss": 0.7191,
+      "step": 7326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016939176200974712,
+      "loss": 0.8564,
+      "step": 7327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001693838399004874,
+      "loss": 0.7739,
+      "step": 7328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001693759169514519,
+      "loss": 0.89,
+      "step": 7329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016936799316273656,
+      "loss": 0.748,
+      "step": 7330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016936006853443729,
+      "loss": 0.8737,
+      "step": 7331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016935214306664994,
+      "loss": 0.8304,
+      "step": 7332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016934421675947052,
+      "loss": 0.8113,
+      "step": 7333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001693362896129949,
+      "loss": 0.7795,
+      "step": 7334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016932836162731905,
+      "loss": 0.8415,
+      "step": 7335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001693204328025389,
+      "loss": 0.763,
+      "step": 7336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001693125031387505,
+      "loss": 0.7825,
+      "step": 7337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001693045726360497,
+      "loss": 0.7228,
+      "step": 7338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016929664129453257,
+      "loss": 0.8089,
+      "step": 7339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016928870911429508,
+      "loss": 0.8743,
+      "step": 7340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016928077609543323,
+      "loss": 0.8532,
+      "step": 7341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016927284223804305,
+      "loss": 0.9407,
+      "step": 7342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016926490754222057,
+      "loss": 0.7854,
+      "step": 7343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001692569720080618,
+      "loss": 0.8469,
+      "step": 7344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016924903563566283,
+      "loss": 0.8365,
+      "step": 7345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016924109842511969,
+      "loss": 0.7684,
+      "step": 7346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016923316037652843,
+      "loss": 0.7386,
+      "step": 7347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016922522148998516,
+      "loss": 0.7671,
+      "step": 7348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016921728176558592,
+      "loss": 0.7505,
+      "step": 7349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001692093412034269,
+      "loss": 0.7636,
+      "step": 7350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016920139980360408,
+      "loss": 0.7929,
+      "step": 7351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016919345756621367,
+      "loss": 0.7614,
+      "step": 7352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016918551449135176,
+      "loss": 0.8027,
+      "step": 7353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001691775705791145,
+      "loss": 0.8319,
+      "step": 7354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016916962582959804,
+      "loss": 0.8747,
+      "step": 7355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016916168024289855,
+      "loss": 0.8454,
+      "step": 7356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016915373381911218,
+      "loss": 0.8377,
+      "step": 7357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016914578655833511,
+      "loss": 0.8205,
+      "step": 7358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001691378384606635,
+      "loss": 0.7997,
+      "step": 7359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001691298895261936,
+      "loss": 0.8033,
+      "step": 7360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001691219397550216,
+      "loss": 0.856,
+      "step": 7361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001691139891472437,
+      "loss": 0.9995,
+      "step": 7362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016910603770295613,
+      "loss": 0.7402,
+      "step": 7363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001690980854222552,
+      "loss": 0.7534,
+      "step": 7364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016909013230523705,
+      "loss": 0.9687,
+      "step": 7365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000169082178351998,
+      "loss": 0.8143,
+      "step": 7366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016907422356263432,
+      "loss": 0.8357,
+      "step": 7367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016906626793724224,
+      "loss": 0.8036,
+      "step": 7368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016905831147591814,
+      "loss": 0.782,
+      "step": 7369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016905035417875825,
+      "loss": 0.8012,
+      "step": 7370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001690423960458589,
+      "loss": 0.8479,
+      "step": 7371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001690344370773164,
+      "loss": 0.753,
+      "step": 7372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016902647727322712,
+      "loss": 0.6567,
+      "step": 7373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016901851663368733,
+      "loss": 0.9717,
+      "step": 7374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016901055515879346,
+      "loss": 0.7626,
+      "step": 7375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016900259284864178,
+      "loss": 0.694,
+      "step": 7376
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016899462970332873,
+      "loss": 0.7815,
+      "step": 7377
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016898666572295068,
+      "loss": 0.7883,
+      "step": 7378
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000168978700907604,
+      "loss": 0.6531,
+      "step": 7379
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001689707352573851,
+      "loss": 0.7166,
+      "step": 7380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016896276877239038,
+      "loss": 0.7047,
+      "step": 7381
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001689548014527163,
+      "loss": 0.7891,
+      "step": 7382
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016894683329845927,
+      "loss": 0.6982,
+      "step": 7383
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016893886430971572,
+      "loss": 0.8448,
+      "step": 7384
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016893089448658208,
+      "loss": 0.8864,
+      "step": 7385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001689229238291549,
+      "loss": 0.761,
+      "step": 7386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016891495233753055,
+      "loss": 0.8774,
+      "step": 7387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016890698001180554,
+      "loss": 0.757,
+      "step": 7388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016889900685207638,
+      "loss": 0.7312,
+      "step": 7389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016889103285843958,
+      "loss": 0.882,
+      "step": 7390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016888305803099162,
+      "loss": 0.7518,
+      "step": 7391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016887508236982908,
+      "loss": 0.7335,
+      "step": 7392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001688671058750484,
+      "loss": 0.7824,
+      "step": 7393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016885912854674618,
+      "loss": 0.8038,
+      "step": 7394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016885115038501903,
+      "loss": 0.772,
+      "step": 7395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016884317138996337,
+      "loss": 0.7633,
+      "step": 7396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016883519156167588,
+      "loss": 0.6851,
+      "step": 7397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016882721090025313,
+      "loss": 0.7691,
+      "step": 7398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016881922940579166,
+      "loss": 0.8617,
+      "step": 7399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016881124707838815,
+      "loss": 0.8256,
+      "step": 7400
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 0.7859690189361572,
+      "eval_runtime": 854.0764,
+      "eval_samples_per_second": 28.171,
+      "eval_steps_per_second": 14.085,
+      "step": 7400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016880326391813916,
+      "loss": 0.7026,
+      "step": 7401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016879527992514132,
+      "loss": 0.7938,
+      "step": 7402
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016878729509949125,
+      "loss": 0.7414,
+      "step": 7403
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016877930944128564,
+      "loss": 0.7811,
+      "step": 7404
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016877132295062107,
+      "loss": 0.8011,
+      "step": 7405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001687633356275943,
+      "loss": 0.7561,
+      "step": 7406
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016875534747230191,
+      "loss": 0.7187,
+      "step": 7407
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016874735848484065,
+      "loss": 0.8162,
+      "step": 7408
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016873936866530716,
+      "loss": 0.8658,
+      "step": 7409
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001687313780137982,
+      "loss": 0.6558,
+      "step": 7410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001687233865304104,
+      "loss": 0.7106,
+      "step": 7411
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016871539421524058,
+      "loss": 0.8774,
+      "step": 7412
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016870740106838544,
+      "loss": 0.8686,
+      "step": 7413
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001686994070899417,
+      "loss": 0.8904,
+      "step": 7414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016869141228000612,
+      "loss": 0.7156,
+      "step": 7415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016868341663867548,
+      "loss": 0.8118,
+      "step": 7416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016867542016604653,
+      "loss": 0.7603,
+      "step": 7417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016866742286221608,
+      "loss": 0.8108,
+      "step": 7418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016865942472728095,
+      "loss": 0.8116,
+      "step": 7419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016865142576133786,
+      "loss": 0.8306,
+      "step": 7420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001686434259644837,
+      "loss": 0.785,
+      "step": 7421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016863542533681523,
+      "loss": 0.8692,
+      "step": 7422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016862742387842934,
+      "loss": 0.8527,
+      "step": 7423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016861942158942288,
+      "loss": 0.7467,
+      "step": 7424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016861141846989265,
+      "loss": 0.6885,
+      "step": 7425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016860341451993555,
+      "loss": 0.7528,
+      "step": 7426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016859540973964844,
+      "loss": 0.782,
+      "step": 7427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016858740412912826,
+      "loss": 0.7716,
+      "step": 7428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001685793976884718,
+      "loss": 0.7919,
+      "step": 7429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016857139041777605,
+      "loss": 0.9108,
+      "step": 7430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016856338231713787,
+      "loss": 0.8553,
+      "step": 7431
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016855537338665423,
+      "loss": 0.764,
+      "step": 7432
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016854736362642206,
+      "loss": 0.7505,
+      "step": 7433
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016853935303653828,
+      "loss": 0.7736,
+      "step": 7434
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016853134161709985,
+      "loss": 0.736,
+      "step": 7435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016852332936820376,
+      "loss": 0.7745,
+      "step": 7436
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016851531628994695,
+      "loss": 0.7782,
+      "step": 7437
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001685073023824264,
+      "loss": 0.7378,
+      "step": 7438
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016849928764573916,
+      "loss": 0.8615,
+      "step": 7439
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016849127207998218,
+      "loss": 0.7963,
+      "step": 7440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016848325568525252,
+      "loss": 0.7828,
+      "step": 7441
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016847523846164717,
+      "loss": 0.729,
+      "step": 7442
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016846722040926316,
+      "loss": 0.8349,
+      "step": 7443
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016845920152819757,
+      "loss": 0.6862,
+      "step": 7444
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016845118181854743,
+      "loss": 0.8286,
+      "step": 7445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001684431612804098,
+      "loss": 0.7558,
+      "step": 7446
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016843513991388176,
+      "loss": 0.7397,
+      "step": 7447
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016842711771906043,
+      "loss": 0.9133,
+      "step": 7448
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016841909469604284,
+      "loss": 0.8635,
+      "step": 7449
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016841107084492618,
+      "loss": 0.8242,
+      "step": 7450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001684030461658075,
+      "loss": 0.7452,
+      "step": 7451
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016839502065878393,
+      "loss": 0.8797,
+      "step": 7452
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016838699432395263,
+      "loss": 0.6886,
+      "step": 7453
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001683789671614107,
+      "loss": 0.8321,
+      "step": 7454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001683709391712554,
+      "loss": 0.8259,
+      "step": 7455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016836291035358375,
+      "loss": 0.7692,
+      "step": 7456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016835488070849307,
+      "loss": 0.7539,
+      "step": 7457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016834685023608042,
+      "loss": 0.7727,
+      "step": 7458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016833881893644307,
+      "loss": 0.7794,
+      "step": 7459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001683307868096782,
+      "loss": 0.7982,
+      "step": 7460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016832275385588306,
+      "loss": 0.7703,
+      "step": 7461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016831472007515483,
+      "loss": 0.825,
+      "step": 7462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016830668546759077,
+      "loss": 0.8126,
+      "step": 7463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001682986500332881,
+      "loss": 0.7295,
+      "step": 7464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016829061377234412,
+      "loss": 0.6771,
+      "step": 7465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016828257668485605,
+      "loss": 0.813,
+      "step": 7466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001682745387709212,
+      "loss": 0.7342,
+      "step": 7467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016826650003063686,
+      "loss": 0.7705,
+      "step": 7468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016825846046410024,
+      "loss": 0.7583,
+      "step": 7469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016825042007140875,
+      "loss": 0.853,
+      "step": 7470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016824237885265967,
+      "loss": 0.8176,
+      "step": 7471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016823433680795033,
+      "loss": 0.7575,
+      "step": 7472
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016822629393737804,
+      "loss": 0.805,
+      "step": 7473
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001682182502410402,
+      "loss": 0.7652,
+      "step": 7474
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001682102057190341,
+      "loss": 0.7892,
+      "step": 7475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016820216037145716,
+      "loss": 0.8071,
+      "step": 7476
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016819411419840668,
+      "loss": 0.7768,
+      "step": 7477
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016818606719998014,
+      "loss": 0.6866,
+      "step": 7478
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001681780193762749,
+      "loss": 0.6737,
+      "step": 7479
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001681699707273883,
+      "loss": 0.8358,
+      "step": 7480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016816192125341788,
+      "loss": 0.7434,
+      "step": 7481
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016815387095446096,
+      "loss": 0.7047,
+      "step": 7482
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016814581983061504,
+      "loss": 0.8534,
+      "step": 7483
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001681377678819775,
+      "loss": 0.9578,
+      "step": 7484
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016812971510864587,
+      "loss": 0.78,
+      "step": 7485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016812166151071757,
+      "loss": 0.7759,
+      "step": 7486
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016811360708829007,
+      "loss": 0.703,
+      "step": 7487
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001681055518414609,
+      "loss": 0.7285,
+      "step": 7488
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001680974957703275,
+      "loss": 0.7804,
+      "step": 7489
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016808943887498741,
+      "loss": 0.8076,
+      "step": 7490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016808138115553814,
+      "loss": 0.8369,
+      "step": 7491
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016807332261207722,
+      "loss": 0.8015,
+      "step": 7492
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016806526324470217,
+      "loss": 0.8078,
+      "step": 7493
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016805720305351055,
+      "loss": 0.6943,
+      "step": 7494
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001680491420385999,
+      "loss": 0.9081,
+      "step": 7495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016804108020006776,
+      "loss": 0.8439,
+      "step": 7496
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016803301753801178,
+      "loss": 0.8391,
+      "step": 7497
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001680249540525295,
+      "loss": 0.7634,
+      "step": 7498
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001680168897437185,
+      "loss": 0.7946,
+      "step": 7499
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016800882461167645,
+      "loss": 0.7735,
+      "step": 7500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016800075865650087,
+      "loss": 0.6973,
+      "step": 7501
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016799269187828944,
+      "loss": 0.7879,
+      "step": 7502
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001679846242771398,
+      "loss": 0.8205,
+      "step": 7503
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016797655585314958,
+      "loss": 0.7858,
+      "step": 7504
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016796848660641643,
+      "loss": 0.7885,
+      "step": 7505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016796041653703805,
+      "loss": 0.7616,
+      "step": 7506
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016795234564511205,
+      "loss": 0.7895,
+      "step": 7507
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016794427393073614,
+      "loss": 0.8819,
+      "step": 7508
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016793620139400806,
+      "loss": 0.8157,
+      "step": 7509
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001679281280350255,
+      "loss": 0.772,
+      "step": 7510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016792005385388612,
+      "loss": 0.8539,
+      "step": 7511
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016791197885068768,
+      "loss": 0.7502,
+      "step": 7512
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016790390302552792,
+      "loss": 0.6735,
+      "step": 7513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001678958263785046,
+      "loss": 0.9375,
+      "step": 7514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016788774890971544,
+      "loss": 0.8182,
+      "step": 7515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001678796706192582,
+      "loss": 0.8954,
+      "step": 7516
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001678715915072307,
+      "loss": 0.8972,
+      "step": 7517
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016786351157373067,
+      "loss": 0.8028,
+      "step": 7518
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016785543081885597,
+      "loss": 0.867,
+      "step": 7519
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016784734924270433,
+      "loss": 0.7767,
+      "step": 7520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001678392668453736,
+      "loss": 0.6374,
+      "step": 7521
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016783118362696163,
+      "loss": 0.8176,
+      "step": 7522
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001678230995875662,
+      "loss": 0.7929,
+      "step": 7523
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016781501472728518,
+      "loss": 0.7511,
+      "step": 7524
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016780692904621642,
+      "loss": 0.8625,
+      "step": 7525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001677988425444578,
+      "loss": 0.7828,
+      "step": 7526
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001677907552221072,
+      "loss": 0.857,
+      "step": 7527
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016778266707926243,
+      "loss": 0.6855,
+      "step": 7528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016777457811602148,
+      "loss": 0.7728,
+      "step": 7529
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016776648833248222,
+      "loss": 0.6759,
+      "step": 7530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016775839772874255,
+      "loss": 0.8212,
+      "step": 7531
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016775030630490037,
+      "loss": 0.8871,
+      "step": 7532
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016774221406105365,
+      "loss": 0.7727,
+      "step": 7533
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016773412099730032,
+      "loss": 0.7873,
+      "step": 7534
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016772602711373838,
+      "loss": 0.7082,
+      "step": 7535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001677179324104657,
+      "loss": 0.7906,
+      "step": 7536
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016770983688758033,
+      "loss": 0.7415,
+      "step": 7537
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016770174054518022,
+      "loss": 0.6891,
+      "step": 7538
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016769364338336339,
+      "loss": 0.8753,
+      "step": 7539
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001676855454022278,
+      "loss": 0.7657,
+      "step": 7540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016767744660187148,
+      "loss": 0.8061,
+      "step": 7541
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016766934698239245,
+      "loss": 0.8194,
+      "step": 7542
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016766124654388879,
+      "loss": 0.6987,
+      "step": 7543
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016765314528645846,
+      "loss": 0.8561,
+      "step": 7544
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016764504321019956,
+      "loss": 0.7833,
+      "step": 7545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016763694031521014,
+      "loss": 0.7195,
+      "step": 7546
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016762883660158831,
+      "loss": 0.8649,
+      "step": 7547
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001676207320694321,
+      "loss": 0.6298,
+      "step": 7548
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016761262671883962,
+      "loss": 0.8761,
+      "step": 7549
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016760452054990897,
+      "loss": 0.7997,
+      "step": 7550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016759641356273828,
+      "loss": 0.7944,
+      "step": 7551
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016758830575742563,
+      "loss": 0.7417,
+      "step": 7552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016758019713406918,
+      "loss": 0.663,
+      "step": 7553
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001675720876927671,
+      "loss": 0.6937,
+      "step": 7554
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001675639774336175,
+      "loss": 0.7977,
+      "step": 7555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016755586635671855,
+      "loss": 0.6735,
+      "step": 7556
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016754775446216843,
+      "loss": 0.7862,
+      "step": 7557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001675396417500653,
+      "loss": 0.731,
+      "step": 7558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016753152822050735,
+      "loss": 0.7913,
+      "step": 7559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016752341387359283,
+      "loss": 0.8731,
+      "step": 7560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016751529870941988,
+      "loss": 0.7649,
+      "step": 7561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001675071827280868,
+      "loss": 0.7598,
+      "step": 7562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016749906592969174,
+      "loss": 0.7509,
+      "step": 7563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000167490948314333,
+      "loss": 0.7055,
+      "step": 7564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016748282988210877,
+      "loss": 0.7559,
+      "step": 7565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001674747106331174,
+      "loss": 0.7132,
+      "step": 7566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001674665905674571,
+      "loss": 0.7546,
+      "step": 7567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016745846968522612,
+      "loss": 0.7406,
+      "step": 7568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016745034798652282,
+      "loss": 0.8818,
+      "step": 7569
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016744222547144546,
+      "loss": 0.9921,
+      "step": 7570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016743410214009233,
+      "loss": 0.9484,
+      "step": 7571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00016742597799256182,
+      "loss": 0.7517,
+      "step": 7572
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001674178530289522,
+      "loss": 0.6604,
+      "step": 7573
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016740972724936182,
+      "loss": 0.8546,
+      "step": 7574
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016740160065388903,
+      "loss": 0.7802,
+      "step": 7575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001673934732426322,
+      "loss": 0.7801,
+      "step": 7576
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016738534501568967,
+      "loss": 0.7504,
+      "step": 7577
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016737721597315984,
+      "loss": 0.7335,
+      "step": 7578
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016736908611514112,
+      "loss": 0.8724,
+      "step": 7579
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016736095544173186,
+      "loss": 0.7402,
+      "step": 7580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016735282395303054,
+      "loss": 0.7563,
+      "step": 7581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016734469164913552,
+      "loss": 0.806,
+      "step": 7582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016733655853014523,
+      "loss": 0.7601,
+      "step": 7583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016732842459615813,
+      "loss": 0.7476,
+      "step": 7584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001673202898472726,
+      "loss": 0.6924,
+      "step": 7585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001673121542835872,
+      "loss": 0.8348,
+      "step": 7586
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016730401790520036,
+      "loss": 0.7655,
+      "step": 7587
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016729588071221055,
+      "loss": 0.8157,
+      "step": 7588
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016728774270471624,
+      "loss": 0.7617,
+      "step": 7589
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016727960388281593,
+      "loss": 0.8796,
+      "step": 7590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016727146424660817,
+      "loss": 0.799,
+      "step": 7591
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016726332379619143,
+      "loss": 0.7673,
+      "step": 7592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016725518253166426,
+      "loss": 0.7858,
+      "step": 7593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016724704045312518,
+      "loss": 0.7984,
+      "step": 7594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016723889756067276,
+      "loss": 0.7567,
+      "step": 7595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016723075385440553,
+      "loss": 0.8365,
+      "step": 7596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016722260933442208,
+      "loss": 0.6957,
+      "step": 7597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016721446400082095,
+      "loss": 0.7274,
+      "step": 7598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016720631785370078,
+      "loss": 0.7557,
+      "step": 7599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016719817089316013,
+      "loss": 0.7186,
+      "step": 7600
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 0.7758554816246033,
+      "eval_runtime": 849.5166,
+      "eval_samples_per_second": 28.322,
+      "eval_steps_per_second": 14.161,
+      "step": 7600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016719002311929758,
+      "loss": 0.8232,
+      "step": 7601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016718187453221178,
+      "loss": 0.8548,
+      "step": 7602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001671737251320014,
+      "loss": 0.7812,
+      "step": 7603
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016716557491876498,
+      "loss": 0.7537,
+      "step": 7604
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016715742389260122,
+      "loss": 0.754,
+      "step": 7605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016714927205360876,
+      "loss": 0.808,
+      "step": 7606
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001671411194018863,
+      "loss": 0.8202,
+      "step": 7607
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016713296593753247,
+      "loss": 0.8207,
+      "step": 7608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016712481166064594,
+      "loss": 0.8283,
+      "step": 7609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016711665657132544,
+      "loss": 0.7995,
+      "step": 7610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016710850066966966,
+      "loss": 0.9201,
+      "step": 7611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016710034395577735,
+      "loss": 0.716,
+      "step": 7612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016709218642974721,
+      "loss": 0.7707,
+      "step": 7613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016708402809167792,
+      "loss": 0.8374,
+      "step": 7614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001670758689416683,
+      "loss": 0.7803,
+      "step": 7615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016706770897981707,
+      "loss": 0.8559,
+      "step": 7616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016705954820622298,
+      "loss": 0.8909,
+      "step": 7617
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016705138662098486,
+      "loss": 0.818,
+      "step": 7618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001670432242242014,
+      "loss": 0.719,
+      "step": 7619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016703506101597144,
+      "loss": 0.8402,
+      "step": 7620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016702689699639383,
+      "loss": 0.7435,
+      "step": 7621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016701873216556728,
+      "loss": 0.7543,
+      "step": 7622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001670105665235907,
+      "loss": 0.8329,
+      "step": 7623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016700240007056286,
+      "loss": 0.8149,
+      "step": 7624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016699423280658267,
+      "loss": 0.7696,
+      "step": 7625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016698606473174895,
+      "loss": 0.8391,
+      "step": 7626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001669778958461605,
+      "loss": 0.7818,
+      "step": 7627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016696972614991626,
+      "loss": 0.856,
+      "step": 7628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001669615556431151,
+      "loss": 0.7638,
+      "step": 7629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001669533843258559,
+      "loss": 0.7443,
+      "step": 7630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016694521219823757,
+      "loss": 0.7772,
+      "step": 7631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016693703926035902,
+      "loss": 0.7288,
+      "step": 7632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016692886551231918,
+      "loss": 0.746,
+      "step": 7633
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016692069095421692,
+      "loss": 0.6759,
+      "step": 7634
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016691251558615124,
+      "loss": 0.8865,
+      "step": 7635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016690433940822106,
+      "loss": 0.7613,
+      "step": 7636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016689616242052536,
+      "loss": 0.7386,
+      "step": 7637
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016688798462316312,
+      "loss": 0.7759,
+      "step": 7638
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001668798060162333,
+      "loss": 0.696,
+      "step": 7639
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016687162659983487,
+      "loss": 0.7422,
+      "step": 7640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016686344637406684,
+      "loss": 0.8945,
+      "step": 7641
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016685526533902822,
+      "loss": 0.825,
+      "step": 7642
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016684708349481804,
+      "loss": 0.8319,
+      "step": 7643
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016683890084153534,
+      "loss": 0.7445,
+      "step": 7644
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016683071737927908,
+      "loss": 0.7798,
+      "step": 7645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001668225331081484,
+      "loss": 0.8072,
+      "step": 7646
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001668143480282423,
+      "loss": 0.955,
+      "step": 7647
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016680616213965988,
+      "loss": 0.8436,
+      "step": 7648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016679797544250023,
+      "loss": 0.8665,
+      "step": 7649
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016678978793686238,
+      "loss": 0.8461,
+      "step": 7650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016678159962284545,
+      "loss": 0.8251,
+      "step": 7651
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016677341050054856,
+      "loss": 0.6797,
+      "step": 7652
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016676522057007081,
+      "loss": 0.8542,
+      "step": 7653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016675702983151136,
+      "loss": 0.5544,
+      "step": 7654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001667488382849693,
+      "loss": 0.9289,
+      "step": 7655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001667406459305438,
+      "loss": 0.7578,
+      "step": 7656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000166732452768334,
+      "loss": 0.808,
+      "step": 7657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016672425879843904,
+      "loss": 0.7927,
+      "step": 7658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016671606402095818,
+      "loss": 0.7167,
+      "step": 7659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016670786843599053,
+      "loss": 0.7676,
+      "step": 7660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001666996720436353,
+      "loss": 0.7003,
+      "step": 7661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001666914748439917,
+      "loss": 0.8689,
+      "step": 7662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016668327683715895,
+      "loss": 0.7443,
+      "step": 7663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016667507802323623,
+      "loss": 0.8529,
+      "step": 7664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016666687840232288,
+      "loss": 0.7584,
+      "step": 7665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016665867797451803,
+      "loss": 0.93,
+      "step": 7666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016665047673992094,
+      "loss": 0.8556,
+      "step": 7667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016664227469863097,
+      "loss": 0.7479,
+      "step": 7668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016663407185074725,
+      "loss": 0.8216,
+      "step": 7669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001666258681963692,
+      "loss": 0.7196,
+      "step": 7670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000166617663735596,
+      "loss": 0.7538,
+      "step": 7671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016660945846852708,
+      "loss": 0.7735,
+      "step": 7672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016660125239526162,
+      "loss": 0.7926,
+      "step": 7673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000166593045515899,
+      "loss": 0.7439,
+      "step": 7674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016658483783053854,
+      "loss": 0.7497,
+      "step": 7675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001665766293392796,
+      "loss": 0.7518,
+      "step": 7676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001665684200422215,
+      "loss": 0.8543,
+      "step": 7677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016656020993946363,
+      "loss": 0.7865,
+      "step": 7678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016655199903110534,
+      "loss": 0.7582,
+      "step": 7679
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000166543787317246,
+      "loss": 0.8306,
+      "step": 7680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016653557479798502,
+      "loss": 0.8182,
+      "step": 7681
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001665273614734218,
+      "loss": 0.9017,
+      "step": 7682
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016651914734365573,
+      "loss": 0.7411,
+      "step": 7683
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016651093240878625,
+      "loss": 0.8245,
+      "step": 7684
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016650271666891279,
+      "loss": 0.685,
+      "step": 7685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016649450012413476,
+      "loss": 0.8302,
+      "step": 7686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001664862827745516,
+      "loss": 0.731,
+      "step": 7687
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001664780646202628,
+      "loss": 0.7805,
+      "step": 7688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016646984566136782,
+      "loss": 0.8156,
+      "step": 7689
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016646162589796615,
+      "loss": 0.8302,
+      "step": 7690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016645340533015724,
+      "loss": 0.7527,
+      "step": 7691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016644518395804062,
+      "loss": 0.8409,
+      "step": 7692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016643696178171577,
+      "loss": 0.7871,
+      "step": 7693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016642873880128224,
+      "loss": 0.6804,
+      "step": 7694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016642051501683953,
+      "loss": 0.775,
+      "step": 7695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016641229042848717,
+      "loss": 0.7435,
+      "step": 7696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001664040650363247,
+      "loss": 0.7771,
+      "step": 7697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001663958388404517,
+      "loss": 0.7829,
+      "step": 7698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016638761184096775,
+      "loss": 0.7365,
+      "step": 7699
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016637938403797235,
+      "loss": 0.7408,
+      "step": 7700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016637115543156514,
+      "loss": 0.828,
+      "step": 7701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016636292602184574,
+      "loss": 0.8243,
+      "step": 7702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016635469580891367,
+      "loss": 0.7474,
+      "step": 7703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001663464647928686,
+      "loss": 0.776,
+      "step": 7704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016633823297381017,
+      "loss": 0.7566,
+      "step": 7705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016633000035183799,
+      "loss": 0.7913,
+      "step": 7706
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001663217669270517,
+      "loss": 0.7602,
+      "step": 7707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001663135326995509,
+      "loss": 0.6928,
+      "step": 7708
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016630529766943535,
+      "loss": 0.7057,
+      "step": 7709
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016629706183680465,
+      "loss": 0.7862,
+      "step": 7710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001662888252017585,
+      "loss": 0.7816,
+      "step": 7711
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001662805877643966,
+      "loss": 0.8273,
+      "step": 7712
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016627234952481867,
+      "loss": 0.7639,
+      "step": 7713
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016626411048312436,
+      "loss": 0.8059,
+      "step": 7714
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016625587063941346,
+      "loss": 0.73,
+      "step": 7715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016624762999378564,
+      "loss": 0.7855,
+      "step": 7716
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016623938854634066,
+      "loss": 0.8489,
+      "step": 7717
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016623114629717826,
+      "loss": 0.7609,
+      "step": 7718
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016622290324639827,
+      "loss": 0.7136,
+      "step": 7719
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016621465939410033,
+      "loss": 0.8112,
+      "step": 7720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016620641474038433,
+      "loss": 0.8258,
+      "step": 7721
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016619816928535,
+      "loss": 0.7249,
+      "step": 7722
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016618992302909718,
+      "loss": 0.745,
+      "step": 7723
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016618167597172562,
+      "loss": 0.691,
+      "step": 7724
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016617342811333518,
+      "loss": 0.8794,
+      "step": 7725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001661651794540257,
+      "loss": 0.8319,
+      "step": 7726
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016615692999389693,
+      "loss": 0.6417,
+      "step": 7727
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001661486797330488,
+      "loss": 0.7132,
+      "step": 7728
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016614042867158117,
+      "loss": 0.8437,
+      "step": 7729
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016613217680959386,
+      "loss": 0.7572,
+      "step": 7730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016612392414718676,
+      "loss": 0.8353,
+      "step": 7731
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016611567068445976,
+      "loss": 0.8683,
+      "step": 7732
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016610741642151275,
+      "loss": 0.7495,
+      "step": 7733
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016609916135844563,
+      "loss": 0.8168,
+      "step": 7734
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016609090549535833,
+      "loss": 0.731,
+      "step": 7735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016608264883235073,
+      "loss": 0.8248,
+      "step": 7736
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016607439136952286,
+      "loss": 0.8242,
+      "step": 7737
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016606613310697453,
+      "loss": 0.85,
+      "step": 7738
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016605787404480582,
+      "loss": 0.7246,
+      "step": 7739
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016604961418311656,
+      "loss": 0.8368,
+      "step": 7740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016604135352200688,
+      "loss": 0.8011,
+      "step": 7741
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016603309206157662,
+      "loss": 0.6976,
+      "step": 7742
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016602482980192584,
+      "loss": 0.8304,
+      "step": 7743
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016601656674315453,
+      "loss": 0.793,
+      "step": 7744
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016600830288536272,
+      "loss": 0.8335,
+      "step": 7745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001660000382286504,
+      "loss": 0.8105,
+      "step": 7746
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001659917727731176,
+      "loss": 0.8473,
+      "step": 7747
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016598350651886437,
+      "loss": 0.8051,
+      "step": 7748
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016597523946599077,
+      "loss": 0.6996,
+      "step": 7749
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016596697161459683,
+      "loss": 0.67,
+      "step": 7750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016595870296478264,
+      "loss": 0.7436,
+      "step": 7751
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016595043351664827,
+      "loss": 0.6331,
+      "step": 7752
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016594216327029383,
+      "loss": 0.8225,
+      "step": 7753
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016593389222581938,
+      "loss": 0.7842,
+      "step": 7754
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016592562038332507,
+      "loss": 0.88,
+      "step": 7755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016591734774291095,
+      "loss": 0.7089,
+      "step": 7756
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016590907430467727,
+      "loss": 0.9068,
+      "step": 7757
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016590080006872402,
+      "loss": 0.7656,
+      "step": 7758
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016589252503515143,
+      "loss": 0.7354,
+      "step": 7759
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016588424920405964,
+      "loss": 0.8151,
+      "step": 7760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001658759725755488,
+      "loss": 0.7438,
+      "step": 7761
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001658676951497191,
+      "loss": 0.7583,
+      "step": 7762
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016585941692667073,
+      "loss": 0.9028,
+      "step": 7763
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016585113790650388,
+      "loss": 0.8233,
+      "step": 7764
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016584285808931873,
+      "loss": 0.8287,
+      "step": 7765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016583457747521553,
+      "loss": 0.7105,
+      "step": 7766
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001658262960642945,
+      "loss": 0.76,
+      "step": 7767
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016581801385665584,
+      "loss": 0.7522,
+      "step": 7768
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001658097308523998,
+      "loss": 0.7568,
+      "step": 7769
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016580144705162665,
+      "loss": 0.748,
+      "step": 7770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016579316245443667,
+      "loss": 0.7037,
+      "step": 7771
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016578487706093008,
+      "loss": 0.7215,
+      "step": 7772
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001657765908712072,
+      "loss": 0.7726,
+      "step": 7773
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016576830388536831,
+      "loss": 0.7774,
+      "step": 7774
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001657600161035137,
+      "loss": 0.8311,
+      "step": 7775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016575172752574368,
+      "loss": 0.6522,
+      "step": 7776
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016574343815215859,
+      "loss": 0.8768,
+      "step": 7777
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016573514798285874,
+      "loss": 0.8635,
+      "step": 7778
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016572685701794448,
+      "loss": 0.8478,
+      "step": 7779
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016571856525751614,
+      "loss": 0.6649,
+      "step": 7780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001657102727016741,
+      "loss": 0.7638,
+      "step": 7781
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001657019793505187,
+      "loss": 0.7942,
+      "step": 7782
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016569368520415035,
+      "loss": 0.7793,
+      "step": 7783
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016568539026266945,
+      "loss": 0.6441,
+      "step": 7784
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016567709452617633,
+      "loss": 0.8234,
+      "step": 7785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016566879799477147,
+      "loss": 0.6625,
+      "step": 7786
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001656605006685552,
+      "loss": 0.6988,
+      "step": 7787
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016565220254762805,
+      "loss": 0.8021,
+      "step": 7788
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016564390363209036,
+      "loss": 0.7485,
+      "step": 7789
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016563560392204266,
+      "loss": 0.6933,
+      "step": 7790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016562730341758533,
+      "loss": 0.7805,
+      "step": 7791
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016561900211881886,
+      "loss": 0.7435,
+      "step": 7792
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016561070002584372,
+      "loss": 0.8134,
+      "step": 7793
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001656023971387604,
+      "loss": 0.7898,
+      "step": 7794
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016559409345766942,
+      "loss": 0.7619,
+      "step": 7795
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016558578898267126,
+      "loss": 0.7385,
+      "step": 7796
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016557748371386638,
+      "loss": 0.8179,
+      "step": 7797
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016556917765135536,
+      "loss": 0.707,
+      "step": 7798
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016556087079523874,
+      "loss": 0.8027,
+      "step": 7799
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016555256314561704,
+      "loss": 0.8891,
+      "step": 7800
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 0.7663759589195251,
+      "eval_runtime": 851.9669,
+      "eval_samples_per_second": 28.241,
+      "eval_steps_per_second": 14.12,
+      "step": 7800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001655442547025908,
+      "loss": 0.7211,
+      "step": 7801
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016553594546626061,
+      "loss": 0.8017,
+      "step": 7802
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000165527635436727,
+      "loss": 0.7575,
+      "step": 7803
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016551932461409056,
+      "loss": 0.8638,
+      "step": 7804
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001655110129984519,
+      "loss": 0.8173,
+      "step": 7805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001655027005899116,
+      "loss": 0.7374,
+      "step": 7806
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016549438738857025,
+      "loss": 0.8129,
+      "step": 7807
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016548607339452853,
+      "loss": 0.7028,
+      "step": 7808
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000165477758607887,
+      "loss": 0.8346,
+      "step": 7809
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016546944302874635,
+      "loss": 0.7802,
+      "step": 7810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016546112665720718,
+      "loss": 0.8033,
+      "step": 7811
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016545280949337014,
+      "loss": 0.8219,
+      "step": 7812
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016544449153733598,
+      "loss": 0.8919,
+      "step": 7813
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016543617278920526,
+      "loss": 0.8879,
+      "step": 7814
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016542785324907876,
+      "loss": 0.7664,
+      "step": 7815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016541953291705715,
+      "loss": 0.8102,
+      "step": 7816
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016541121179324106,
+      "loss": 0.8339,
+      "step": 7817
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001654028898777313,
+      "loss": 0.7587,
+      "step": 7818
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016539456717062855,
+      "loss": 0.8106,
+      "step": 7819
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016538624367203357,
+      "loss": 0.7993,
+      "step": 7820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016537791938204705,
+      "loss": 0.7915,
+      "step": 7821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016536959430076976,
+      "loss": 0.7977,
+      "step": 7822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001653612684283025,
+      "loss": 0.705,
+      "step": 7823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000165352941764746,
+      "loss": 0.8395,
+      "step": 7824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016534461431020106,
+      "loss": 0.8388,
+      "step": 7825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016533628606476848,
+      "loss": 0.7634,
+      "step": 7826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016532795702854902,
+      "loss": 0.8033,
+      "step": 7827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016531962720164352,
+      "loss": 0.7247,
+      "step": 7828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001653112965841528,
+      "loss": 0.7774,
+      "step": 7829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016530296517617768,
+      "loss": 0.6701,
+      "step": 7830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016529463297781897,
+      "loss": 0.7294,
+      "step": 7831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016528629998917754,
+      "loss": 0.7797,
+      "step": 7832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001652779662103543,
+      "loss": 0.7651,
+      "step": 7833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016526963164145003,
+      "loss": 0.7624,
+      "step": 7834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016526129628256566,
+      "loss": 0.7702,
+      "step": 7835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016525296013380204,
+      "loss": 0.8302,
+      "step": 7836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016524462319526012,
+      "loss": 0.7312,
+      "step": 7837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016523628546704074,
+      "loss": 0.862,
+      "step": 7838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016522794694924487,
+      "loss": 0.6627,
+      "step": 7839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016521960764197336,
+      "loss": 0.6922,
+      "step": 7840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016521126754532724,
+      "loss": 0.8735,
+      "step": 7841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001652029266594074,
+      "loss": 0.7525,
+      "step": 7842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016519458498431478,
+      "loss": 0.7597,
+      "step": 7843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016518624252015037,
+      "loss": 0.8558,
+      "step": 7844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016517789926701512,
+      "loss": 0.833,
+      "step": 7845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016516955522501001,
+      "loss": 0.7884,
+      "step": 7846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016516121039423606,
+      "loss": 0.7755,
+      "step": 7847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016515286477479423,
+      "loss": 0.7264,
+      "step": 7848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001651445183667856,
+      "loss": 0.8368,
+      "step": 7849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001651361711703111,
+      "loss": 0.692,
+      "step": 7850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016512782318547182,
+      "loss": 0.7607,
+      "step": 7851
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016511947441236877,
+      "loss": 0.7587,
+      "step": 7852
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000165111124851103,
+      "loss": 0.7486,
+      "step": 7853
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001651027745017756,
+      "loss": 0.6887,
+      "step": 7854
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001650944233644876,
+      "loss": 0.7453,
+      "step": 7855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016508607143934006,
+      "loss": 0.701,
+      "step": 7856
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016507771872643413,
+      "loss": 0.755,
+      "step": 7857
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00016506936522587087,
+      "loss": 0.7816,
+      "step": 7858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016506101093775137,
+      "loss": 0.7108,
+      "step": 7859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016505265586217677,
+      "loss": 0.6386,
+      "step": 7860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016504429999924817,
+      "loss": 0.7384,
+      "step": 7861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016503594334906673,
+      "loss": 0.7625,
+      "step": 7862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016502758591173356,
+      "loss": 0.7756,
+      "step": 7863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016501922768734988,
+      "loss": 0.8006,
+      "step": 7864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016501086867601677,
+      "loss": 0.8112,
+      "step": 7865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016500250887783544,
+      "loss": 0.7771,
+      "step": 7866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001649941482929071,
+      "loss": 0.7297,
+      "step": 7867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001649857869213329,
+      "loss": 0.7528,
+      "step": 7868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016497742476321404,
+      "loss": 0.8049,
+      "step": 7869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016496906181865173,
+      "loss": 0.7346,
+      "step": 7870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016496069808774721,
+      "loss": 0.6947,
+      "step": 7871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016495233357060173,
+      "loss": 0.7135,
+      "step": 7872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016494396826731645,
+      "loss": 0.869,
+      "step": 7873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001649356021779927,
+      "loss": 0.7715,
+      "step": 7874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001649272353027317,
+      "loss": 0.8613,
+      "step": 7875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016491886764163468,
+      "loss": 0.822,
+      "step": 7876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000164910499194803,
+      "loss": 0.7344,
+      "step": 7877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001649021299623379,
+      "loss": 0.821,
+      "step": 7878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016489375994434066,
+      "loss": 0.7712,
+      "step": 7879
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016488538914091264,
+      "loss": 0.7195,
+      "step": 7880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001648770175521551,
+      "loss": 0.734,
+      "step": 7881
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016486864517816936,
+      "loss": 0.6967,
+      "step": 7882
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001648602720190568,
+      "loss": 0.8391,
+      "step": 7883
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016485189807491875,
+      "loss": 0.8486,
+      "step": 7884
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016484352334585653,
+      "loss": 0.7107,
+      "step": 7885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016483514783197155,
+      "loss": 0.7037,
+      "step": 7886
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016482677153336513,
+      "loss": 0.7293,
+      "step": 7887
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016481839445013868,
+      "loss": 0.8651,
+      "step": 7888
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016481001658239362,
+      "loss": 0.6405,
+      "step": 7889
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001648016379302313,
+      "loss": 0.7112,
+      "step": 7890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016479325849375314,
+      "loss": 0.8545,
+      "step": 7891
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016478487827306055,
+      "loss": 0.8436,
+      "step": 7892
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016477649726825502,
+      "loss": 0.7623,
+      "step": 7893
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016476811547943792,
+      "loss": 0.7699,
+      "step": 7894
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016475973290671072,
+      "loss": 0.71,
+      "step": 7895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001647513495501749,
+      "loss": 0.7439,
+      "step": 7896
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016474296540993185,
+      "loss": 0.8216,
+      "step": 7897
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016473458048608315,
+      "loss": 0.7549,
+      "step": 7898
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016472619477873025,
+      "loss": 0.7841,
+      "step": 7899
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016471780828797457,
+      "loss": 0.8083,
+      "step": 7900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016470942101391774,
+      "loss": 0.7893,
+      "step": 7901
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016470103295666117,
+      "loss": 0.6678,
+      "step": 7902
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016469264411630644,
+      "loss": 0.7422,
+      "step": 7903
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016468425449295504,
+      "loss": 0.7738,
+      "step": 7904
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016467586408670858,
+      "loss": 0.799,
+      "step": 7905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016466747289766852,
+      "loss": 0.8305,
+      "step": 7906
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016465908092593652,
+      "loss": 0.8099,
+      "step": 7907
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016465068817161409,
+      "loss": 0.8511,
+      "step": 7908
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016464229463480283,
+      "loss": 0.7406,
+      "step": 7909
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016463390031560432,
+      "loss": 0.6964,
+      "step": 7910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016462550521412014,
+      "loss": 0.8122,
+      "step": 7911
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016461710933045193,
+      "loss": 0.8094,
+      "step": 7912
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001646087126647013,
+      "loss": 0.8303,
+      "step": 7913
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001646003152169699,
+      "loss": 0.858,
+      "step": 7914
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016459191698735932,
+      "loss": 0.7864,
+      "step": 7915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016458351797597124,
+      "loss": 0.9528,
+      "step": 7916
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016457511818290728,
+      "loss": 0.7847,
+      "step": 7917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016456671760826916,
+      "loss": 0.6732,
+      "step": 7918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016455831625215853,
+      "loss": 0.7544,
+      "step": 7919
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016454991411467707,
+      "loss": 0.8595,
+      "step": 7920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016454151119592647,
+      "loss": 0.8003,
+      "step": 7921
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016453310749600842,
+      "loss": 0.9295,
+      "step": 7922
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001645247030150247,
+      "loss": 0.8662,
+      "step": 7923
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016451629775307694,
+      "loss": 0.8151,
+      "step": 7924
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016450789171026692,
+      "loss": 0.7393,
+      "step": 7925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016449948488669639,
+      "loss": 0.6787,
+      "step": 7926
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016449107728246708,
+      "loss": 0.8316,
+      "step": 7927
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001644826688976808,
+      "loss": 0.7789,
+      "step": 7928
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001644742597324392,
+      "loss": 0.8293,
+      "step": 7929
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001644658497868442,
+      "loss": 0.8566,
+      "step": 7930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016445743906099745,
+      "loss": 0.7675,
+      "step": 7931
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001644490275550009,
+      "loss": 0.7195,
+      "step": 7932
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016444061526895624,
+      "loss": 0.7473,
+      "step": 7933
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016443220220296534,
+      "loss": 0.7995,
+      "step": 7934
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016442378835713,
+      "loss": 0.7377,
+      "step": 7935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016441537373155206,
+      "loss": 0.7443,
+      "step": 7936
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016440695832633338,
+      "loss": 0.7913,
+      "step": 7937
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001643985421415758,
+      "loss": 0.8212,
+      "step": 7938
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016439012517738123,
+      "loss": 0.8411,
+      "step": 7939
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016438170743385147,
+      "loss": 0.7907,
+      "step": 7940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016437328891108843,
+      "loss": 0.7739,
+      "step": 7941
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016436486960919405,
+      "loss": 0.7654,
+      "step": 7942
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016435644952827013,
+      "loss": 0.7349,
+      "step": 7943
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001643480286684187,
+      "loss": 0.7617,
+      "step": 7944
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016433960702974155,
+      "loss": 0.6836,
+      "step": 7945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016433118461234075,
+      "loss": 0.7381,
+      "step": 7946
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016432276141631818,
+      "loss": 0.7947,
+      "step": 7947
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001643143374417757,
+      "loss": 0.9124,
+      "step": 7948
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016430591268881543,
+      "loss": 0.8259,
+      "step": 7949
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001642974871575392,
+      "loss": 0.732,
+      "step": 7950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016428906084804906,
+      "loss": 0.7221,
+      "step": 7951
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000164280633760447,
+      "loss": 0.7426,
+      "step": 7952
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016427220589483498,
+      "loss": 0.7921,
+      "step": 7953
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000164263777251315,
+      "loss": 0.8769,
+      "step": 7954
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016425534782998913,
+      "loss": 0.6203,
+      "step": 7955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016424691763095936,
+      "loss": 0.8891,
+      "step": 7956
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001642384866543277,
+      "loss": 0.8345,
+      "step": 7957
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016423005490019622,
+      "loss": 0.7439,
+      "step": 7958
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016422162236866698,
+      "loss": 0.6674,
+      "step": 7959
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016421318905984202,
+      "loss": 0.7915,
+      "step": 7960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016420475497382344,
+      "loss": 0.8891,
+      "step": 7961
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001641963201107133,
+      "loss": 0.767,
+      "step": 7962
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016418788447061363,
+      "loss": 0.7268,
+      "step": 7963
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016417944805362665,
+      "loss": 0.7908,
+      "step": 7964
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001641710108598544,
+      "loss": 0.8156,
+      "step": 7965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000164162572889399,
+      "loss": 0.8042,
+      "step": 7966
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001641541341423626,
+      "loss": 0.7318,
+      "step": 7967
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001641456946188473,
+      "loss": 0.8788,
+      "step": 7968
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001641372543189553,
+      "loss": 0.7408,
+      "step": 7969
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001641288132427887,
+      "loss": 0.7881,
+      "step": 7970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001641203713904497,
+      "loss": 0.7503,
+      "step": 7971
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016411192876204043,
+      "loss": 0.6924,
+      "step": 7972
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016410348535766316,
+      "loss": 0.7943,
+      "step": 7973
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016409504117742,
+      "loss": 0.6749,
+      "step": 7974
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016408659622141317,
+      "loss": 0.7939,
+      "step": 7975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016407815048974494,
+      "loss": 0.7029,
+      "step": 7976
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016406970398251747,
+      "loss": 0.7345,
+      "step": 7977
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016406125669983298,
+      "loss": 0.7805,
+      "step": 7978
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016405280864179378,
+      "loss": 0.7638,
+      "step": 7979
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016404435980850205,
+      "loss": 0.7496,
+      "step": 7980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001640359102000601,
+      "loss": 0.679,
+      "step": 7981
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016402745981657018,
+      "loss": 0.9313,
+      "step": 7982
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016401900865813454,
+      "loss": 0.7412,
+      "step": 7983
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001640105567248555,
+      "loss": 0.813,
+      "step": 7984
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016400210401683535,
+      "loss": 0.7794,
+      "step": 7985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016399365053417636,
+      "loss": 0.7421,
+      "step": 7986
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016398519627698095,
+      "loss": 0.7747,
+      "step": 7987
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016397674124535132,
+      "loss": 0.7792,
+      "step": 7988
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016396828543938988,
+      "loss": 0.7394,
+      "step": 7989
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016395982885919895,
+      "loss": 0.7391,
+      "step": 7990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016395137150488086,
+      "loss": 0.7479,
+      "step": 7991
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016394291337653801,
+      "loss": 0.8246,
+      "step": 7992
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016393445447427277,
+      "loss": 0.8312,
+      "step": 7993
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001639259947981875,
+      "loss": 0.8472,
+      "step": 7994
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001639175343483846,
+      "loss": 0.7178,
+      "step": 7995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016390907312496646,
+      "loss": 0.8833,
+      "step": 7996
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001639006111280355,
+      "loss": 0.8594,
+      "step": 7997
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016389214835769414,
+      "loss": 0.7593,
+      "step": 7998
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001638836848140448,
+      "loss": 0.7076,
+      "step": 7999
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016387522049718994,
+      "loss": 0.7919,
+      "step": 8000
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 0.7580129504203796,
+      "eval_runtime": 848.617,
+      "eval_samples_per_second": 28.352,
+      "eval_steps_per_second": 14.176,
+      "step": 8000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016386675540723197,
+      "loss": 0.8154,
+      "step": 8001
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001638582895442734,
+      "loss": 0.7586,
+      "step": 8002
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001638498229084166,
+      "loss": 0.7752,
+      "step": 8003
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001638413554997641,
+      "loss": 0.9047,
+      "step": 8004
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016383288731841843,
+      "loss": 0.7838,
+      "step": 8005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016382441836448202,
+      "loss": 0.6608,
+      "step": 8006
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001638159486380574,
+      "loss": 0.7845,
+      "step": 8007
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016380747813924708,
+      "loss": 0.9333,
+      "step": 8008
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016379900686815354,
+      "loss": 0.7312,
+      "step": 8009
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016379053482487937,
+      "loss": 0.7276,
+      "step": 8010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016378206200952708,
+      "loss": 0.8733,
+      "step": 8011
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016377358842219926,
+      "loss": 0.7027,
+      "step": 8012
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001637651140629984,
+      "loss": 0.7213,
+      "step": 8013
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001637566389320271,
+      "loss": 0.7038,
+      "step": 8014
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016374816302938797,
+      "loss": 0.7371,
+      "step": 8015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016373968635518353,
+      "loss": 0.8513,
+      "step": 8016
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016373120890951642,
+      "loss": 0.7866,
+      "step": 8017
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016372273069248925,
+      "loss": 0.7027,
+      "step": 8018
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001637142517042046,
+      "loss": 0.7548,
+      "step": 8019
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016370577194476513,
+      "loss": 0.7405,
+      "step": 8020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016369729141427346,
+      "loss": 0.6814,
+      "step": 8021
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016368881011283223,
+      "loss": 0.7205,
+      "step": 8022
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001636803280405441,
+      "loss": 0.8248,
+      "step": 8023
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001636718451975117,
+      "loss": 0.8112,
+      "step": 8024
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016366336158383778,
+      "loss": 0.8927,
+      "step": 8025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016365487719962492,
+      "loss": 0.8979,
+      "step": 8026
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016364639204497587,
+      "loss": 0.7555,
+      "step": 8027
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001636379061199933,
+      "loss": 0.6475,
+      "step": 8028
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016362941942477997,
+      "loss": 0.7602,
+      "step": 8029
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016362093195943853,
+      "loss": 0.726,
+      "step": 8030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016361244372407172,
+      "loss": 0.7825,
+      "step": 8031
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016360395471878234,
+      "loss": 0.756,
+      "step": 8032
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016359546494367305,
+      "loss": 0.6687,
+      "step": 8033
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016358697439884665,
+      "loss": 0.7009,
+      "step": 8034
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016357848308440592,
+      "loss": 0.6314,
+      "step": 8035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016356999100045357,
+      "loss": 0.7534,
+      "step": 8036
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016356149814709243,
+      "loss": 0.7718,
+      "step": 8037
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001635530045244253,
+      "loss": 0.7069,
+      "step": 8038
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016354451013255496,
+      "loss": 0.7423,
+      "step": 8039
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001635360149715842,
+      "loss": 0.8327,
+      "step": 8040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001635275190416159,
+      "loss": 0.6772,
+      "step": 8041
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016351902234275283,
+      "loss": 0.8648,
+      "step": 8042
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016351052487509788,
+      "loss": 0.7068,
+      "step": 8043
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016350202663875386,
+      "loss": 0.7716,
+      "step": 8044
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001634935276338236,
+      "loss": 0.6914,
+      "step": 8045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016348502786041005,
+      "loss": 0.7196,
+      "step": 8046
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016347652731861604,
+      "loss": 0.5981,
+      "step": 8047
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016346802600854444,
+      "loss": 0.627,
+      "step": 8048
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016345952393029812,
+      "loss": 0.6384,
+      "step": 8049
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016345102108398005,
+      "loss": 0.7982,
+      "step": 8050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016344251746969312,
+      "loss": 0.7597,
+      "step": 8051
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016343401308754026,
+      "loss": 0.7555,
+      "step": 8052
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016342550793762434,
+      "loss": 0.6855,
+      "step": 8053
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001634170020200484,
+      "loss": 0.7619,
+      "step": 8054
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016340849533491532,
+      "loss": 0.6085,
+      "step": 8055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016339998788232808,
+      "loss": 0.6389,
+      "step": 8056
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016339147966238962,
+      "loss": 0.7298,
+      "step": 8057
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016338297067520298,
+      "loss": 0.636,
+      "step": 8058
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016337446092087106,
+      "loss": 0.7075,
+      "step": 8059
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016336595039949695,
+      "loss": 0.8183,
+      "step": 8060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016335743911118363,
+      "loss": 0.7892,
+      "step": 8061
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016334892705603408,
+      "loss": 0.7896,
+      "step": 8062
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016334041423415134,
+      "loss": 0.8279,
+      "step": 8063
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001633319006456384,
+      "loss": 0.739,
+      "step": 8064
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016332338629059843,
+      "loss": 0.8047,
+      "step": 8065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016331487116913435,
+      "loss": 0.864,
+      "step": 8066
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001633063552813493,
+      "loss": 0.6756,
+      "step": 8067
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001632978386273463,
+      "loss": 0.7314,
+      "step": 8068
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016328932120722846,
+      "loss": 0.7493,
+      "step": 8069
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016328080302109888,
+      "loss": 0.7774,
+      "step": 8070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001632722840690606,
+      "loss": 0.687,
+      "step": 8071
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016326376435121679,
+      "loss": 0.7274,
+      "step": 8072
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016325524386767053,
+      "loss": 0.7545,
+      "step": 8073
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016324672261852497,
+      "loss": 0.8161,
+      "step": 8074
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016323820060388323,
+      "loss": 0.75,
+      "step": 8075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016322967782384851,
+      "loss": 0.7545,
+      "step": 8076
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016322115427852384,
+      "loss": 0.8818,
+      "step": 8077
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016321262996801252,
+      "loss": 0.8356,
+      "step": 8078
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016320410489241763,
+      "loss": 0.736,
+      "step": 8079
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016319557905184238,
+      "loss": 0.7802,
+      "step": 8080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016318705244638997,
+      "loss": 0.7484,
+      "step": 8081
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001631785250761636,
+      "loss": 0.7692,
+      "step": 8082
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016316999694126647,
+      "loss": 0.7231,
+      "step": 8083
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001631614680418018,
+      "loss": 0.774,
+      "step": 8084
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016315293837787284,
+      "loss": 0.775,
+      "step": 8085
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001631444079495828,
+      "loss": 0.8443,
+      "step": 8086
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001631358767570349,
+      "loss": 0.7513,
+      "step": 8087
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001631273448003325,
+      "loss": 0.7381,
+      "step": 8088
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016311881207957873,
+      "loss": 0.7145,
+      "step": 8089
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016311027859487696,
+      "loss": 0.7159,
+      "step": 8090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016310174434633044,
+      "loss": 0.709,
+      "step": 8091
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016309320933404248,
+      "loss": 0.7452,
+      "step": 8092
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016308467355811637,
+      "loss": 0.743,
+      "step": 8093
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001630761370186554,
+      "loss": 0.5904,
+      "step": 8094
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016306759971576292,
+      "loss": 0.6716,
+      "step": 8095
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016305906164954223,
+      "loss": 0.7872,
+      "step": 8096
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016305052282009668,
+      "loss": 0.7353,
+      "step": 8097
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001630419832275297,
+      "loss": 0.6797,
+      "step": 8098
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016303344287194453,
+      "loss": 0.6577,
+      "step": 8099
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016302490175344457,
+      "loss": 0.733,
+      "step": 8100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016301635987213322,
+      "loss": 0.7767,
+      "step": 8101
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016300781722811385,
+      "loss": 0.7478,
+      "step": 8102
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016299927382148989,
+      "loss": 0.7827,
+      "step": 8103
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001629907296523647,
+      "loss": 0.6672,
+      "step": 8104
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016298218472084167,
+      "loss": 0.8377,
+      "step": 8105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016297363902702426,
+      "loss": 0.7312,
+      "step": 8106
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016296509257101595,
+      "loss": 0.7716,
+      "step": 8107
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001629565453529201,
+      "loss": 0.7718,
+      "step": 8108
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016294799737284017,
+      "loss": 0.8034,
+      "step": 8109
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016293944863087964,
+      "loss": 0.8468,
+      "step": 8110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016293089912714198,
+      "loss": 0.8611,
+      "step": 8111
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016292234886173068,
+      "loss": 0.7034,
+      "step": 8112
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001629137978347492,
+      "loss": 0.8625,
+      "step": 8113
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016290524604630106,
+      "loss": 0.7224,
+      "step": 8114
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016289669349648972,
+      "loss": 0.6305,
+      "step": 8115
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016288814018541875,
+      "loss": 0.7989,
+      "step": 8116
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016287958611319167,
+      "loss": 0.7755,
+      "step": 8117
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016287103127991194,
+      "loss": 0.7627,
+      "step": 8118
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016286247568568317,
+      "loss": 0.6888,
+      "step": 8119
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016285391933060895,
+      "loss": 0.8599,
+      "step": 8120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016284536221479275,
+      "loss": 0.6466,
+      "step": 8121
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016283680433833815,
+      "loss": 0.7514,
+      "step": 8122
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016282824570134876,
+      "loss": 0.8828,
+      "step": 8123
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016281968630392819,
+      "loss": 0.7455,
+      "step": 8124
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016281112614618,
+      "loss": 0.7319,
+      "step": 8125
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016280256522820778,
+      "loss": 0.6674,
+      "step": 8126
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001627940035501152,
+      "loss": 0.8646,
+      "step": 8127
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016278544111200582,
+      "loss": 0.7239,
+      "step": 8128
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016277687791398335,
+      "loss": 0.7766,
+      "step": 8129
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016276831395615137,
+      "loss": 0.6805,
+      "step": 8130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016275974923861357,
+      "loss": 0.8056,
+      "step": 8131
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001627511837614736,
+      "loss": 0.738,
+      "step": 8132
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016274261752483512,
+      "loss": 0.8129,
+      "step": 8133
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016273405052880184,
+      "loss": 0.5753,
+      "step": 8134
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001627254827734774,
+      "loss": 0.6882,
+      "step": 8135
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016271691425896555,
+      "loss": 0.6938,
+      "step": 8136
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016270834498536996,
+      "loss": 0.7732,
+      "step": 8137
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016269977495279435,
+      "loss": 0.8549,
+      "step": 8138
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016269120416134247,
+      "loss": 0.7542,
+      "step": 8139
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016268263261111804,
+      "loss": 0.7837,
+      "step": 8140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001626740603022248,
+      "loss": 0.8545,
+      "step": 8141
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016266548723476652,
+      "loss": 0.8346,
+      "step": 8142
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016265691340884694,
+      "loss": 0.7224,
+      "step": 8143
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016264833882456987,
+      "loss": 0.7792,
+      "step": 8144
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016263976348203905,
+      "loss": 0.7293,
+      "step": 8145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016263118738135828,
+      "loss": 0.6778,
+      "step": 8146
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016262261052263135,
+      "loss": 0.8462,
+      "step": 8147
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016261403290596208,
+      "loss": 0.8161,
+      "step": 8148
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016260545453145432,
+      "loss": 0.742,
+      "step": 8149
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016259687539921187,
+      "loss": 0.6296,
+      "step": 8150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016258829550933857,
+      "loss": 0.6949,
+      "step": 8151
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016257971486193822,
+      "loss": 0.7503,
+      "step": 8152
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016257113345711472,
+      "loss": 0.7461,
+      "step": 8153
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016256255129497196,
+      "loss": 0.7425,
+      "step": 8154
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016255396837561376,
+      "loss": 0.7388,
+      "step": 8155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000162545384699144,
+      "loss": 0.7036,
+      "step": 8156
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016253680026566663,
+      "loss": 0.8417,
+      "step": 8157
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016252821507528552,
+      "loss": 0.7385,
+      "step": 8158
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016251962912810453,
+      "loss": 0.7047,
+      "step": 8159
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016251104242422768,
+      "loss": 0.7772,
+      "step": 8160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016250245496375878,
+      "loss": 0.8509,
+      "step": 8161
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016249386674680184,
+      "loss": 0.7995,
+      "step": 8162
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016248527777346084,
+      "loss": 0.8017,
+      "step": 8163
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016247668804383964,
+      "loss": 0.7792,
+      "step": 8164
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016246809755804224,
+      "loss": 0.8439,
+      "step": 8165
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016245950631617266,
+      "loss": 0.7885,
+      "step": 8166
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016245091431833485,
+      "loss": 0.7386,
+      "step": 8167
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016244232156463278,
+      "loss": 0.8148,
+      "step": 8168
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016243372805517048,
+      "loss": 0.8015,
+      "step": 8169
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016242513379005193,
+      "loss": 0.7334,
+      "step": 8170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016241653876938118,
+      "loss": 0.5909,
+      "step": 8171
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016240794299326227,
+      "loss": 0.8422,
+      "step": 8172
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016239934646179922,
+      "loss": 0.7445,
+      "step": 8173
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016239074917509606,
+      "loss": 0.7427,
+      "step": 8174
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016238215113325682,
+      "loss": 0.8269,
+      "step": 8175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016237355233638567,
+      "loss": 0.7428,
+      "step": 8176
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016236495278458657,
+      "loss": 0.776,
+      "step": 8177
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016235635247796365,
+      "loss": 0.6963,
+      "step": 8178
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016234775141662103,
+      "loss": 0.7273,
+      "step": 8179
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016233914960066276,
+      "loss": 0.7376,
+      "step": 8180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000162330547030193,
+      "loss": 0.7795,
+      "step": 8181
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016232194370531584,
+      "loss": 0.8609,
+      "step": 8182
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001623133396261354,
+      "loss": 0.7711,
+      "step": 8183
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016230473479275585,
+      "loss": 0.7033,
+      "step": 8184
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001622961292052813,
+      "loss": 0.676,
+      "step": 8185
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001622875228638159,
+      "loss": 0.7067,
+      "step": 8186
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016227891576846393,
+      "loss": 0.855,
+      "step": 8187
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001622703079193294,
+      "loss": 0.7767,
+      "step": 8188
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001622616993165166,
+      "loss": 0.6684,
+      "step": 8189
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016225308996012968,
+      "loss": 0.8339,
+      "step": 8190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016224447985027288,
+      "loss": 0.8763,
+      "step": 8191
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001622358689870503,
+      "loss": 0.7811,
+      "step": 8192
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016222725737056632,
+      "loss": 0.7135,
+      "step": 8193
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001622186450009251,
+      "loss": 0.8409,
+      "step": 8194
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016221003187823088,
+      "loss": 0.7514,
+      "step": 8195
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016220141800258784,
+      "loss": 0.7881,
+      "step": 8196
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016219280337410034,
+      "loss": 0.8501,
+      "step": 8197
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016218418799287258,
+      "loss": 0.6892,
+      "step": 8198
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001621755718590089,
+      "loss": 0.6724,
+      "step": 8199
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016216695497261348,
+      "loss": 0.8378,
+      "step": 8200
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 0.7487373352050781,
+      "eval_runtime": 847.1108,
+      "eval_samples_per_second": 28.402,
+      "eval_steps_per_second": 14.201,
+      "step": 8200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001621583373337907,
+      "loss": 0.8322,
+      "step": 8201
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016214971894264486,
+      "loss": 0.7669,
+      "step": 8202
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016214109979928025,
+      "loss": 0.7055,
+      "step": 8203
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016213247990380116,
+      "loss": 0.7119,
+      "step": 8204
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016212385925631198,
+      "loss": 0.8403,
+      "step": 8205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016211523785691697,
+      "loss": 0.831,
+      "step": 8206
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016210661570572053,
+      "loss": 0.7332,
+      "step": 8207
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016209799280282705,
+      "loss": 0.8016,
+      "step": 8208
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016208936914834086,
+      "loss": 0.7171,
+      "step": 8209
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016208074474236627,
+      "loss": 0.7313,
+      "step": 8210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016207211958500775,
+      "loss": 0.7447,
+      "step": 8211
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001620634936763697,
+      "loss": 0.7613,
+      "step": 8212
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016205486701655647,
+      "loss": 0.8395,
+      "step": 8213
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016204623960567252,
+      "loss": 0.7829,
+      "step": 8214
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001620376114438222,
+      "loss": 0.7775,
+      "step": 8215
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016202898253110998,
+      "loss": 0.8217,
+      "step": 8216
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016202035286764032,
+      "loss": 0.7841,
+      "step": 8217
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016201172245351766,
+      "loss": 0.8313,
+      "step": 8218
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016200309128884644,
+      "loss": 0.7142,
+      "step": 8219
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016199445937373113,
+      "loss": 0.674,
+      "step": 8220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016198582670827618,
+      "loss": 0.6849,
+      "step": 8221
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016197719329258612,
+      "loss": 0.7902,
+      "step": 8222
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016196855912676544,
+      "loss": 0.7998,
+      "step": 8223
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001619599242109186,
+      "loss": 0.8162,
+      "step": 8224
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016195128854515014,
+      "loss": 0.7032,
+      "step": 8225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016194265212956458,
+      "loss": 0.755,
+      "step": 8226
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016193401496426645,
+      "loss": 0.6581,
+      "step": 8227
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016192537704936029,
+      "loss": 0.7698,
+      "step": 8228
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016191673838495066,
+      "loss": 0.8353,
+      "step": 8229
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016190809897114204,
+      "loss": 0.8716,
+      "step": 8230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001618994588080391,
+      "loss": 0.7643,
+      "step": 8231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001618908178957464,
+      "loss": 0.7341,
+      "step": 8232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016188217623436843,
+      "loss": 0.8699,
+      "step": 8233
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016187353382400986,
+      "loss": 0.6853,
+      "step": 8234
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016186489066477528,
+      "loss": 0.7792,
+      "step": 8235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016185624675676933,
+      "loss": 0.7735,
+      "step": 8236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016184760210009657,
+      "loss": 0.8713,
+      "step": 8237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016183895669486166,
+      "loss": 0.6961,
+      "step": 8238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016183031054116924,
+      "loss": 0.8471,
+      "step": 8239
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016182166363912397,
+      "loss": 0.7307,
+      "step": 8240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016181301598883048,
+      "loss": 0.7705,
+      "step": 8241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016180436759039344,
+      "loss": 0.7519,
+      "step": 8242
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001617957184439175,
+      "loss": 0.7589,
+      "step": 8243
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016178706854950745,
+      "loss": 0.8588,
+      "step": 8244
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016177841790726785,
+      "loss": 0.7579,
+      "step": 8245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016176976651730346,
+      "loss": 0.7598,
+      "step": 8246
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000161761114379719,
+      "loss": 0.784,
+      "step": 8247
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001617524614946192,
+      "loss": 0.7309,
+      "step": 8248
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016174380786210875,
+      "loss": 0.7671,
+      "step": 8249
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001617351534822924,
+      "loss": 0.7323,
+      "step": 8250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001617264983552749,
+      "loss": 0.7362,
+      "step": 8251
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016171784248116103,
+      "loss": 0.6727,
+      "step": 8252
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001617091858600555,
+      "loss": 0.7342,
+      "step": 8253
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016170052849206315,
+      "loss": 0.7796,
+      "step": 8254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016169187037728868,
+      "loss": 0.627,
+      "step": 8255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016168321151583697,
+      "loss": 0.6448,
+      "step": 8256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016167455190781277,
+      "loss": 0.8894,
+      "step": 8257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016166589155332093,
+      "loss": 0.6518,
+      "step": 8258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016165723045246622,
+      "loss": 0.7422,
+      "step": 8259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001616485686053535,
+      "loss": 0.7565,
+      "step": 8260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016163990601208755,
+      "loss": 0.6413,
+      "step": 8261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001616312426727733,
+      "loss": 0.784,
+      "step": 8262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001616225785875156,
+      "loss": 0.6819,
+      "step": 8263
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016161391375641926,
+      "loss": 0.9238,
+      "step": 8264
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016160524817958915,
+      "loss": 0.7718,
+      "step": 8265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001615965818571302,
+      "loss": 0.7434,
+      "step": 8266
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001615879147891473,
+      "loss": 0.7615,
+      "step": 8267
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001615792469757453,
+      "loss": 0.7459,
+      "step": 8268
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016157057841702916,
+      "loss": 0.8033,
+      "step": 8269
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016156190911310376,
+      "loss": 0.8972,
+      "step": 8270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016155323906407408,
+      "loss": 0.7222,
+      "step": 8271
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000161544568270045,
+      "loss": 0.7922,
+      "step": 8272
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001615358967311215,
+      "loss": 0.7294,
+      "step": 8273
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016152722444740848,
+      "loss": 0.7253,
+      "step": 8274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016151855141901096,
+      "loss": 0.7892,
+      "step": 8275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016150987764603395,
+      "loss": 0.7566,
+      "step": 8276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016150120312858232,
+      "loss": 0.6911,
+      "step": 8277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016149252786676116,
+      "loss": 0.76,
+      "step": 8278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001614838518606754,
+      "loss": 0.7135,
+      "step": 8279
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001614751751104301,
+      "loss": 0.79,
+      "step": 8280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016146649761613024,
+      "loss": 0.7802,
+      "step": 8281
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001614578193778809,
+      "loss": 0.7035,
+      "step": 8282
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016144914039578704,
+      "loss": 0.7217,
+      "step": 8283
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016144046066995376,
+      "loss": 0.6531,
+      "step": 8284
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001614317802004861,
+      "loss": 0.7543,
+      "step": 8285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001614230989874891,
+      "loss": 0.9127,
+      "step": 8286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001614144170310679,
+      "loss": 0.7756,
+      "step": 8287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016140573433132748,
+      "loss": 0.7517,
+      "step": 8288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000161397050888373,
+      "loss": 0.744,
+      "step": 8289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016138836670230954,
+      "loss": 0.7954,
+      "step": 8290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001613796817732422,
+      "loss": 0.7664,
+      "step": 8291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016137099610127612,
+      "loss": 0.6658,
+      "step": 8292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001613623096865164,
+      "loss": 0.6617,
+      "step": 8293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001613536225290682,
+      "loss": 0.9434,
+      "step": 8294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016134493462903663,
+      "loss": 0.7747,
+      "step": 8295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016133624598652684,
+      "loss": 0.768,
+      "step": 8296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016132755660164407,
+      "loss": 0.7252,
+      "step": 8297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001613188664744934,
+      "loss": 0.7192,
+      "step": 8298
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016131017560518004,
+      "loss": 0.8596,
+      "step": 8299
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016130148399380918,
+      "loss": 0.7424,
+      "step": 8300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016129279164048602,
+      "loss": 0.8533,
+      "step": 8301
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016128409854531577,
+      "loss": 0.784,
+      "step": 8302
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016127540470840365,
+      "loss": 0.6961,
+      "step": 8303
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016126671012985485,
+      "loss": 0.6872,
+      "step": 8304
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016125801480977464,
+      "loss": 0.6892,
+      "step": 8305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016124931874826826,
+      "loss": 0.7755,
+      "step": 8306
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016124062194544094,
+      "loss": 0.8024,
+      "step": 8307
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016123192440139797,
+      "loss": 0.7889,
+      "step": 8308
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016122322611624457,
+      "loss": 0.7705,
+      "step": 8309
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016121452709008606,
+      "loss": 0.8351,
+      "step": 8310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016120582732302773,
+      "loss": 0.8385,
+      "step": 8311
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016119712681517484,
+      "loss": 0.6326,
+      "step": 8312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016118842556663273,
+      "loss": 0.698,
+      "step": 8313
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016117972357750673,
+      "loss": 0.6968,
+      "step": 8314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001611710208479021,
+      "loss": 0.6879,
+      "step": 8315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001611623173779242,
+      "loss": 0.7905,
+      "step": 8316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001611536131676784,
+      "loss": 0.8373,
+      "step": 8317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016114490821727005,
+      "loss": 0.6341,
+      "step": 8318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016113620252680445,
+      "loss": 0.8127,
+      "step": 8319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000161127496096387,
+      "loss": 0.7029,
+      "step": 8320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001611187889261231,
+      "loss": 0.6705,
+      "step": 8321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001611100810161181,
+      "loss": 0.9087,
+      "step": 8322
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016110137236647744,
+      "loss": 0.8363,
+      "step": 8323
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016109266297730649,
+      "loss": 0.7062,
+      "step": 8324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016108395284871065,
+      "loss": 0.7184,
+      "step": 8325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016107524198079537,
+      "loss": 0.7994,
+      "step": 8326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001610665303736661,
+      "loss": 0.6649,
+      "step": 8327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001610578180274282,
+      "loss": 0.718,
+      "step": 8328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001610491049421872,
+      "loss": 0.8094,
+      "step": 8329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016104039111804853,
+      "loss": 0.7168,
+      "step": 8330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001610316765551176,
+      "loss": 0.7665,
+      "step": 8331
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001610229612535,
+      "loss": 0.709,
+      "step": 8332
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016101424521330112,
+      "loss": 0.8566,
+      "step": 8333
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001610055284346265,
+      "loss": 0.8022,
+      "step": 8334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016099681091758162,
+      "loss": 0.7585,
+      "step": 8335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000160988092662272,
+      "loss": 0.7612,
+      "step": 8336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016097937366880318,
+      "loss": 0.726,
+      "step": 8337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016097065393728065,
+      "loss": 0.7713,
+      "step": 8338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016096193346780993,
+      "loss": 0.908,
+      "step": 8339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016095321226049664,
+      "loss": 0.7423,
+      "step": 8340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001609444903154463,
+      "loss": 0.8225,
+      "step": 8341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016093576763276444,
+      "loss": 0.7689,
+      "step": 8342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016092704421255668,
+      "loss": 0.8501,
+      "step": 8343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016091832005492862,
+      "loss": 0.6927,
+      "step": 8344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001609095951599858,
+      "loss": 0.6989,
+      "step": 8345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001609008695278338,
+      "loss": 0.6892,
+      "step": 8346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016089214315857833,
+      "loss": 0.8762,
+      "step": 8347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001608834160523249,
+      "loss": 0.6656,
+      "step": 8348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001608746882091792,
+      "loss": 0.6907,
+      "step": 8349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016086595962924686,
+      "loss": 0.7468,
+      "step": 8350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001608572303126335,
+      "loss": 0.7615,
+      "step": 8351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001608485002594448,
+      "loss": 0.7077,
+      "step": 8352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001608397694697864,
+      "loss": 0.8116,
+      "step": 8353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000160831037943764,
+      "loss": 0.7845,
+      "step": 8354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016082230568148322,
+      "loss": 0.8111,
+      "step": 8355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016081357268304985,
+      "loss": 0.8169,
+      "step": 8356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001608048389485695,
+      "loss": 0.8848,
+      "step": 8357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001607961044781479,
+      "loss": 0.755,
+      "step": 8358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001607873692718908,
+      "loss": 0.8312,
+      "step": 8359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001607786333299039,
+      "loss": 0.7596,
+      "step": 8360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016076989665229295,
+      "loss": 0.7263,
+      "step": 8361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016076115923916362,
+      "loss": 0.8086,
+      "step": 8362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016075242109062177,
+      "loss": 0.7077,
+      "step": 8363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016074368220677306,
+      "loss": 0.9911,
+      "step": 8364
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016073494258772336,
+      "loss": 0.7826,
+      "step": 8365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001607262022335784,
+      "loss": 0.7054,
+      "step": 8366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016071746114444392,
+      "loss": 0.816,
+      "step": 8367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016070871932042578,
+      "loss": 0.7435,
+      "step": 8368
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001606999767616298,
+      "loss": 0.7934,
+      "step": 8369
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001606912334681617,
+      "loss": 0.7305,
+      "step": 8370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016068248944012741,
+      "loss": 0.6569,
+      "step": 8371
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001606737446776327,
+      "loss": 0.6924,
+      "step": 8372
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016066499918078342,
+      "loss": 0.7445,
+      "step": 8373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016065625294968544,
+      "loss": 0.9096,
+      "step": 8374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001606475059844446,
+      "loss": 0.7475,
+      "step": 8375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016063875828516676,
+      "loss": 0.7484,
+      "step": 8376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016063000985195783,
+      "loss": 0.8209,
+      "step": 8377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016062126068492366,
+      "loss": 0.7852,
+      "step": 8378
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016061251078417017,
+      "loss": 0.8619,
+      "step": 8379
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016060376014980326,
+      "loss": 0.7932,
+      "step": 8380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001605950087819288,
+      "loss": 0.7274,
+      "step": 8381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016058625668065275,
+      "loss": 0.6823,
+      "step": 8382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016057750384608106,
+      "loss": 0.6906,
+      "step": 8383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001605687502783196,
+      "loss": 0.7374,
+      "step": 8384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001605599959774744,
+      "loss": 0.6817,
+      "step": 8385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016055124094365133,
+      "loss": 0.7467,
+      "step": 8386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016054248517695645,
+      "loss": 0.8596,
+      "step": 8387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016053372867749565,
+      "loss": 0.8161,
+      "step": 8388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016052497144537498,
+      "loss": 0.6993,
+      "step": 8389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016051621348070033,
+      "loss": 0.8059,
+      "step": 8390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016050745478357782,
+      "loss": 0.6621,
+      "step": 8391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016049869535411345,
+      "loss": 0.711,
+      "step": 8392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016048993519241313,
+      "loss": 0.7629,
+      "step": 8393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016048117429858297,
+      "loss": 0.6447,
+      "step": 8394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000160472412672729,
+      "loss": 0.753,
+      "step": 8395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016046365031495725,
+      "loss": 0.6464,
+      "step": 8396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016045488722537377,
+      "loss": 0.7754,
+      "step": 8397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016044612340408466,
+      "loss": 0.6515,
+      "step": 8398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016043735885119592,
+      "loss": 0.7831,
+      "step": 8399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016042859356681368,
+      "loss": 0.6804,
+      "step": 8400
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 0.7389582395553589,
+      "eval_runtime": 845.0308,
+      "eval_samples_per_second": 28.472,
+      "eval_steps_per_second": 14.236,
+      "step": 8400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016041982755104405,
+      "loss": 0.7157,
+      "step": 8401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016041106080399309,
+      "loss": 0.6699,
+      "step": 8402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016040229332576691,
+      "loss": 0.6415,
+      "step": 8403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016039352511647162,
+      "loss": 0.7075,
+      "step": 8404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001603847561762134,
+      "loss": 0.6814,
+      "step": 8405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001603759865050983,
+      "loss": 0.8936,
+      "step": 8406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016036721610323252,
+      "loss": 0.6906,
+      "step": 8407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001603584449707222,
+      "loss": 0.7066,
+      "step": 8408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001603496731076735,
+      "loss": 0.7254,
+      "step": 8409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016034090051419257,
+      "loss": 0.8299,
+      "step": 8410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001603321271903856,
+      "loss": 0.8019,
+      "step": 8411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016032335313635878,
+      "loss": 0.7746,
+      "step": 8412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016031457835221827,
+      "loss": 0.675,
+      "step": 8413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016030580283807035,
+      "loss": 0.7124,
+      "step": 8414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016029702659402117,
+      "loss": 0.6466,
+      "step": 8415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016028824962017695,
+      "loss": 0.6163,
+      "step": 8416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000160279471916644,
+      "loss": 0.6717,
+      "step": 8417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016027069348352845,
+      "loss": 0.7091,
+      "step": 8418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016026191432093662,
+      "loss": 0.6736,
+      "step": 8419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016025313442897475,
+      "loss": 0.8486,
+      "step": 8420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016024435380774908,
+      "loss": 0.7407,
+      "step": 8421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016023557245736593,
+      "loss": 0.6946,
+      "step": 8422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016022679037793155,
+      "loss": 0.7505,
+      "step": 8423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016021800756955225,
+      "loss": 0.6799,
+      "step": 8424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016020922403233435,
+      "loss": 0.7492,
+      "step": 8425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001602004397663841,
+      "loss": 0.7285,
+      "step": 8426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016019165477180786,
+      "loss": 0.9025,
+      "step": 8427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016018286904871193,
+      "loss": 0.7534,
+      "step": 8428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016017408259720272,
+      "loss": 0.7541,
+      "step": 8429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001601652954173865,
+      "loss": 0.7458,
+      "step": 8430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001601565075093696,
+      "loss": 0.6478,
+      "step": 8431
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016014771887325847,
+      "loss": 0.8409,
+      "step": 8432
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016013892950915947,
+      "loss": 0.8391,
+      "step": 8433
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016013013941717895,
+      "loss": 0.7382,
+      "step": 8434
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016012134859742325,
+      "loss": 0.7362,
+      "step": 8435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016011255704999884,
+      "loss": 0.7824,
+      "step": 8436
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016010376477501213,
+      "loss": 0.7549,
+      "step": 8437
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016009497177256948,
+      "loss": 0.7057,
+      "step": 8438
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001600861780427774,
+      "loss": 0.6665,
+      "step": 8439
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016007738358574222,
+      "loss": 0.6584,
+      "step": 8440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016006858840157047,
+      "loss": 0.7983,
+      "step": 8441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016005979249036854,
+      "loss": 0.7661,
+      "step": 8442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016005099585224291,
+      "loss": 0.7254,
+      "step": 8443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016004219848730008,
+      "loss": 0.8556,
+      "step": 8444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016003340039564648,
+      "loss": 0.8017,
+      "step": 8445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001600246015773886,
+      "loss": 0.6044,
+      "step": 8446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000160015802032633,
+      "loss": 0.829,
+      "step": 8447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001600070017614861,
+      "loss": 0.7718,
+      "step": 8448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015999820076405448,
+      "loss": 0.7576,
+      "step": 8449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001599893990404446,
+      "loss": 0.7252,
+      "step": 8450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015998059659076303,
+      "loss": 0.7937,
+      "step": 8451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001599717934151163,
+      "loss": 0.7312,
+      "step": 8452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015996298951361095,
+      "loss": 0.7191,
+      "step": 8453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015995418488635353,
+      "loss": 0.6887,
+      "step": 8454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015994537953345066,
+      "loss": 0.7294,
+      "step": 8455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015993657345500886,
+      "loss": 0.8791,
+      "step": 8456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001599277666511347,
+      "loss": 0.7667,
+      "step": 8457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001599189591219348,
+      "loss": 0.6896,
+      "step": 8458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015991015086751578,
+      "loss": 0.7426,
+      "step": 8459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015990134188798423,
+      "loss": 0.875,
+      "step": 8460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015989253218344675,
+      "loss": 0.7608,
+      "step": 8461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015988372175400998,
+      "loss": 0.7917,
+      "step": 8462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015987491059978054,
+      "loss": 0.6803,
+      "step": 8463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001598660987208651,
+      "loss": 0.764,
+      "step": 8464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015985728611737034,
+      "loss": 0.6833,
+      "step": 8465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015984847278940287,
+      "loss": 0.7614,
+      "step": 8466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015983965873706937,
+      "loss": 0.8396,
+      "step": 8467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015983084396047653,
+      "loss": 0.8639,
+      "step": 8468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015982202845973105,
+      "loss": 0.6823,
+      "step": 8469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015981321223493958,
+      "loss": 0.822,
+      "step": 8470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015980439528620887,
+      "loss": 0.6612,
+      "step": 8471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015979557761364562,
+      "loss": 0.7114,
+      "step": 8472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015978675921735658,
+      "loss": 0.88,
+      "step": 8473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015977794009744845,
+      "loss": 0.8035,
+      "step": 8474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015976912025402796,
+      "loss": 0.6778,
+      "step": 8475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001597602996872019,
+      "loss": 0.7324,
+      "step": 8476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015975147839707697,
+      "loss": 0.7401,
+      "step": 8477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015974265638375997,
+      "loss": 0.7479,
+      "step": 8478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015973383364735773,
+      "loss": 0.7409,
+      "step": 8479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015972501018797695,
+      "loss": 0.6951,
+      "step": 8480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015971618600572445,
+      "loss": 0.8301,
+      "step": 8481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015970736110070704,
+      "loss": 0.8354,
+      "step": 8482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015969853547303156,
+      "loss": 0.7403,
+      "step": 8483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015968970912280477,
+      "loss": 0.7305,
+      "step": 8484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015968088205013355,
+      "loss": 0.7541,
+      "step": 8485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015967205425512467,
+      "loss": 0.8065,
+      "step": 8486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015966322573788506,
+      "loss": 0.7235,
+      "step": 8487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015965439649852153,
+      "loss": 0.7744,
+      "step": 8488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015964556653714094,
+      "loss": 0.6757,
+      "step": 8489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015963673585385016,
+      "loss": 0.7107,
+      "step": 8490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001596279044487561,
+      "loss": 0.7569,
+      "step": 8491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001596190723219656,
+      "loss": 0.7617,
+      "step": 8492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001596102394735856,
+      "loss": 0.6989,
+      "step": 8493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000159601405903723,
+      "loss": 0.8243,
+      "step": 8494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015959257161248473,
+      "loss": 0.6615,
+      "step": 8495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015958373659997766,
+      "loss": 0.8584,
+      "step": 8496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015957490086630877,
+      "loss": 0.8134,
+      "step": 8497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015956606441158502,
+      "loss": 0.7382,
+      "step": 8498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015955722723591328,
+      "loss": 0.7814,
+      "step": 8499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015954838933940056,
+      "loss": 0.7296,
+      "step": 8500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015953955072215385,
+      "loss": 0.8279,
+      "step": 8501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015953071138428014,
+      "loss": 0.8089,
+      "step": 8502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001595218713258863,
+      "loss": 0.7136,
+      "step": 8503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015951303054707948,
+      "loss": 0.8381,
+      "step": 8504
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015950418904796658,
+      "loss": 0.7815,
+      "step": 8505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001594953468286546,
+      "loss": 0.7174,
+      "step": 8506
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015948650388925066,
+      "loss": 0.6491,
+      "step": 8507
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015947766022986169,
+      "loss": 0.7103,
+      "step": 8508
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015946881585059478,
+      "loss": 0.7031,
+      "step": 8509
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015945997075155694,
+      "loss": 0.7611,
+      "step": 8510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015945112493285528,
+      "loss": 0.7877,
+      "step": 8511
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001594422783945968,
+      "loss": 0.7171,
+      "step": 8512
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015943343113688862,
+      "loss": 0.8212,
+      "step": 8513
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001594245831598378,
+      "loss": 0.8636,
+      "step": 8514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015941573446355144,
+      "loss": 0.6458,
+      "step": 8515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015940688504813662,
+      "loss": 0.8482,
+      "step": 8516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015939803491370048,
+      "loss": 0.7366,
+      "step": 8517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001593891840603501,
+      "loss": 0.7289,
+      "step": 8518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001593803324881926,
+      "loss": 0.7933,
+      "step": 8519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001593714801973352,
+      "loss": 0.6949,
+      "step": 8520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015936262718788496,
+      "loss": 0.7604,
+      "step": 8521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015935377345994902,
+      "loss": 0.7342,
+      "step": 8522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015934491901363457,
+      "loss": 0.8166,
+      "step": 8523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001593360638490488,
+      "loss": 0.6918,
+      "step": 8524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015932720796629885,
+      "loss": 0.7146,
+      "step": 8525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015931835136549194,
+      "loss": 0.766,
+      "step": 8526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001593094940467352,
+      "loss": 0.6606,
+      "step": 8527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015930063601013588,
+      "loss": 0.7582,
+      "step": 8528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015929177725580121,
+      "loss": 0.7806,
+      "step": 8529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001592829177838384,
+      "loss": 0.722,
+      "step": 8530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015927405759435468,
+      "loss": 0.737,
+      "step": 8531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015926519668745724,
+      "loss": 0.7559,
+      "step": 8532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015925633506325338,
+      "loss": 0.9812,
+      "step": 8533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015924747272185035,
+      "loss": 0.8583,
+      "step": 8534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015923860966335536,
+      "loss": 0.7172,
+      "step": 8535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015922974588787576,
+      "loss": 0.7601,
+      "step": 8536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015922088139551876,
+      "loss": 0.8513,
+      "step": 8537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015921201618639174,
+      "loss": 0.6837,
+      "step": 8538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001592031502606019,
+      "loss": 0.7294,
+      "step": 8539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015919428361825663,
+      "loss": 0.7408,
+      "step": 8540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015918541625946318,
+      "loss": 0.721,
+      "step": 8541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001591765481843289,
+      "loss": 0.674,
+      "step": 8542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015916767939296112,
+      "loss": 0.7064,
+      "step": 8543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015915880988546719,
+      "loss": 0.7808,
+      "step": 8544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015914993966195445,
+      "loss": 0.7754,
+      "step": 8545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015914106872253028,
+      "loss": 0.8003,
+      "step": 8546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015913219706730205,
+      "loss": 0.7021,
+      "step": 8547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001591233246963771,
+      "loss": 0.7721,
+      "step": 8548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015911445160986284,
+      "loss": 0.7204,
+      "step": 8549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015910557780786667,
+      "loss": 0.7566,
+      "step": 8550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015909670329049597,
+      "loss": 0.8163,
+      "step": 8551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015908782805785818,
+      "loss": 0.8047,
+      "step": 8552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015907895211006068,
+      "loss": 0.7139,
+      "step": 8553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015907007544721093,
+      "loss": 0.6707,
+      "step": 8554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015906119806941636,
+      "loss": 0.8187,
+      "step": 8555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015905231997678443,
+      "loss": 0.7422,
+      "step": 8556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015904344116942257,
+      "loss": 0.7649,
+      "step": 8557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015903456164743826,
+      "loss": 0.6344,
+      "step": 8558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015902568141093896,
+      "loss": 0.75,
+      "step": 8559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015901680046003216,
+      "loss": 0.7977,
+      "step": 8560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015900791879482531,
+      "loss": 0.7932,
+      "step": 8561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015899903641542599,
+      "loss": 0.7565,
+      "step": 8562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015899015332194164,
+      "loss": 0.6355,
+      "step": 8563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001589812695144798,
+      "loss": 0.8391,
+      "step": 8564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000158972384993148,
+      "loss": 0.8492,
+      "step": 8565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015896349975805376,
+      "loss": 0.786,
+      "step": 8566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001589546138093046,
+      "loss": 0.6728,
+      "step": 8567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001589457271470081,
+      "loss": 0.7494,
+      "step": 8568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001589368397712718,
+      "loss": 0.647,
+      "step": 8569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015892795168220326,
+      "loss": 0.7621,
+      "step": 8570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015891906287991011,
+      "loss": 0.8127,
+      "step": 8571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001589101733644999,
+      "loss": 0.6677,
+      "step": 8572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015890128313608018,
+      "loss": 0.6824,
+      "step": 8573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001588923921947586,
+      "loss": 0.7288,
+      "step": 8574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015888350054064275,
+      "loss": 0.7049,
+      "step": 8575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015887460817384027,
+      "loss": 0.6655,
+      "step": 8576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001588657150944588,
+      "loss": 0.7982,
+      "step": 8577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015885682130260592,
+      "loss": 0.8598,
+      "step": 8578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015884792679838933,
+      "loss": 0.7315,
+      "step": 8579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015883903158191664,
+      "loss": 0.7608,
+      "step": 8580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015883013565329553,
+      "loss": 0.8462,
+      "step": 8581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015882123901263368,
+      "loss": 0.7683,
+      "step": 8582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015881234166003876,
+      "loss": 0.758,
+      "step": 8583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015880344359561844,
+      "loss": 0.6618,
+      "step": 8584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001587945448194805,
+      "loss": 0.7301,
+      "step": 8585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015878564533173251,
+      "loss": 0.6538,
+      "step": 8586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015877674513248226,
+      "loss": 0.6992,
+      "step": 8587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001587678442218375,
+      "loss": 0.6954,
+      "step": 8588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015875894259990587,
+      "loss": 0.7196,
+      "step": 8589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001587500402667952,
+      "loss": 0.8491,
+      "step": 8590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001587411372226132,
+      "loss": 0.6664,
+      "step": 8591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015873223346746763,
+      "loss": 0.7081,
+      "step": 8592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015872332900146625,
+      "loss": 0.7611,
+      "step": 8593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015871442382471687,
+      "loss": 0.7924,
+      "step": 8594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015870551793732717,
+      "loss": 0.6592,
+      "step": 8595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015869661133940504,
+      "loss": 0.6314,
+      "step": 8596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015868770403105827,
+      "loss": 0.7294,
+      "step": 8597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015867879601239466,
+      "loss": 0.7251,
+      "step": 8598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000158669887283522,
+      "loss": 0.712,
+      "step": 8599
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015866097784454812,
+      "loss": 0.6921,
+      "step": 8600
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 0.7307732701301575,
+      "eval_runtime": 844.7155,
+      "eval_samples_per_second": 28.483,
+      "eval_steps_per_second": 14.241,
+      "step": 8600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015865206769558084,
+      "loss": 0.7673,
+      "step": 8601
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015864315683672807,
+      "loss": 0.6944,
+      "step": 8602
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015863424526809763,
+      "loss": 0.6777,
+      "step": 8603
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015862533298979735,
+      "loss": 0.6815,
+      "step": 8604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001586164200019351,
+      "loss": 0.7906,
+      "step": 8605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001586075063046188,
+      "loss": 0.6688,
+      "step": 8606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015859859189795627,
+      "loss": 0.5681,
+      "step": 8607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015858967678205547,
+      "loss": 0.6749,
+      "step": 8608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015858076095702427,
+      "loss": 0.6714,
+      "step": 8609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001585718444229706,
+      "loss": 0.7469,
+      "step": 8610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015856292718000235,
+      "loss": 0.8347,
+      "step": 8611
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015855400922822746,
+      "loss": 0.6918,
+      "step": 8612
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015854509056775388,
+      "loss": 0.8152,
+      "step": 8613
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015853617119868956,
+      "loss": 0.7357,
+      "step": 8614
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015852725112114243,
+      "loss": 0.8438,
+      "step": 8615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015851833033522046,
+      "loss": 0.6463,
+      "step": 8616
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001585094088410316,
+      "loss": 0.6991,
+      "step": 8617
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001585004866386839,
+      "loss": 0.7508,
+      "step": 8618
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001584915637282853,
+      "loss": 0.7571,
+      "step": 8619
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015848264010994375,
+      "loss": 0.6982,
+      "step": 8620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015847371578376734,
+      "loss": 0.8079,
+      "step": 8621
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015846479074986406,
+      "loss": 0.8363,
+      "step": 8622
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015845586500834192,
+      "loss": 0.6997,
+      "step": 8623
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015844693855930893,
+      "loss": 0.6722,
+      "step": 8624
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015843801140287317,
+      "loss": 0.701,
+      "step": 8625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015842908353914264,
+      "loss": 0.7192,
+      "step": 8626
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001584201549682255,
+      "loss": 0.7531,
+      "step": 8627
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001584112256902297,
+      "loss": 0.8781,
+      "step": 8628
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015840229570526337,
+      "loss": 0.7842,
+      "step": 8629
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015839336501343457,
+      "loss": 0.8068,
+      "step": 8630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001583844336148514,
+      "loss": 0.7565,
+      "step": 8631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015837550150962196,
+      "loss": 0.772,
+      "step": 8632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001583665686978544,
+      "loss": 0.72,
+      "step": 8633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015835763517965673,
+      "loss": 0.8477,
+      "step": 8634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015834870095513716,
+      "loss": 0.8386,
+      "step": 8635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015833976602440384,
+      "loss": 0.6869,
+      "step": 8636
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015833083038756485,
+      "loss": 0.6566,
+      "step": 8637
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015832189404472836,
+      "loss": 0.7296,
+      "step": 8638
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001583129569960026,
+      "loss": 0.6653,
+      "step": 8639
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001583040192414956,
+      "loss": 0.6907,
+      "step": 8640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015829508078131567,
+      "loss": 0.6677,
+      "step": 8641
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015828614161557088,
+      "loss": 0.6526,
+      "step": 8642
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015827720174436954,
+      "loss": 0.7057,
+      "step": 8643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015826826116781978,
+      "loss": 0.6328,
+      "step": 8644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015825931988602983,
+      "loss": 0.7188,
+      "step": 8645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001582503778991079,
+      "loss": 0.812,
+      "step": 8646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001582414352071622,
+      "loss": 0.7177,
+      "step": 8647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000158232491810301,
+      "loss": 0.7157,
+      "step": 8648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015822354770863255,
+      "loss": 0.7071,
+      "step": 8649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015821460290226509,
+      "loss": 0.718,
+      "step": 8650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015820565739130686,
+      "loss": 0.7718,
+      "step": 8651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015819671117586615,
+      "loss": 0.81,
+      "step": 8652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015818776425605125,
+      "loss": 0.7394,
+      "step": 8653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015817881663197044,
+      "loss": 0.7549,
+      "step": 8654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000158169868303732,
+      "loss": 0.6472,
+      "step": 8655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015816091927144425,
+      "loss": 0.7525,
+      "step": 8656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015815196953521552,
+      "loss": 0.7574,
+      "step": 8657
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001581430190951541,
+      "loss": 0.7003,
+      "step": 8658
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015813406795136831,
+      "loss": 0.7158,
+      "step": 8659
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001581251161039665,
+      "loss": 0.721,
+      "step": 8660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015811616355305709,
+      "loss": 0.7357,
+      "step": 8661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015810721029874833,
+      "loss": 0.7058,
+      "step": 8662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015809825634114866,
+      "loss": 0.6826,
+      "step": 8663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015808930168036638,
+      "loss": 0.755,
+      "step": 8664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015808034631650992,
+      "loss": 0.7733,
+      "step": 8665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015807139024968768,
+      "loss": 0.7797,
+      "step": 8666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015806243348000804,
+      "loss": 0.8046,
+      "step": 8667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001580534760075794,
+      "loss": 0.6108,
+      "step": 8668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015804451783251014,
+      "loss": 0.8705,
+      "step": 8669
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015803555895490878,
+      "loss": 0.8084,
+      "step": 8670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015802659937488368,
+      "loss": 0.6927,
+      "step": 8671
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015801763909254333,
+      "loss": 0.6428,
+      "step": 8672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015800867810799611,
+      "loss": 0.6918,
+      "step": 8673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015799971642135052,
+      "loss": 0.7685,
+      "step": 8674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015799075403271506,
+      "loss": 0.7513,
+      "step": 8675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015798179094219812,
+      "loss": 0.7395,
+      "step": 8676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015797282714990823,
+      "loss": 0.7533,
+      "step": 8677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015796386265595392,
+      "loss": 0.6869,
+      "step": 8678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015795489746044363,
+      "loss": 0.7832,
+      "step": 8679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015794593156348588,
+      "loss": 0.7197,
+      "step": 8680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001579369649651892,
+      "loss": 0.6338,
+      "step": 8681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015792799766566214,
+      "loss": 0.8052,
+      "step": 8682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001579190296650132,
+      "loss": 0.7657,
+      "step": 8683
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015791006096335091,
+      "loss": 0.7339,
+      "step": 8684
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015790109156078386,
+      "loss": 0.8014,
+      "step": 8685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001578921214574206,
+      "loss": 0.7393,
+      "step": 8686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015788315065336965,
+      "loss": 0.9278,
+      "step": 8687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015787417914873967,
+      "loss": 0.6602,
+      "step": 8688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015786520694363918,
+      "loss": 0.7326,
+      "step": 8689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001578562340381768,
+      "loss": 0.7433,
+      "step": 8690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015784726043246115,
+      "loss": 0.6405,
+      "step": 8691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015783828612660075,
+      "loss": 0.8028,
+      "step": 8692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015782931112070435,
+      "loss": 0.7313,
+      "step": 8693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015782033541488047,
+      "loss": 0.7825,
+      "step": 8694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015781135900923784,
+      "loss": 0.7855,
+      "step": 8695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015780238190388503,
+      "loss": 0.7807,
+      "step": 8696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001577934040989307,
+      "loss": 0.7027,
+      "step": 8697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015778442559448357,
+      "loss": 0.8425,
+      "step": 8698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015777544639065223,
+      "loss": 0.7228,
+      "step": 8699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015776646648754542,
+      "loss": 0.7526,
+      "step": 8700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001577574858852718,
+      "loss": 0.7079,
+      "step": 8701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015774850458394007,
+      "loss": 0.7882,
+      "step": 8702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001577395225836589,
+      "loss": 0.7957,
+      "step": 8703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001577305398845371,
+      "loss": 0.7673,
+      "step": 8704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015772155648668328,
+      "loss": 0.7846,
+      "step": 8705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015771257239020623,
+      "loss": 0.8685,
+      "step": 8706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015770358759521465,
+      "loss": 0.8852,
+      "step": 8707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015769460210181732,
+      "loss": 0.6958,
+      "step": 8708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000157685615910123,
+      "loss": 0.6532,
+      "step": 8709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001576766290202404,
+      "loss": 0.7069,
+      "step": 8710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015766764143227837,
+      "loss": 0.8379,
+      "step": 8711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015765865314634563,
+      "loss": 0.6992,
+      "step": 8712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000157649664162551,
+      "loss": 0.8232,
+      "step": 8713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001576406744810032,
+      "loss": 0.733,
+      "step": 8714
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00015763168410181115,
+      "loss": 0.7454,
+      "step": 8715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001576226930250836,
+      "loss": 0.7478,
+      "step": 8716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001576137012509294,
+      "loss": 0.7262,
+      "step": 8717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015760470877945732,
+      "loss": 0.848,
+      "step": 8718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015759571561077627,
+      "loss": 0.7784,
+      "step": 8719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015758672174499507,
+      "loss": 0.6387,
+      "step": 8720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015757772718222256,
+      "loss": 0.6904,
+      "step": 8721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015756873192256763,
+      "loss": 0.8478,
+      "step": 8722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015755973596613912,
+      "loss": 0.7992,
+      "step": 8723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015755073931304593,
+      "loss": 0.7793,
+      "step": 8724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000157541741963397,
+      "loss": 0.6803,
+      "step": 8725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015753274391730116,
+      "loss": 0.817,
+      "step": 8726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015752374517486733,
+      "loss": 0.734,
+      "step": 8727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001575147457362044,
+      "loss": 0.8073,
+      "step": 8728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015750574560142137,
+      "loss": 0.7105,
+      "step": 8729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015749674477062709,
+      "loss": 0.7062,
+      "step": 8730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015748774324393056,
+      "loss": 0.711,
+      "step": 8731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001574787410214407,
+      "loss": 0.8303,
+      "step": 8732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015746973810326647,
+      "loss": 0.7381,
+      "step": 8733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015746073448951684,
+      "loss": 0.717,
+      "step": 8734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015745173018030077,
+      "loss": 0.6496,
+      "step": 8735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015744272517572724,
+      "loss": 0.6869,
+      "step": 8736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001574337194759053,
+      "loss": 0.7299,
+      "step": 8737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015742471308094388,
+      "loss": 0.689,
+      "step": 8738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015741570599095197,
+      "loss": 0.6808,
+      "step": 8739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015740669820603866,
+      "loss": 0.7782,
+      "step": 8740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001573976897263129,
+      "loss": 0.7165,
+      "step": 8741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015738868055188384,
+      "loss": 0.716,
+      "step": 8742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001573796706828604,
+      "loss": 0.751,
+      "step": 8743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001573706601193517,
+      "loss": 0.6807,
+      "step": 8744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015736164886146674,
+      "loss": 0.7421,
+      "step": 8745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001573526369093146,
+      "loss": 0.7081,
+      "step": 8746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001573436242630044,
+      "loss": 0.8252,
+      "step": 8747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015733461092264518,
+      "loss": 0.7279,
+      "step": 8748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015732559688834606,
+      "loss": 0.7111,
+      "step": 8749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001573165821602161,
+      "loss": 0.7213,
+      "step": 8750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015730756673836447,
+      "loss": 0.6584,
+      "step": 8751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015729855062290022,
+      "loss": 0.6939,
+      "step": 8752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015728953381393253,
+      "loss": 0.8201,
+      "step": 8753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015728051631157047,
+      "loss": 0.8255,
+      "step": 8754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015727149811592324,
+      "loss": 0.7181,
+      "step": 8755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001572624792271,
+      "loss": 0.7462,
+      "step": 8756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015725345964520985,
+      "loss": 0.8345,
+      "step": 8757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000157244439370362,
+      "loss": 0.7084,
+      "step": 8758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001572354184026656,
+      "loss": 0.6384,
+      "step": 8759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015722639674222986,
+      "loss": 0.7511,
+      "step": 8760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015721737438916395,
+      "loss": 0.7373,
+      "step": 8761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001572083513435771,
+      "loss": 0.648,
+      "step": 8762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015719932760557845,
+      "loss": 0.7326,
+      "step": 8763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015719030317527731,
+      "loss": 0.725,
+      "step": 8764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015718127805278284,
+      "loss": 0.6724,
+      "step": 8765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015717225223820427,
+      "loss": 0.7751,
+      "step": 8766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015716322573165091,
+      "loss": 0.6893,
+      "step": 8767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015715419853323196,
+      "loss": 0.7404,
+      "step": 8768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015714517064305668,
+      "loss": 0.6591,
+      "step": 8769
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015713614206123435,
+      "loss": 0.7079,
+      "step": 8770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015712711278787422,
+      "loss": 0.8174,
+      "step": 8771
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001571180828230856,
+      "loss": 0.657,
+      "step": 8772
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001571090521669778,
+      "loss": 0.8794,
+      "step": 8773
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001571000208196601,
+      "loss": 0.6825,
+      "step": 8774
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015709098878124178,
+      "loss": 0.7152,
+      "step": 8775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001570819560518322,
+      "loss": 0.5998,
+      "step": 8776
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015707292263154068,
+      "loss": 0.8222,
+      "step": 8777
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015706388852047652,
+      "loss": 0.8742,
+      "step": 8778
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001570548537187491,
+      "loss": 0.7795,
+      "step": 8779
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001570458182264678,
+      "loss": 0.7268,
+      "step": 8780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001570367820437419,
+      "loss": 0.7832,
+      "step": 8781
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015702774517068082,
+      "loss": 0.7196,
+      "step": 8782
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001570187076073939,
+      "loss": 0.6487,
+      "step": 8783
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015700966935399058,
+      "loss": 0.6367,
+      "step": 8784
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015700063041058023,
+      "loss": 0.8379,
+      "step": 8785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001569915907772722,
+      "loss": 0.7629,
+      "step": 8786
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015698255045417598,
+      "loss": 0.8107,
+      "step": 8787
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015697350944140093,
+      "loss": 0.6005,
+      "step": 8788
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015696446773905652,
+      "loss": 0.6937,
+      "step": 8789
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015695542534725215,
+      "loss": 0.624,
+      "step": 8790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015694638226609726,
+      "loss": 0.6285,
+      "step": 8791
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001569373384957013,
+      "loss": 0.8974,
+      "step": 8792
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015692829403617383,
+      "loss": 0.6486,
+      "step": 8793
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015691924888762415,
+      "loss": 0.6273,
+      "step": 8794
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015691020305016188,
+      "loss": 0.8409,
+      "step": 8795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001569011565238964,
+      "loss": 0.7926,
+      "step": 8796
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015689210930893728,
+      "loss": 0.7566,
+      "step": 8797
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015688306140539395,
+      "loss": 0.6909,
+      "step": 8798
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000156874012813376,
+      "loss": 0.6244,
+      "step": 8799
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001568649635329929,
+      "loss": 0.7263,
+      "step": 8800
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 0.722107470035553,
+      "eval_runtime": 848.9678,
+      "eval_samples_per_second": 28.34,
+      "eval_steps_per_second": 14.17,
+      "step": 8800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015685591356435417,
+      "loss": 0.6709,
+      "step": 8801
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015684686290756936,
+      "loss": 0.8612,
+      "step": 8802
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015683781156274802,
+      "loss": 0.6891,
+      "step": 8803
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001568287595299997,
+      "loss": 0.6835,
+      "step": 8804
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015681970680943397,
+      "loss": 0.7653,
+      "step": 8805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015681065340116034,
+      "loss": 0.8248,
+      "step": 8806
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015680159930528843,
+      "loss": 0.6619,
+      "step": 8807
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015679254452192787,
+      "loss": 0.6643,
+      "step": 8808
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015678348905118818,
+      "loss": 0.7861,
+      "step": 8809
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000156774432893179,
+      "loss": 0.6753,
+      "step": 8810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001567653760480099,
+      "loss": 0.6749,
+      "step": 8811
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015675631851579057,
+      "loss": 0.6952,
+      "step": 8812
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015674726029663055,
+      "loss": 0.7069,
+      "step": 8813
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001567382013906395,
+      "loss": 0.6318,
+      "step": 8814
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015672914179792712,
+      "loss": 0.8008,
+      "step": 8815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015672008151860302,
+      "loss": 0.7396,
+      "step": 8816
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015671102055277684,
+      "loss": 0.6994,
+      "step": 8817
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015670195890055827,
+      "loss": 0.6953,
+      "step": 8818
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000156692896562057,
+      "loss": 0.7441,
+      "step": 8819
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001566838335373827,
+      "loss": 0.8259,
+      "step": 8820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015667476982664504,
+      "loss": 0.76,
+      "step": 8821
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015666570542995377,
+      "loss": 0.7371,
+      "step": 8822
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001566566403474185,
+      "loss": 0.7432,
+      "step": 8823
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015664757457914912,
+      "loss": 0.741,
+      "step": 8824
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001566385081252552,
+      "loss": 0.652,
+      "step": 8825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015662944098584653,
+      "loss": 0.6325,
+      "step": 8826
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015662037316103286,
+      "loss": 0.714,
+      "step": 8827
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015661130465092392,
+      "loss": 0.742,
+      "step": 8828
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015660223545562954,
+      "loss": 0.8081,
+      "step": 8829
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015659316557525935,
+      "loss": 0.8427,
+      "step": 8830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015658409500992327,
+      "loss": 0.662,
+      "step": 8831
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000156575023759731,
+      "loss": 0.7317,
+      "step": 8832
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015656595182479235,
+      "loss": 0.7021,
+      "step": 8833
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015655687920521712,
+      "loss": 0.7451,
+      "step": 8834
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015654780590111512,
+      "loss": 0.7979,
+      "step": 8835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015653873191259616,
+      "loss": 0.8094,
+      "step": 8836
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001565296572397701,
+      "loss": 0.7354,
+      "step": 8837
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015652058188274674,
+      "loss": 0.6726,
+      "step": 8838
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015651150584163593,
+      "loss": 0.7024,
+      "step": 8839
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001565024291165475,
+      "loss": 0.7921,
+      "step": 8840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015649335170759136,
+      "loss": 0.8122,
+      "step": 8841
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001564842736148773,
+      "loss": 0.7096,
+      "step": 8842
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001564751948385153,
+      "loss": 0.7723,
+      "step": 8843
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015646611537861515,
+      "loss": 0.5967,
+      "step": 8844
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001564570352352868,
+      "loss": 0.6648,
+      "step": 8845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001564479544086401,
+      "loss": 0.7567,
+      "step": 8846
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015643887289878503,
+      "loss": 0.6694,
+      "step": 8847
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001564297907058314,
+      "loss": 0.7724,
+      "step": 8848
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015642070782988923,
+      "loss": 0.7445,
+      "step": 8849
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015641162427106842,
+      "loss": 0.7523,
+      "step": 8850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015640254002947894,
+      "loss": 0.7213,
+      "step": 8851
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015639345510523066,
+      "loss": 0.7702,
+      "step": 8852
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001563843694984336,
+      "loss": 0.6819,
+      "step": 8853
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015637528320919773,
+      "loss": 0.6908,
+      "step": 8854
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000156366196237633,
+      "loss": 0.644,
+      "step": 8855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001563571085838494,
+      "loss": 0.7233,
+      "step": 8856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015634802024795693,
+      "loss": 0.744,
+      "step": 8857
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001563389312300656,
+      "loss": 0.7028,
+      "step": 8858
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015632984153028537,
+      "loss": 0.7247,
+      "step": 8859
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015632075114872628,
+      "loss": 0.6816,
+      "step": 8860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015631166008549835,
+      "loss": 0.7528,
+      "step": 8861
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015630256834071163,
+      "loss": 0.6244,
+      "step": 8862
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015629347591447618,
+      "loss": 0.7217,
+      "step": 8863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000156284382806902,
+      "loss": 0.6918,
+      "step": 8864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015627528901809918,
+      "loss": 0.7772,
+      "step": 8865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015626619454817776,
+      "loss": 0.7003,
+      "step": 8866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015625709939724782,
+      "loss": 0.7836,
+      "step": 8867
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015624800356541947,
+      "loss": 0.7813,
+      "step": 8868
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015623890705280274,
+      "loss": 0.5958,
+      "step": 8869
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001562298098595078,
+      "loss": 0.7123,
+      "step": 8870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001562207119856447,
+      "loss": 0.5786,
+      "step": 8871
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015621161343132357,
+      "loss": 0.7379,
+      "step": 8872
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015620251419665458,
+      "loss": 0.8107,
+      "step": 8873
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001561934142817478,
+      "loss": 0.8206,
+      "step": 8874
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015618431368671334,
+      "loss": 0.7249,
+      "step": 8875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015617521241166146,
+      "loss": 0.7352,
+      "step": 8876
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015616611045670225,
+      "loss": 0.785,
+      "step": 8877
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015615700782194584,
+      "loss": 0.7012,
+      "step": 8878
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015614790450750246,
+      "loss": 0.7458,
+      "step": 8879
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001561388005134823,
+      "loss": 0.7046,
+      "step": 8880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015612969583999546,
+      "loss": 0.7374,
+      "step": 8881
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001561205904871522,
+      "loss": 0.6497,
+      "step": 8882
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015611148445506275,
+      "loss": 0.7389,
+      "step": 8883
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001561023777438373,
+      "loss": 0.7751,
+      "step": 8884
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015609327035358603,
+      "loss": 0.6907,
+      "step": 8885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001560841622844192,
+      "loss": 0.7007,
+      "step": 8886
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015607505353644707,
+      "loss": 0.6072,
+      "step": 8887
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015606594410977987,
+      "loss": 0.7819,
+      "step": 8888
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015605683400452786,
+      "loss": 0.6592,
+      "step": 8889
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001560477232208013,
+      "loss": 0.7337,
+      "step": 8890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015603861175871045,
+      "loss": 0.721,
+      "step": 8891
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001560294996183656,
+      "loss": 0.6583,
+      "step": 8892
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015602038679987703,
+      "loss": 0.777,
+      "step": 8893
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015601127330335503,
+      "loss": 0.6549,
+      "step": 8894
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001560021591289099,
+      "loss": 0.7648,
+      "step": 8895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000155993044276652,
+      "loss": 0.7235,
+      "step": 8896
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001559839287466916,
+      "loss": 0.6852,
+      "step": 8897
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015597481253913905,
+      "loss": 0.6817,
+      "step": 8898
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015596569565410465,
+      "loss": 0.8149,
+      "step": 8899
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001559565780916988,
+      "loss": 0.7283,
+      "step": 8900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001559474598520318,
+      "loss": 0.6538,
+      "step": 8901
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015593834093521406,
+      "loss": 0.7329,
+      "step": 8902
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015592922134135593,
+      "loss": 0.7162,
+      "step": 8903
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015592010107056775,
+      "loss": 0.8021,
+      "step": 8904
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015591098012296,
+      "loss": 0.6264,
+      "step": 8905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015590185849864295,
+      "loss": 0.798,
+      "step": 8906
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001558927361977271,
+      "loss": 0.7329,
+      "step": 8907
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015588361322032283,
+      "loss": 0.7082,
+      "step": 8908
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015587448956654056,
+      "loss": 0.7878,
+      "step": 8909
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001558653652364907,
+      "loss": 0.7306,
+      "step": 8910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015585624023028372,
+      "loss": 0.8273,
+      "step": 8911
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015584711454802997,
+      "loss": 0.8429,
+      "step": 8912
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015583798818984002,
+      "loss": 0.8286,
+      "step": 8913
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001558288611558243,
+      "loss": 0.7223,
+      "step": 8914
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015581973344609321,
+      "loss": 0.741,
+      "step": 8915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015581060506075732,
+      "loss": 0.7364,
+      "step": 8916
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015580147599992704,
+      "loss": 0.7776,
+      "step": 8917
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015579234626371288,
+      "loss": 0.7914,
+      "step": 8918
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015578321585222536,
+      "loss": 0.8127,
+      "step": 8919
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000155774084765575,
+      "loss": 0.6474,
+      "step": 8920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015576495300387226,
+      "loss": 0.6173,
+      "step": 8921
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001557558205672277,
+      "loss": 0.7989,
+      "step": 8922
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015574668745575187,
+      "loss": 0.7498,
+      "step": 8923
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001557375536695553,
+      "loss": 0.6437,
+      "step": 8924
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015572841920874852,
+      "loss": 0.6929,
+      "step": 8925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001557192840734421,
+      "loss": 0.7017,
+      "step": 8926
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001557101482637466,
+      "loss": 0.7147,
+      "step": 8927
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015570101177977258,
+      "loss": 0.8348,
+      "step": 8928
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001556918746216307,
+      "loss": 0.6804,
+      "step": 8929
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015568273678943145,
+      "loss": 0.8366,
+      "step": 8930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015567359828328546,
+      "loss": 0.8034,
+      "step": 8931
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015566445910330335,
+      "loss": 0.6221,
+      "step": 8932
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015565531924959577,
+      "loss": 0.7719,
+      "step": 8933
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015564617872227326,
+      "loss": 0.7081,
+      "step": 8934
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015563703752144648,
+      "loss": 0.7299,
+      "step": 8935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015562789564722612,
+      "loss": 0.7861,
+      "step": 8936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015561875309972277,
+      "loss": 0.7452,
+      "step": 8937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001556096098790471,
+      "loss": 0.6125,
+      "step": 8938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015560046598530982,
+      "loss": 0.5617,
+      "step": 8939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015559132141862154,
+      "loss": 0.7135,
+      "step": 8940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015558217617909294,
+      "loss": 0.8693,
+      "step": 8941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015557303026683474,
+      "loss": 0.6599,
+      "step": 8942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015556388368195763,
+      "loss": 0.7958,
+      "step": 8943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015555473642457233,
+      "loss": 0.784,
+      "step": 8944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015554558849478948,
+      "loss": 0.6888,
+      "step": 8945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001555364398927199,
+      "loss": 0.7931,
+      "step": 8946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015552729061847425,
+      "loss": 0.7754,
+      "step": 8947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015551814067216326,
+      "loss": 0.7894,
+      "step": 8948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015550899005389773,
+      "loss": 0.746,
+      "step": 8949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015549983876378835,
+      "loss": 0.6662,
+      "step": 8950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015549068680194595,
+      "loss": 0.9028,
+      "step": 8951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015548153416848127,
+      "loss": 0.7562,
+      "step": 8952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015547238086350507,
+      "loss": 0.7369,
+      "step": 8953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015546322688712812,
+      "loss": 0.7626,
+      "step": 8954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015545407223946126,
+      "loss": 0.7071,
+      "step": 8955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015544491692061528,
+      "loss": 0.6746,
+      "step": 8956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015543576093070098,
+      "loss": 0.753,
+      "step": 8957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015542660426982917,
+      "loss": 0.7739,
+      "step": 8958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001554174469381107,
+      "loss": 0.6914,
+      "step": 8959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015540828893565638,
+      "loss": 0.8003,
+      "step": 8960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015539913026257706,
+      "loss": 0.712,
+      "step": 8961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001553899709189836,
+      "loss": 0.7717,
+      "step": 8962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001553808109049869,
+      "loss": 0.7633,
+      "step": 8963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015537165022069772,
+      "loss": 0.6842,
+      "step": 8964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015536248886622697,
+      "loss": 0.598,
+      "step": 8965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001553533268416856,
+      "loss": 0.6677,
+      "step": 8966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015534416414718447,
+      "loss": 0.7589,
+      "step": 8967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015533500078283443,
+      "loss": 0.7391,
+      "step": 8968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015532583674874646,
+      "loss": 0.7087,
+      "step": 8969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015531667204503143,
+      "loss": 0.6828,
+      "step": 8970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015530750667180025,
+      "loss": 0.8309,
+      "step": 8971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015529834062916388,
+      "loss": 0.7838,
+      "step": 8972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015528917391723329,
+      "loss": 0.7926,
+      "step": 8973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015528000653611935,
+      "loss": 0.7739,
+      "step": 8974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015527083848593308,
+      "loss": 0.7535,
+      "step": 8975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015526166976678543,
+      "loss": 0.6827,
+      "step": 8976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015525250037878732,
+      "loss": 0.5951,
+      "step": 8977
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001552433303220498,
+      "loss": 0.7343,
+      "step": 8978
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015523415959668385,
+      "loss": 0.7241,
+      "step": 8979
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015522498820280044,
+      "loss": 0.6682,
+      "step": 8980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015521581614051058,
+      "loss": 0.829,
+      "step": 8981
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015520664340992526,
+      "loss": 0.6906,
+      "step": 8982
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015519747001115555,
+      "loss": 0.662,
+      "step": 8983
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015518829594431246,
+      "loss": 0.6715,
+      "step": 8984
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000155179121209507,
+      "loss": 0.7023,
+      "step": 8985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015516994580685024,
+      "loss": 0.6945,
+      "step": 8986
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015516076973645326,
+      "loss": 0.7048,
+      "step": 8987
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015515159299842707,
+      "loss": 0.6267,
+      "step": 8988
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015514241559288277,
+      "loss": 0.6564,
+      "step": 8989
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001551332375199314,
+      "loss": 0.7406,
+      "step": 8990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001551240587796841,
+      "loss": 0.8389,
+      "step": 8991
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015511487937225194,
+      "loss": 0.713,
+      "step": 8992
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015510569929774598,
+      "loss": 0.7349,
+      "step": 8993
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001550965185562774,
+      "loss": 0.6679,
+      "step": 8994
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015508733714795727,
+      "loss": 0.7088,
+      "step": 8995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015507815507289672,
+      "loss": 0.7345,
+      "step": 8996
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001550689723312069,
+      "loss": 0.7044,
+      "step": 8997
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015505978892299896,
+      "loss": 0.6599,
+      "step": 8998
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015505060484838403,
+      "loss": 0.6853,
+      "step": 8999
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015504142010747327,
+      "loss": 0.6268,
+      "step": 9000
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 0.7145230770111084,
+      "eval_runtime": 849.5107,
+      "eval_samples_per_second": 28.322,
+      "eval_steps_per_second": 14.161,
+      "step": 9000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00015503223470037785,
+      "loss": 0.8292,
+      "step": 9001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015502304862720893,
+      "loss": 0.7209,
+      "step": 9002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015501386188807771,
+      "loss": 0.6438,
+      "step": 9003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001550046744830954,
+      "loss": 0.757,
+      "step": 9004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015499548641237314,
+      "loss": 0.6901,
+      "step": 9005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015498629767602217,
+      "loss": 0.7182,
+      "step": 9006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015497710827415374,
+      "loss": 0.6948,
+      "step": 9007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015496791820687902,
+      "loss": 0.8279,
+      "step": 9008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015495872747430924,
+      "loss": 0.7591,
+      "step": 9009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015494953607655563,
+      "loss": 0.7123,
+      "step": 9010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015494034401372953,
+      "loss": 0.662,
+      "step": 9011
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001549311512859421,
+      "loss": 0.6088,
+      "step": 9012
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015492195789330464,
+      "loss": 0.7721,
+      "step": 9013
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001549127638359284,
+      "loss": 0.696,
+      "step": 9014
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015490356911392465,
+      "loss": 0.7054,
+      "step": 9015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015489437372740472,
+      "loss": 0.614,
+      "step": 9016
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015488517767647987,
+      "loss": 0.6229,
+      "step": 9017
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001548759809612614,
+      "loss": 0.8421,
+      "step": 9018
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015486678358186064,
+      "loss": 0.8047,
+      "step": 9019
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001548575855383889,
+      "loss": 0.8101,
+      "step": 9020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015484838683095753,
+      "loss": 0.7767,
+      "step": 9021
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015483918745967784,
+      "loss": 0.8398,
+      "step": 9022
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015482998742466118,
+      "loss": 0.7579,
+      "step": 9023
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015482078672601887,
+      "loss": 0.729,
+      "step": 9024
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015481158536386231,
+      "loss": 0.696,
+      "step": 9025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015480238333830289,
+      "loss": 0.7107,
+      "step": 9026
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001547931806494519,
+      "loss": 0.6792,
+      "step": 9027
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001547839772974208,
+      "loss": 0.8115,
+      "step": 9028
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015477477328232095,
+      "loss": 0.6734,
+      "step": 9029
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015476556860426376,
+      "loss": 0.6762,
+      "step": 9030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015475636326336064,
+      "loss": 0.7061,
+      "step": 9031
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015474715725972299,
+      "loss": 0.6201,
+      "step": 9032
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015473795059346224,
+      "loss": 0.7233,
+      "step": 9033
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001547287432646898,
+      "loss": 0.7177,
+      "step": 9034
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015471953527351718,
+      "loss": 0.7467,
+      "step": 9035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015471032662005572,
+      "loss": 0.676,
+      "step": 9036
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015470111730441698,
+      "loss": 0.6649,
+      "step": 9037
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015469190732671234,
+      "loss": 0.5515,
+      "step": 9038
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015468269668705332,
+      "loss": 0.7245,
+      "step": 9039
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015467348538555141,
+      "loss": 0.8319,
+      "step": 9040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015466427342231807,
+      "loss": 0.7258,
+      "step": 9041
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001546550607974648,
+      "loss": 0.7322,
+      "step": 9042
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001546458475111031,
+      "loss": 0.7955,
+      "step": 9043
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015463663356334446,
+      "loss": 0.5944,
+      "step": 9044
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015462741895430043,
+      "loss": 0.6568,
+      "step": 9045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015461820368408257,
+      "loss": 0.77,
+      "step": 9046
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015460898775280236,
+      "loss": 0.7252,
+      "step": 9047
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015459977116057136,
+      "loss": 0.7176,
+      "step": 9048
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015459055390750113,
+      "loss": 0.7064,
+      "step": 9049
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015458133599370323,
+      "loss": 0.7441,
+      "step": 9050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015457211741928917,
+      "loss": 0.6214,
+      "step": 9051
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015456289818437062,
+      "loss": 0.6873,
+      "step": 9052
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015455367828905912,
+      "loss": 0.7294,
+      "step": 9053
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015454445773346626,
+      "loss": 0.7962,
+      "step": 9054
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015453523651770364,
+      "loss": 0.7274,
+      "step": 9055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015452601464188286,
+      "loss": 0.6784,
+      "step": 9056
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015451679210611553,
+      "loss": 0.7141,
+      "step": 9057
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015450756891051333,
+      "loss": 0.7698,
+      "step": 9058
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015449834505518778,
+      "loss": 0.6825,
+      "step": 9059
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015448912054025064,
+      "loss": 0.7074,
+      "step": 9060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015447989536581348,
+      "loss": 0.713,
+      "step": 9061
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015447066953198801,
+      "loss": 0.6993,
+      "step": 9062
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015446144303888582,
+      "loss": 0.6656,
+      "step": 9063
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015445221588661866,
+      "loss": 0.644,
+      "step": 9064
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015444298807529817,
+      "loss": 0.7694,
+      "step": 9065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015443375960503603,
+      "loss": 0.7812,
+      "step": 9066
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015442453047594397,
+      "loss": 0.7024,
+      "step": 9067
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015441530068813363,
+      "loss": 0.7253,
+      "step": 9068
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001544060702417168,
+      "loss": 0.804,
+      "step": 9069
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015439683913680513,
+      "loss": 0.7406,
+      "step": 9070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001543876073735104,
+      "loss": 0.6582,
+      "step": 9071
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001543783749519443,
+      "loss": 0.6994,
+      "step": 9072
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001543691418722186,
+      "loss": 0.7053,
+      "step": 9073
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015435990813444505,
+      "loss": 0.7162,
+      "step": 9074
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001543506737387354,
+      "loss": 0.7648,
+      "step": 9075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015434143868520142,
+      "loss": 0.685,
+      "step": 9076
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015433220297395492,
+      "loss": 0.8008,
+      "step": 9077
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015432296660510765,
+      "loss": 0.7336,
+      "step": 9078
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015431372957877136,
+      "loss": 0.7567,
+      "step": 9079
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015430449189505793,
+      "loss": 0.7741,
+      "step": 9080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001542952535540791,
+      "loss": 0.7692,
+      "step": 9081
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015428601455594673,
+      "loss": 0.7207,
+      "step": 9082
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001542767749007726,
+      "loss": 0.752,
+      "step": 9083
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015426753458866859,
+      "loss": 0.6834,
+      "step": 9084
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001542582936197465,
+      "loss": 0.7187,
+      "step": 9085
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015424905199411818,
+      "loss": 0.6604,
+      "step": 9086
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001542398097118955,
+      "loss": 0.601,
+      "step": 9087
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015423056677319036,
+      "loss": 0.7312,
+      "step": 9088
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015422132317811452,
+      "loss": 0.6906,
+      "step": 9089
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015421207892677997,
+      "loss": 0.6202,
+      "step": 9090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015420283401929851,
+      "loss": 0.7628,
+      "step": 9091
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001541935884557821,
+      "loss": 0.7375,
+      "step": 9092
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015418434223634264,
+      "loss": 0.7393,
+      "step": 9093
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000154175095361092,
+      "loss": 0.6159,
+      "step": 9094
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001541658478301421,
+      "loss": 0.6172,
+      "step": 9095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015415659964360492,
+      "loss": 0.6735,
+      "step": 9096
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001541473508015923,
+      "loss": 0.8412,
+      "step": 9097
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001541381013042163,
+      "loss": 0.6338,
+      "step": 9098
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015412885115158878,
+      "loss": 0.8549,
+      "step": 9099
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015411960034382173,
+      "loss": 0.745,
+      "step": 9100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001541103488810271,
+      "loss": 0.6362,
+      "step": 9101
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001541010967633169,
+      "loss": 0.6871,
+      "step": 9102
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000154091843990803,
+      "loss": 0.6638,
+      "step": 9103
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015408259056359758,
+      "loss": 0.6311,
+      "step": 9104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015407333648181248,
+      "loss": 0.7831,
+      "step": 9105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015406408174555976,
+      "loss": 0.7465,
+      "step": 9106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015405482635495147,
+      "loss": 0.7253,
+      "step": 9107
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015404557031009955,
+      "loss": 0.6126,
+      "step": 9108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015403631361111608,
+      "loss": 0.7298,
+      "step": 9109
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001540270562581131,
+      "loss": 0.6634,
+      "step": 9110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015401779825120263,
+      "loss": 0.7371,
+      "step": 9111
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015400853959049673,
+      "loss": 0.6583,
+      "step": 9112
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015399928027610749,
+      "loss": 0.7411,
+      "step": 9113
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015399002030814694,
+      "loss": 0.7407,
+      "step": 9114
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015398075968672717,
+      "loss": 0.6821,
+      "step": 9115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015397149841196028,
+      "loss": 0.7517,
+      "step": 9116
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001539622364839583,
+      "loss": 0.8846,
+      "step": 9117
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015395297390283347,
+      "loss": 0.7603,
+      "step": 9118
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015394371066869772,
+      "loss": 0.5982,
+      "step": 9119
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015393444678166328,
+      "loss": 0.6249,
+      "step": 9120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015392518224184224,
+      "loss": 0.723,
+      "step": 9121
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015391591704934677,
+      "loss": 0.8711,
+      "step": 9122
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015390665120428897,
+      "loss": 0.7128,
+      "step": 9123
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015389738470678098,
+      "loss": 0.7093,
+      "step": 9124
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015388811755693492,
+      "loss": 0.6499,
+      "step": 9125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015387884975486305,
+      "loss": 0.6846,
+      "step": 9126
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015386958130067752,
+      "loss": 0.7958,
+      "step": 9127
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015386031219449047,
+      "loss": 0.7548,
+      "step": 9128
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001538510424364141,
+      "loss": 0.6887,
+      "step": 9129
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001538417720265606,
+      "loss": 0.6456,
+      "step": 9130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015383250096504217,
+      "loss": 0.7561,
+      "step": 9131
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015382322925197107,
+      "loss": 0.7702,
+      "step": 9132
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015381395688745942,
+      "loss": 0.6871,
+      "step": 9133
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015380468387161958,
+      "loss": 0.7952,
+      "step": 9134
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015379541020456365,
+      "loss": 0.7002,
+      "step": 9135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015378613588640393,
+      "loss": 0.7969,
+      "step": 9136
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015377686091725272,
+      "loss": 0.8003,
+      "step": 9137
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001537675852972222,
+      "loss": 0.7405,
+      "step": 9138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015375830902642466,
+      "loss": 0.7272,
+      "step": 9139
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001537490321049724,
+      "loss": 0.6747,
+      "step": 9140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015373975453297767,
+      "loss": 0.7959,
+      "step": 9141
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015373047631055277,
+      "loss": 0.6741,
+      "step": 9142
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015372119743781,
+      "loss": 0.6961,
+      "step": 9143
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015371191791486167,
+      "loss": 0.8316,
+      "step": 9144
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015370263774182007,
+      "loss": 0.5661,
+      "step": 9145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015369335691879757,
+      "loss": 0.6986,
+      "step": 9146
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015368407544590643,
+      "loss": 0.5111,
+      "step": 9147
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015367479332325907,
+      "loss": 0.6889,
+      "step": 9148
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015366551055096775,
+      "loss": 0.7696,
+      "step": 9149
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001536562271291449,
+      "loss": 0.5661,
+      "step": 9150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001536469430579028,
+      "loss": 0.8709,
+      "step": 9151
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015363765833735385,
+      "loss": 0.7063,
+      "step": 9152
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001536283729676105,
+      "loss": 0.7088,
+      "step": 9153
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015361908694878503,
+      "loss": 0.71,
+      "step": 9154
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015360980028098983,
+      "loss": 0.7439,
+      "step": 9155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001536005129643374,
+      "loss": 0.6529,
+      "step": 9156
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015359122499894007,
+      "loss": 0.8129,
+      "step": 9157
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015358193638491025,
+      "loss": 0.7554,
+      "step": 9158
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015357264712236043,
+      "loss": 0.7298,
+      "step": 9159
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000153563357211403,
+      "loss": 0.6613,
+      "step": 9160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015355406665215035,
+      "loss": 0.6647,
+      "step": 9161
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000153544775444715,
+      "loss": 0.6462,
+      "step": 9162
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015353548358920937,
+      "loss": 0.6374,
+      "step": 9163
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015352619108574596,
+      "loss": 0.7056,
+      "step": 9164
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015351689793443714,
+      "loss": 0.7926,
+      "step": 9165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015350760413539557,
+      "loss": 0.7858,
+      "step": 9166
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015349830968873355,
+      "loss": 0.7088,
+      "step": 9167
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015348901459456363,
+      "loss": 0.776,
+      "step": 9168
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015347971885299838,
+      "loss": 0.7952,
+      "step": 9169
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015347042246415023,
+      "loss": 0.6776,
+      "step": 9170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015346112542813175,
+      "loss": 0.5893,
+      "step": 9171
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015345182774505545,
+      "loss": 0.6098,
+      "step": 9172
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001534425294150338,
+      "loss": 0.7544,
+      "step": 9173
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015343323043817946,
+      "loss": 0.7099,
+      "step": 9174
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001534239308146049,
+      "loss": 0.7503,
+      "step": 9175
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015341463054442266,
+      "loss": 0.7796,
+      "step": 9176
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015340532962774533,
+      "loss": 0.6579,
+      "step": 9177
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001533960280646855,
+      "loss": 0.6841,
+      "step": 9178
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015338672585535579,
+      "loss": 0.7314,
+      "step": 9179
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001533774229998687,
+      "loss": 0.7088,
+      "step": 9180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015336811949833682,
+      "loss": 0.7392,
+      "step": 9181
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015335881535087284,
+      "loss": 0.8505,
+      "step": 9182
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015334951055758931,
+      "loss": 0.6626,
+      "step": 9183
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001533402051185989,
+      "loss": 0.7057,
+      "step": 9184
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015333089903401417,
+      "loss": 0.8154,
+      "step": 9185
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015332159230394777,
+      "loss": 0.7169,
+      "step": 9186
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015331228492851238,
+      "loss": 0.717,
+      "step": 9187
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015330297690782063,
+      "loss": 0.6861,
+      "step": 9188
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015329366824198517,
+      "loss": 0.6639,
+      "step": 9189
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015328435893111866,
+      "loss": 0.6276,
+      "step": 9190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001532750489753338,
+      "loss": 0.6262,
+      "step": 9191
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015326573837474324,
+      "loss": 0.7326,
+      "step": 9192
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015325642712945973,
+      "loss": 0.7985,
+      "step": 9193
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015324711523959588,
+      "loss": 0.7127,
+      "step": 9194
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015323780270526448,
+      "loss": 0.7086,
+      "step": 9195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015322848952657815,
+      "loss": 0.7555,
+      "step": 9196
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001532191757036497,
+      "loss": 0.7696,
+      "step": 9197
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001532098612365918,
+      "loss": 0.7118,
+      "step": 9198
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015320054612551722,
+      "loss": 0.6565,
+      "step": 9199
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015319123037053868,
+      "loss": 0.6954,
+      "step": 9200
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 0.7067068219184875,
+      "eval_runtime": 852.677,
+      "eval_samples_per_second": 28.217,
+      "eval_steps_per_second": 14.109,
+      "step": 9200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015318191397176896,
+      "loss": 0.6549,
+      "step": 9201
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015317259692932078,
+      "loss": 0.7741,
+      "step": 9202
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015316327924330697,
+      "loss": 0.728,
+      "step": 9203
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001531539609138402,
+      "loss": 0.7331,
+      "step": 9204
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015314464194103337,
+      "loss": 0.6615,
+      "step": 9205
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001531353223249992,
+      "loss": 0.6314,
+      "step": 9206
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001531260020658505,
+      "loss": 0.7431,
+      "step": 9207
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001531166811637001,
+      "loss": 0.8746,
+      "step": 9208
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015310735961866076,
+      "loss": 0.6087,
+      "step": 9209
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015309803743084537,
+      "loss": 0.6956,
+      "step": 9210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015308871460036674,
+      "loss": 0.644,
+      "step": 9211
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015307939112733763,
+      "loss": 0.7669,
+      "step": 9212
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015307006701187104,
+      "loss": 0.791,
+      "step": 9213
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015306074225407968,
+      "loss": 0.7628,
+      "step": 9214
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015305141685407648,
+      "loss": 0.7107,
+      "step": 9215
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015304209081197425,
+      "loss": 0.7239,
+      "step": 9216
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015303276412788596,
+      "loss": 0.7067,
+      "step": 9217
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015302343680192443,
+      "loss": 0.7493,
+      "step": 9218
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015301410883420254,
+      "loss": 0.6693,
+      "step": 9219
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015300478022483324,
+      "loss": 0.6725,
+      "step": 9220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015299545097392938,
+      "loss": 0.5833,
+      "step": 9221
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015298612108160392,
+      "loss": 0.7508,
+      "step": 9222
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015297679054796979,
+      "loss": 0.7768,
+      "step": 9223
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015296745937313987,
+      "loss": 0.7177,
+      "step": 9224
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015295812755722717,
+      "loss": 0.701,
+      "step": 9225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015294879510034453,
+      "loss": 0.6267,
+      "step": 9226
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000152939462002605,
+      "loss": 0.7664,
+      "step": 9227
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001529301282641215,
+      "loss": 0.7454,
+      "step": 9228
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015292079388500702,
+      "loss": 0.616,
+      "step": 9229
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015291145886537453,
+      "loss": 0.639,
+      "step": 9230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000152902123205337,
+      "loss": 0.6372,
+      "step": 9231
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015289278690500742,
+      "loss": 0.6713,
+      "step": 9232
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015288344996449883,
+      "loss": 0.6996,
+      "step": 9233
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015287411238392416,
+      "loss": 0.7682,
+      "step": 9234
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015286477416339656,
+      "loss": 0.71,
+      "step": 9235
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015285543530302892,
+      "loss": 0.6591,
+      "step": 9236
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015284609580293432,
+      "loss": 0.8557,
+      "step": 9237
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015283675566322578,
+      "loss": 0.7349,
+      "step": 9238
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015282741488401643,
+      "loss": 0.767,
+      "step": 9239
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001528180734654192,
+      "loss": 0.635,
+      "step": 9240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015280873140754726,
+      "loss": 0.7373,
+      "step": 9241
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001527993887105136,
+      "loss": 0.6911,
+      "step": 9242
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015279004537443135,
+      "loss": 0.7488,
+      "step": 9243
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001527807013994136,
+      "loss": 0.6956,
+      "step": 9244
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001527713567855734,
+      "loss": 0.6914,
+      "step": 9245
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015276201153302388,
+      "loss": 0.7163,
+      "step": 9246
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001527526656418781,
+      "loss": 0.6848,
+      "step": 9247
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001527433191122493,
+      "loss": 0.8065,
+      "step": 9248
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015273397194425045,
+      "loss": 0.7784,
+      "step": 9249
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015272462413799476,
+      "loss": 0.78,
+      "step": 9250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015271527569359538,
+      "loss": 0.7455,
+      "step": 9251
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015270592661116542,
+      "loss": 0.628,
+      "step": 9252
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001526965768908181,
+      "loss": 0.7469,
+      "step": 9253
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015268722653266652,
+      "loss": 0.6832,
+      "step": 9254
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015267787553682386,
+      "loss": 0.6907,
+      "step": 9255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015266852390340328,
+      "loss": 0.6695,
+      "step": 9256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015265917163251804,
+      "loss": 0.6858,
+      "step": 9257
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015264981872428125,
+      "loss": 0.6945,
+      "step": 9258
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015264046517880618,
+      "loss": 0.6959,
+      "step": 9259
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015263111099620598,
+      "loss": 0.6479,
+      "step": 9260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015262175617659392,
+      "loss": 0.6179,
+      "step": 9261
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015261240072008317,
+      "loss": 0.8853,
+      "step": 9262
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000152603044626787,
+      "loss": 0.7617,
+      "step": 9263
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015259368789681866,
+      "loss": 0.6392,
+      "step": 9264
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015258433053029136,
+      "loss": 0.7833,
+      "step": 9265
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015257497252731835,
+      "loss": 0.6511,
+      "step": 9266
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015256561388801297,
+      "loss": 0.7622,
+      "step": 9267
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015255625461248838,
+      "loss": 0.6262,
+      "step": 9268
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015254689470085796,
+      "loss": 0.6885,
+      "step": 9269
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015253753415323494,
+      "loss": 0.7461,
+      "step": 9270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015252817296973264,
+      "loss": 0.7288,
+      "step": 9271
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015251881115046435,
+      "loss": 0.7621,
+      "step": 9272
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001525094486955433,
+      "loss": 0.6041,
+      "step": 9273
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015250008560508297,
+      "loss": 0.7403,
+      "step": 9274
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015249072187919658,
+      "loss": 0.6981,
+      "step": 9275
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015248135751799743,
+      "loss": 0.7123,
+      "step": 9276
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015247199252159895,
+      "loss": 0.7272,
+      "step": 9277
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015246262689011446,
+      "loss": 0.6098,
+      "step": 9278
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015245326062365727,
+      "loss": 0.8289,
+      "step": 9279
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015244389372234083,
+      "loss": 0.6811,
+      "step": 9280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001524345261862784,
+      "loss": 0.7102,
+      "step": 9281
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015242515801558345,
+      "loss": 0.6583,
+      "step": 9282
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001524157892103693,
+      "loss": 0.6384,
+      "step": 9283
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015240641977074941,
+      "loss": 0.6967,
+      "step": 9284
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001523970496968371,
+      "loss": 0.7472,
+      "step": 9285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015238767898874587,
+      "loss": 0.7521,
+      "step": 9286
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015237830764658908,
+      "loss": 0.7056,
+      "step": 9287
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015236893567048017,
+      "loss": 0.7461,
+      "step": 9288
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015235956306053258,
+      "loss": 0.6567,
+      "step": 9289
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001523501898168597,
+      "loss": 0.7467,
+      "step": 9290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015234081593957504,
+      "loss": 0.5716,
+      "step": 9291
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015233144142879203,
+      "loss": 0.772,
+      "step": 9292
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015232206628462412,
+      "loss": 0.8196,
+      "step": 9293
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001523126905071848,
+      "loss": 0.6728,
+      "step": 9294
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001523033140965876,
+      "loss": 0.6891,
+      "step": 9295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015229393705294588,
+      "loss": 0.7439,
+      "step": 9296
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015228455937637322,
+      "loss": 0.585,
+      "step": 9297
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015227518106698308,
+      "loss": 0.6816,
+      "step": 9298
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015226580212488906,
+      "loss": 0.6726,
+      "step": 9299
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015225642255020457,
+      "loss": 0.7343,
+      "step": 9300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015224704234304316,
+      "loss": 0.7274,
+      "step": 9301
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015223766150351838,
+      "loss": 0.7813,
+      "step": 9302
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015222828003174377,
+      "loss": 0.6219,
+      "step": 9303
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001522188979278329,
+      "loss": 0.6328,
+      "step": 9304
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015220951519189925,
+      "loss": 0.7861,
+      "step": 9305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015220013182405643,
+      "loss": 0.7414,
+      "step": 9306
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015219074782441805,
+      "loss": 0.6475,
+      "step": 9307
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015218136319309762,
+      "loss": 0.7463,
+      "step": 9308
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015217197793020875,
+      "loss": 0.6351,
+      "step": 9309
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015216259203586506,
+      "loss": 0.629,
+      "step": 9310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015215320551018006,
+      "loss": 0.717,
+      "step": 9311
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015214381835326749,
+      "loss": 0.6936,
+      "step": 9312
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015213443056524088,
+      "loss": 0.6593,
+      "step": 9313
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015212504214621384,
+      "loss": 0.6767,
+      "step": 9314
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015211565309630008,
+      "loss": 0.6734,
+      "step": 9315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015210626341561313,
+      "loss": 0.6895,
+      "step": 9316
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015209687310426676,
+      "loss": 0.7307,
+      "step": 9317
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015208748216237456,
+      "loss": 0.7008,
+      "step": 9318
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015207809059005015,
+      "loss": 0.7937,
+      "step": 9319
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015206869838740728,
+      "loss": 0.8755,
+      "step": 9320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015205930555455962,
+      "loss": 0.6807,
+      "step": 9321
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015204991209162078,
+      "loss": 0.6226,
+      "step": 9322
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015204051799870452,
+      "loss": 0.6736,
+      "step": 9323
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015203112327592447,
+      "loss": 0.823,
+      "step": 9324
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015202172792339445,
+      "loss": 0.6786,
+      "step": 9325
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001520123319412281,
+      "loss": 0.662,
+      "step": 9326
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015200293532953912,
+      "loss": 0.7474,
+      "step": 9327
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015199353808844127,
+      "loss": 0.634,
+      "step": 9328
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015198414021804833,
+      "loss": 0.7878,
+      "step": 9329
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000151974741718474,
+      "loss": 0.8008,
+      "step": 9330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015196534258983202,
+      "loss": 0.752,
+      "step": 9331
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015195594283223614,
+      "loss": 0.7177,
+      "step": 9332
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015194654244580022,
+      "loss": 0.6813,
+      "step": 9333
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015193714143063795,
+      "loss": 0.8367,
+      "step": 9334
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015192773978686314,
+      "loss": 0.7497,
+      "step": 9335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001519183375145896,
+      "loss": 0.7496,
+      "step": 9336
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015190893461393108,
+      "loss": 0.7566,
+      "step": 9337
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001518995310850014,
+      "loss": 0.7091,
+      "step": 9338
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015189012692791445,
+      "loss": 0.7516,
+      "step": 9339
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015188072214278396,
+      "loss": 0.7174,
+      "step": 9340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015187131672972376,
+      "loss": 0.7704,
+      "step": 9341
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015186191068884775,
+      "loss": 0.8457,
+      "step": 9342
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001518525040202697,
+      "loss": 0.6627,
+      "step": 9343
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015184309672410358,
+      "loss": 0.6551,
+      "step": 9344
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015183368880046313,
+      "loss": 0.8055,
+      "step": 9345
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015182428024946226,
+      "loss": 0.6724,
+      "step": 9346
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015181487107121484,
+      "loss": 0.8443,
+      "step": 9347
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001518054612658348,
+      "loss": 0.591,
+      "step": 9348
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001517960508334359,
+      "loss": 0.6062,
+      "step": 9349
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015178663977413222,
+      "loss": 0.6569,
+      "step": 9350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015177722808803755,
+      "loss": 0.7511,
+      "step": 9351
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015176781577526578,
+      "loss": 0.6591,
+      "step": 9352
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001517584028359309,
+      "loss": 0.7059,
+      "step": 9353
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015174898927014678,
+      "loss": 0.6357,
+      "step": 9354
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015173957507802743,
+      "loss": 0.8043,
+      "step": 9355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015173016025968675,
+      "loss": 0.5872,
+      "step": 9356
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015172074481523864,
+      "loss": 0.6468,
+      "step": 9357
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015171132874479716,
+      "loss": 0.7305,
+      "step": 9358
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001517019120484762,
+      "loss": 0.7278,
+      "step": 9359
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015169249472638978,
+      "loss": 0.7224,
+      "step": 9360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015168307677865185,
+      "loss": 0.7082,
+      "step": 9361
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001516736582053764,
+      "loss": 0.6658,
+      "step": 9362
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015166423900667746,
+      "loss": 0.6578,
+      "step": 9363
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000151654819182669,
+      "loss": 0.711,
+      "step": 9364
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000151645398733465,
+      "loss": 0.7914,
+      "step": 9365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015163597765917953,
+      "loss": 0.6959,
+      "step": 9366
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015162655595992664,
+      "loss": 0.7577,
+      "step": 9367
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015161713363582031,
+      "loss": 0.8067,
+      "step": 9368
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001516077106869746,
+      "loss": 0.5696,
+      "step": 9369
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015159828711350354,
+      "loss": 0.6933,
+      "step": 9370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015158886291552124,
+      "loss": 0.5857,
+      "step": 9371
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015157943809314167,
+      "loss": 0.6356,
+      "step": 9372
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015157001264647904,
+      "loss": 0.6427,
+      "step": 9373
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001515605865756473,
+      "loss": 0.7232,
+      "step": 9374
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001515511598807606,
+      "loss": 0.714,
+      "step": 9375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015154173256193303,
+      "loss": 0.6777,
+      "step": 9376
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001515323046192787,
+      "loss": 0.7514,
+      "step": 9377
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015152287605291165,
+      "loss": 0.6347,
+      "step": 9378
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001515134468629461,
+      "loss": 0.8049,
+      "step": 9379
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001515040170494961,
+      "loss": 0.6218,
+      "step": 9380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015149458661267584,
+      "loss": 0.6851,
+      "step": 9381
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001514851555525994,
+      "loss": 0.6839,
+      "step": 9382
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015147572386938095,
+      "loss": 0.6831,
+      "step": 9383
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015146629156313464,
+      "loss": 0.6821,
+      "step": 9384
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015145685863397467,
+      "loss": 0.6529,
+      "step": 9385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015144742508201516,
+      "loss": 0.7348,
+      "step": 9386
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015143799090737035,
+      "loss": 0.6874,
+      "step": 9387
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015142855611015435,
+      "loss": 0.6805,
+      "step": 9388
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001514191206904814,
+      "loss": 0.7432,
+      "step": 9389
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015140968464846566,
+      "loss": 0.7654,
+      "step": 9390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015140024798422137,
+      "loss": 0.725,
+      "step": 9391
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015139081069786276,
+      "loss": 0.7744,
+      "step": 9392
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000151381372789504,
+      "loss": 0.6684,
+      "step": 9393
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001513719342592594,
+      "loss": 0.6806,
+      "step": 9394
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015136249510724315,
+      "loss": 0.8023,
+      "step": 9395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015135305533356944,
+      "loss": 0.6551,
+      "step": 9396
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015134361493835262,
+      "loss": 0.6643,
+      "step": 9397
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015133417392170688,
+      "loss": 0.8069,
+      "step": 9398
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015132473228374654,
+      "loss": 0.722,
+      "step": 9399
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015131529002458585,
+      "loss": 0.8179,
+      "step": 9400
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 0.6991519927978516,
+      "eval_runtime": 845.2941,
+      "eval_samples_per_second": 28.463,
+      "eval_steps_per_second": 14.232,
+      "step": 9400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015130584714433907,
+      "loss": 0.6521,
+      "step": 9401
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015129640364312054,
+      "loss": 0.8208,
+      "step": 9402
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015128695952104453,
+      "loss": 0.7404,
+      "step": 9403
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015127751477822532,
+      "loss": 0.6847,
+      "step": 9404
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015126806941477726,
+      "loss": 0.6558,
+      "step": 9405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015125862343081468,
+      "loss": 0.6373,
+      "step": 9406
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001512491768264519,
+      "loss": 0.6874,
+      "step": 9407
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015123972960180321,
+      "loss": 0.8289,
+      "step": 9408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000151230281756983,
+      "loss": 0.6555,
+      "step": 9409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001512208332921056,
+      "loss": 0.7324,
+      "step": 9410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015121138420728534,
+      "loss": 0.6276,
+      "step": 9411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001512019345026367,
+      "loss": 0.6985,
+      "step": 9412
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015119248417827394,
+      "loss": 0.6843,
+      "step": 9413
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015118303323431145,
+      "loss": 0.6603,
+      "step": 9414
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015117358167086365,
+      "loss": 0.6938,
+      "step": 9415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015116412948804498,
+      "loss": 0.6675,
+      "step": 9416
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015115467668596973,
+      "loss": 0.8538,
+      "step": 9417
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015114522326475238,
+      "loss": 0.6195,
+      "step": 9418
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015113576922450736,
+      "loss": 0.761,
+      "step": 9419
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015112631456534907,
+      "loss": 0.6601,
+      "step": 9420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015111685928739194,
+      "loss": 0.685,
+      "step": 9421
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015110740339075042,
+      "loss": 0.698,
+      "step": 9422
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015109794687553893,
+      "loss": 0.7292,
+      "step": 9423
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015108848974187196,
+      "loss": 0.8037,
+      "step": 9424
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000151079031989864,
+      "loss": 0.677,
+      "step": 9425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015106957361962943,
+      "loss": 0.672,
+      "step": 9426
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015106011463128283,
+      "loss": 0.8406,
+      "step": 9427
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001510506550249386,
+      "loss": 0.7644,
+      "step": 9428
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001510411948007113,
+      "loss": 0.6661,
+      "step": 9429
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015103173395871538,
+      "loss": 0.865,
+      "step": 9430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015102227249906536,
+      "loss": 0.7429,
+      "step": 9431
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015101281042187578,
+      "loss": 0.7711,
+      "step": 9432
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015100334772726113,
+      "loss": 0.7749,
+      "step": 9433
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015099388441533598,
+      "loss": 0.6977,
+      "step": 9434
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015098442048621483,
+      "loss": 0.7837,
+      "step": 9435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015097495594001222,
+      "loss": 0.652,
+      "step": 9436
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015096549077684275,
+      "loss": 0.7551,
+      "step": 9437
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015095602499682093,
+      "loss": 0.895,
+      "step": 9438
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015094655860006136,
+      "loss": 0.6528,
+      "step": 9439
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015093709158667858,
+      "loss": 0.7491,
+      "step": 9440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015092762395678723,
+      "loss": 0.7074,
+      "step": 9441
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015091815571050186,
+      "loss": 0.7068,
+      "step": 9442
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015090868684793706,
+      "loss": 0.7311,
+      "step": 9443
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015089921736920742,
+      "loss": 0.6803,
+      "step": 9444
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001508897472744276,
+      "loss": 0.6649,
+      "step": 9445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015088027656371224,
+      "loss": 0.7259,
+      "step": 9446
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001508708052371759,
+      "loss": 0.742,
+      "step": 9447
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015086133329493323,
+      "loss": 0.6682,
+      "step": 9448
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015085186073709886,
+      "loss": 0.7142,
+      "step": 9449
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015084238756378753,
+      "loss": 0.7112,
+      "step": 9450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015083291377511378,
+      "loss": 0.687,
+      "step": 9451
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015082343937119233,
+      "loss": 0.7431,
+      "step": 9452
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015081396435213785,
+      "loss": 0.7356,
+      "step": 9453
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015080448871806503,
+      "loss": 0.6792,
+      "step": 9454
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015079501246908854,
+      "loss": 0.6636,
+      "step": 9455
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015078553560532307,
+      "loss": 0.7461,
+      "step": 9456
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015077605812688328,
+      "loss": 0.6182,
+      "step": 9457
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000150766580033884,
+      "loss": 0.6585,
+      "step": 9458
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015075710132643987,
+      "loss": 0.6203,
+      "step": 9459
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015074762200466556,
+      "loss": 0.7049,
+      "step": 9460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015073814206867592,
+      "loss": 0.6717,
+      "step": 9461
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015072866151858558,
+      "loss": 0.7795,
+      "step": 9462
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015071918035450938,
+      "loss": 0.7139,
+      "step": 9463
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000150709698576562,
+      "loss": 0.6921,
+      "step": 9464
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001507002161848582,
+      "loss": 0.6382,
+      "step": 9465
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015069073317951282,
+      "loss": 0.6817,
+      "step": 9466
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015068124956064057,
+      "loss": 0.6339,
+      "step": 9467
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001506717653283563,
+      "loss": 0.7069,
+      "step": 9468
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015066228048277473,
+      "loss": 0.5637,
+      "step": 9469
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001506527950240107,
+      "loss": 0.7105,
+      "step": 9470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015064330895217895,
+      "loss": 0.7895,
+      "step": 9471
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015063382226739437,
+      "loss": 0.7745,
+      "step": 9472
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001506243349697718,
+      "loss": 0.7279,
+      "step": 9473
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000150614847059426,
+      "loss": 0.6901,
+      "step": 9474
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001506053585364718,
+      "loss": 0.6086,
+      "step": 9475
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001505958694010241,
+      "loss": 0.8373,
+      "step": 9476
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001505863796531977,
+      "loss": 0.8009,
+      "step": 9477
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015057688929310747,
+      "loss": 0.7599,
+      "step": 9478
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015056739832086834,
+      "loss": 0.8595,
+      "step": 9479
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015055790673659507,
+      "loss": 0.6999,
+      "step": 9480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015054841454040262,
+      "loss": 0.7547,
+      "step": 9481
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015053892173240584,
+      "loss": 0.661,
+      "step": 9482
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015052942831271965,
+      "loss": 0.6465,
+      "step": 9483
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015051993428145897,
+      "loss": 0.6327,
+      "step": 9484
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015051043963873863,
+      "loss": 0.587,
+      "step": 9485
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015050094438467362,
+      "loss": 0.7386,
+      "step": 9486
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015049144851937883,
+      "loss": 0.6448,
+      "step": 9487
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015048195204296922,
+      "loss": 0.6549,
+      "step": 9488
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001504724549555597,
+      "loss": 0.8107,
+      "step": 9489
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001504629572572652,
+      "loss": 0.8073,
+      "step": 9490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015045345894820072,
+      "loss": 0.702,
+      "step": 9491
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015044396002848121,
+      "loss": 0.7112,
+      "step": 9492
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015043446049822166,
+      "loss": 0.6814,
+      "step": 9493
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000150424960357537,
+      "loss": 0.647,
+      "step": 9494
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015041545960654219,
+      "loss": 0.8377,
+      "step": 9495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001504059582453523,
+      "loss": 0.6991,
+      "step": 9496
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015039645627408228,
+      "loss": 0.656,
+      "step": 9497
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015038695369284715,
+      "loss": 0.7381,
+      "step": 9498
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015037745050176189,
+      "loss": 0.6169,
+      "step": 9499
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015036794670094156,
+      "loss": 0.7215,
+      "step": 9500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001503584422905012,
+      "loss": 0.6402,
+      "step": 9501
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001503489372705558,
+      "loss": 0.739,
+      "step": 9502
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015033943164122045,
+      "loss": 0.8109,
+      "step": 9503
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015032992540261016,
+      "loss": 0.6438,
+      "step": 9504
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015032041855483997,
+      "loss": 0.6392,
+      "step": 9505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015031091109802503,
+      "loss": 0.68,
+      "step": 9506
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001503014030322803,
+      "loss": 0.8039,
+      "step": 9507
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015029189435772095,
+      "loss": 0.6365,
+      "step": 9508
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015028238507446202,
+      "loss": 0.6866,
+      "step": 9509
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015027287518261864,
+      "loss": 0.7995,
+      "step": 9510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015026336468230587,
+      "loss": 0.7263,
+      "step": 9511
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001502538535736388,
+      "loss": 0.7288,
+      "step": 9512
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015024434185673263,
+      "loss": 0.6358,
+      "step": 9513
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015023482953170243,
+      "loss": 0.687,
+      "step": 9514
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015022531659866333,
+      "loss": 0.8007,
+      "step": 9515
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001502158030577305,
+      "loss": 0.6711,
+      "step": 9516
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015020628890901905,
+      "loss": 0.699,
+      "step": 9517
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001501967741526441,
+      "loss": 0.7882,
+      "step": 9518
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001501872587887209,
+      "loss": 0.6352,
+      "step": 9519
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015017774281736457,
+      "loss": 0.669,
+      "step": 9520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015016822623869027,
+      "loss": 0.7706,
+      "step": 9521
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001501587090528132,
+      "loss": 0.8179,
+      "step": 9522
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015014919125984854,
+      "loss": 0.8662,
+      "step": 9523
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015013967285991154,
+      "loss": 0.6806,
+      "step": 9524
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015013015385311734,
+      "loss": 0.6001,
+      "step": 9525
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015012063423958113,
+      "loss": 0.8117,
+      "step": 9526
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015011111401941823,
+      "loss": 0.7016,
+      "step": 9527
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015010159319274376,
+      "loss": 0.6726,
+      "step": 9528
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015009207175967305,
+      "loss": 0.6994,
+      "step": 9529
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015008254972032124,
+      "loss": 0.7367,
+      "step": 9530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015007302707480366,
+      "loss": 0.7011,
+      "step": 9531
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015006350382323555,
+      "loss": 0.7264,
+      "step": 9532
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015005397996573213,
+      "loss": 0.5937,
+      "step": 9533
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015004445550240872,
+      "loss": 0.6174,
+      "step": 9534
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015003493043338058,
+      "loss": 0.8008,
+      "step": 9535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015002540475876294,
+      "loss": 0.6399,
+      "step": 9536
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001500158784786712,
+      "loss": 0.6387,
+      "step": 9537
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015000635159322063,
+      "loss": 0.5799,
+      "step": 9538
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014999682410252643,
+      "loss": 0.8557,
+      "step": 9539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014998729600670406,
+      "loss": 0.7376,
+      "step": 9540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014997776730586878,
+      "loss": 0.7861,
+      "step": 9541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014996823800013587,
+      "loss": 0.8592,
+      "step": 9542
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014995870808962077,
+      "loss": 0.7002,
+      "step": 9543
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001499491775744387,
+      "loss": 0.6797,
+      "step": 9544
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014993964645470514,
+      "loss": 0.7186,
+      "step": 9545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014993011473053537,
+      "loss": 0.7098,
+      "step": 9546
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014992058240204475,
+      "loss": 0.7406,
+      "step": 9547
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001499110494693487,
+      "loss": 0.7054,
+      "step": 9548
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014990151593256257,
+      "loss": 0.6657,
+      "step": 9549
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014989198179180178,
+      "loss": 0.6894,
+      "step": 9550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014988244704718163,
+      "loss": 0.636,
+      "step": 9551
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014987291169881765,
+      "loss": 0.674,
+      "step": 9552
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014986337574682517,
+      "loss": 0.7307,
+      "step": 9553
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001498538391913196,
+      "loss": 0.6408,
+      "step": 9554
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001498443020324164,
+      "loss": 0.6618,
+      "step": 9555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000149834764270231,
+      "loss": 0.6915,
+      "step": 9556
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014982522590487887,
+      "loss": 0.6856,
+      "step": 9557
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001498156869364754,
+      "loss": 0.684,
+      "step": 9558
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000149806147365136,
+      "loss": 0.6974,
+      "step": 9559
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014979660719097626,
+      "loss": 0.6569,
+      "step": 9560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014978706641411156,
+      "loss": 0.6696,
+      "step": 9561
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014977752503465735,
+      "loss": 0.7409,
+      "step": 9562
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001497679830527292,
+      "loss": 0.6816,
+      "step": 9563
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014975844046844253,
+      "loss": 0.7689,
+      "step": 9564
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001497488972819129,
+      "loss": 0.7806,
+      "step": 9565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014973935349325577,
+      "loss": 0.624,
+      "step": 9566
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014972980910258664,
+      "loss": 0.6963,
+      "step": 9567
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014972026411002107,
+      "loss": 0.6795,
+      "step": 9568
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014971071851567453,
+      "loss": 0.725,
+      "step": 9569
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001497011723196626,
+      "loss": 0.7003,
+      "step": 9570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014969162552210083,
+      "loss": 0.6989,
+      "step": 9571
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014968207812310474,
+      "loss": 0.7446,
+      "step": 9572
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014967253012278988,
+      "loss": 0.721,
+      "step": 9573
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014966298152127182,
+      "loss": 0.7079,
+      "step": 9574
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014965343231866616,
+      "loss": 0.7789,
+      "step": 9575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014964388251508844,
+      "loss": 0.6543,
+      "step": 9576
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014963433211065424,
+      "loss": 0.7622,
+      "step": 9577
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014962478110547918,
+      "loss": 0.7572,
+      "step": 9578
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014961522949967886,
+      "loss": 0.6639,
+      "step": 9579
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014960567729336882,
+      "loss": 0.7014,
+      "step": 9580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001495961244866648,
+      "loss": 0.6976,
+      "step": 9581
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001495865710796823,
+      "loss": 0.6805,
+      "step": 9582
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000149577017072537,
+      "loss": 0.7168,
+      "step": 9583
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014956746246534456,
+      "loss": 0.6424,
+      "step": 9584
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014955790725822057,
+      "loss": 0.6747,
+      "step": 9585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001495483514512807,
+      "loss": 0.5688,
+      "step": 9586
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001495387950446406,
+      "loss": 0.698,
+      "step": 9587
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000149529238038416,
+      "loss": 0.6463,
+      "step": 9588
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014951968043272246,
+      "loss": 0.7103,
+      "step": 9589
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014951012222767573,
+      "loss": 0.6729,
+      "step": 9590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001495005634233915,
+      "loss": 0.7382,
+      "step": 9591
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001494910040199854,
+      "loss": 0.7215,
+      "step": 9592
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014948144401757323,
+      "loss": 0.8273,
+      "step": 9593
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014947188341627064,
+      "loss": 0.7806,
+      "step": 9594
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001494623222161933,
+      "loss": 0.7364,
+      "step": 9595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000149452760417457,
+      "loss": 0.7262,
+      "step": 9596
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014944319802017748,
+      "loss": 0.7472,
+      "step": 9597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001494336350244704,
+      "loss": 0.7176,
+      "step": 9598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001494240714304516,
+      "loss": 0.7879,
+      "step": 9599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014941450723823677,
+      "loss": 0.5709,
+      "step": 9600
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 0.6929077506065369,
+      "eval_runtime": 844.5492,
+      "eval_samples_per_second": 28.489,
+      "eval_steps_per_second": 14.244,
+      "step": 9600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014940494244794166,
+      "loss": 0.7003,
+      "step": 9601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014939537705968206,
+      "loss": 0.6017,
+      "step": 9602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014938581107357376,
+      "loss": 0.7168,
+      "step": 9603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001493762444897325,
+      "loss": 0.7308,
+      "step": 9604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014936667730827412,
+      "loss": 0.6464,
+      "step": 9605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014935710952931436,
+      "loss": 0.6989,
+      "step": 9606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014934754115296904,
+      "loss": 0.8254,
+      "step": 9607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014933797217935398,
+      "loss": 0.6325,
+      "step": 9608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000149328402608585,
+      "loss": 0.6914,
+      "step": 9609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014931883244077794,
+      "loss": 0.6799,
+      "step": 9610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014930926167604856,
+      "loss": 0.8221,
+      "step": 9611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014929969031451275,
+      "loss": 0.8186,
+      "step": 9612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014929011835628634,
+      "loss": 0.7858,
+      "step": 9613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014928054580148523,
+      "loss": 0.7135,
+      "step": 9614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014927097265022524,
+      "loss": 0.5722,
+      "step": 9615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001492613989026222,
+      "loss": 0.6868,
+      "step": 9616
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014925182455879205,
+      "loss": 0.6297,
+      "step": 9617
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014924224961885066,
+      "loss": 0.7074,
+      "step": 9618
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014923267408291387,
+      "loss": 0.6552,
+      "step": 9619
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014922309795109762,
+      "loss": 0.5837,
+      "step": 9620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001492135212235178,
+      "loss": 0.68,
+      "step": 9621
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014920394390029033,
+      "loss": 0.8422,
+      "step": 9622
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001491943659815311,
+      "loss": 0.7183,
+      "step": 9623
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014918478746735607,
+      "loss": 0.5871,
+      "step": 9624
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014917520835788114,
+      "loss": 0.735,
+      "step": 9625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001491656286532223,
+      "loss": 0.637,
+      "step": 9626
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014915604835349545,
+      "loss": 0.6623,
+      "step": 9627
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014914646745881654,
+      "loss": 0.6917,
+      "step": 9628
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014913688596930155,
+      "loss": 0.716,
+      "step": 9629
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014912730388506645,
+      "loss": 0.6552,
+      "step": 9630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014911772120622721,
+      "loss": 0.7392,
+      "step": 9631
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001491081379328998,
+      "loss": 0.6159,
+      "step": 9632
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014909855406520023,
+      "loss": 0.6524,
+      "step": 9633
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014908896960324453,
+      "loss": 0.7406,
+      "step": 9634
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014907938454714861,
+      "loss": 0.7086,
+      "step": 9635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014906979889702852,
+      "loss": 0.6718,
+      "step": 9636
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014906021265300034,
+      "loss": 0.7077,
+      "step": 9637
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014905062581518002,
+      "loss": 0.7288,
+      "step": 9638
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001490410383836836,
+      "loss": 0.7346,
+      "step": 9639
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001490314503586272,
+      "loss": 0.6233,
+      "step": 9640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014902186174012676,
+      "loss": 0.7514,
+      "step": 9641
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001490122725282984,
+      "loss": 0.6812,
+      "step": 9642
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014900268272325816,
+      "loss": 0.7403,
+      "step": 9643
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001489930923251221,
+      "loss": 0.7396,
+      "step": 9644
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001489835013340063,
+      "loss": 0.5924,
+      "step": 9645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014897390975002686,
+      "loss": 0.7661,
+      "step": 9646
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014896431757329987,
+      "loss": 0.6906,
+      "step": 9647
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014895472480394142,
+      "loss": 0.6078,
+      "step": 9648
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001489451314420676,
+      "loss": 0.8429,
+      "step": 9649
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014893553748779453,
+      "loss": 0.6339,
+      "step": 9650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001489259429412383,
+      "loss": 0.5869,
+      "step": 9651
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014891634780251514,
+      "loss": 0.7365,
+      "step": 9652
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001489067520717411,
+      "loss": 0.7092,
+      "step": 9653
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014889715574903226,
+      "loss": 0.7658,
+      "step": 9654
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001488875588345049,
+      "loss": 0.6706,
+      "step": 9655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001488779613282751,
+      "loss": 0.7744,
+      "step": 9656
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014886836323045906,
+      "loss": 0.6721,
+      "step": 9657
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001488587645411729,
+      "loss": 0.7996,
+      "step": 9658
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001488491652605328,
+      "loss": 0.6958,
+      "step": 9659
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000148839565388655,
+      "loss": 0.8473,
+      "step": 9660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014882996492565566,
+      "loss": 0.654,
+      "step": 9661
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014882036387165094,
+      "loss": 0.6992,
+      "step": 9662
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001488107622267571,
+      "loss": 0.7301,
+      "step": 9663
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001488011599910903,
+      "loss": 0.7138,
+      "step": 9664
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001487915571647668,
+      "loss": 0.6461,
+      "step": 9665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014878195374790285,
+      "loss": 0.6596,
+      "step": 9666
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001487723497406146,
+      "loss": 0.6736,
+      "step": 9667
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014876274514301834,
+      "loss": 0.7443,
+      "step": 9668
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014875313995523035,
+      "loss": 0.6543,
+      "step": 9669
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014874353417736685,
+      "loss": 0.713,
+      "step": 9670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014873392780954408,
+      "loss": 0.6659,
+      "step": 9671
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014872432085187833,
+      "loss": 0.6565,
+      "step": 9672
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014871471330448588,
+      "loss": 0.7451,
+      "step": 9673
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000148705105167483,
+      "loss": 0.8272,
+      "step": 9674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014869549644098602,
+      "loss": 0.7752,
+      "step": 9675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001486858871251112,
+      "loss": 0.6887,
+      "step": 9676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014867627721997482,
+      "loss": 0.658,
+      "step": 9677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014866666672569328,
+      "loss": 0.6826,
+      "step": 9678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014865705564238283,
+      "loss": 0.6666,
+      "step": 9679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001486474439701598,
+      "loss": 0.6032,
+      "step": 9680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014863783170914054,
+      "loss": 0.6453,
+      "step": 9681
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014862821885944135,
+      "loss": 0.642,
+      "step": 9682
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014861860542117868,
+      "loss": 0.7356,
+      "step": 9683
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014860899139446877,
+      "loss": 0.6037,
+      "step": 9684
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014859937677942804,
+      "loss": 0.5945,
+      "step": 9685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014858976157617284,
+      "loss": 0.6479,
+      "step": 9686
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014858014578481956,
+      "loss": 0.7724,
+      "step": 9687
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014857052940548456,
+      "loss": 0.7872,
+      "step": 9688
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014856091243828424,
+      "loss": 0.6421,
+      "step": 9689
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014855129488333506,
+      "loss": 0.5876,
+      "step": 9690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014854167674075333,
+      "loss": 0.6923,
+      "step": 9691
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014853205801065552,
+      "loss": 0.6817,
+      "step": 9692
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014852243869315797,
+      "loss": 0.7301,
+      "step": 9693
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014851281878837725,
+      "loss": 0.7571,
+      "step": 9694
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014850319829642965,
+      "loss": 0.6699,
+      "step": 9695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014849357721743168,
+      "loss": 0.7246,
+      "step": 9696
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014848395555149978,
+      "loss": 0.7692,
+      "step": 9697
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001484743332987504,
+      "loss": 0.7192,
+      "step": 9698
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014846471045929999,
+      "loss": 0.6231,
+      "step": 9699
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014845508703326504,
+      "loss": 0.6794,
+      "step": 9700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000148445463020762,
+      "loss": 0.8043,
+      "step": 9701
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014843583842190738,
+      "loss": 0.6775,
+      "step": 9702
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014842621323681764,
+      "loss": 0.6594,
+      "step": 9703
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014841658746560934,
+      "loss": 0.6769,
+      "step": 9704
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014840696110839886,
+      "loss": 0.7591,
+      "step": 9705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014839733416530286,
+      "loss": 0.692,
+      "step": 9706
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014838770663643772,
+      "loss": 0.7557,
+      "step": 9707
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001483780785219201,
+      "loss": 0.7142,
+      "step": 9708
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014836844982186643,
+      "loss": 0.8,
+      "step": 9709
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001483588205363933,
+      "loss": 0.6373,
+      "step": 9710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014834919066561723,
+      "loss": 0.6468,
+      "step": 9711
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014833956020965478,
+      "loss": 0.7349,
+      "step": 9712
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014832992916862252,
+      "loss": 0.6196,
+      "step": 9713
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014832029754263698,
+      "loss": 0.6989,
+      "step": 9714
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014831066533181478,
+      "loss": 0.7997,
+      "step": 9715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014830103253627252,
+      "loss": 0.654,
+      "step": 9716
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014829139915612676,
+      "loss": 0.7275,
+      "step": 9717
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014828176519149407,
+      "loss": 0.6217,
+      "step": 9718
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014827213064249107,
+      "loss": 0.6263,
+      "step": 9719
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014826249550923436,
+      "loss": 0.8012,
+      "step": 9720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014825285979184062,
+      "loss": 0.643,
+      "step": 9721
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001482432234904264,
+      "loss": 0.6626,
+      "step": 9722
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014823358660510838,
+      "loss": 0.8288,
+      "step": 9723
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014822394913600314,
+      "loss": 0.6958,
+      "step": 9724
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014821431108322738,
+      "loss": 0.7621,
+      "step": 9725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014820467244689779,
+      "loss": 0.6399,
+      "step": 9726
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001481950332271309,
+      "loss": 0.792,
+      "step": 9727
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014818539342404348,
+      "loss": 0.7648,
+      "step": 9728
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014817575303775217,
+      "loss": 0.6655,
+      "step": 9729
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014816611206837368,
+      "loss": 0.7412,
+      "step": 9730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014815647051602464,
+      "loss": 0.6625,
+      "step": 9731
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014814682838082182,
+      "loss": 0.6235,
+      "step": 9732
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014813718566288183,
+      "loss": 0.7462,
+      "step": 9733
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014812754236232146,
+      "loss": 0.6509,
+      "step": 9734
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001481178984792574,
+      "loss": 0.7405,
+      "step": 9735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014810825401380637,
+      "loss": 0.7473,
+      "step": 9736
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001480986089660851,
+      "loss": 0.7345,
+      "step": 9737
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014808896333621033,
+      "loss": 0.691,
+      "step": 9738
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001480793171242988,
+      "loss": 0.7784,
+      "step": 9739
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014806967033046726,
+      "loss": 0.7212,
+      "step": 9740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014806002295483247,
+      "loss": 0.7314,
+      "step": 9741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001480503749975112,
+      "loss": 0.6856,
+      "step": 9742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001480407264586202,
+      "loss": 0.6631,
+      "step": 9743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001480310773382763,
+      "loss": 0.7074,
+      "step": 9744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014802142763659624,
+      "loss": 0.6845,
+      "step": 9745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001480117773536968,
+      "loss": 0.6449,
+      "step": 9746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014800212648969485,
+      "loss": 0.6173,
+      "step": 9747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014799247504470716,
+      "loss": 0.6879,
+      "step": 9748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014798282301885048,
+      "loss": 0.6945,
+      "step": 9749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014797317041224175,
+      "loss": 0.7791,
+      "step": 9750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014796351722499773,
+      "loss": 0.6691,
+      "step": 9751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014795386345723527,
+      "loss": 0.678,
+      "step": 9752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001479442091090712,
+      "loss": 0.7144,
+      "step": 9753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001479345541806224,
+      "loss": 0.6559,
+      "step": 9754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001479248986720057,
+      "loss": 0.7038,
+      "step": 9755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014791524258333795,
+      "loss": 0.7641,
+      "step": 9756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014790558591473607,
+      "loss": 0.6561,
+      "step": 9757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001478959286663169,
+      "loss": 0.6102,
+      "step": 9758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014788627083819731,
+      "loss": 0.6061,
+      "step": 9759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014787661243049428,
+      "loss": 0.6929,
+      "step": 9760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001478669534433246,
+      "loss": 0.7475,
+      "step": 9761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001478572938768052,
+      "loss": 0.7347,
+      "step": 9762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014784763373105304,
+      "loss": 0.6677,
+      "step": 9763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000147837973006185,
+      "loss": 0.5873,
+      "step": 9764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014782831170231803,
+      "loss": 0.6906,
+      "step": 9765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014781864981956906,
+      "loss": 0.7043,
+      "step": 9766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014780898735805505,
+      "loss": 0.6992,
+      "step": 9767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001477993243178929,
+      "loss": 0.594,
+      "step": 9768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014778966069919956,
+      "loss": 0.6933,
+      "step": 9769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014777999650209206,
+      "loss": 0.7234,
+      "step": 9770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001477703317266873,
+      "loss": 0.745,
+      "step": 9771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014776066637310232,
+      "loss": 0.5699,
+      "step": 9772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014775100044145406,
+      "loss": 0.7573,
+      "step": 9773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001477413339318595,
+      "loss": 0.6575,
+      "step": 9774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001477316668444357,
+      "loss": 0.6566,
+      "step": 9775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014772199917929957,
+      "loss": 0.7162,
+      "step": 9776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014771233093656817,
+      "loss": 0.7726,
+      "step": 9777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014770266211635856,
+      "loss": 0.6368,
+      "step": 9778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001476929927187877,
+      "loss": 0.6667,
+      "step": 9779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014768332274397265,
+      "loss": 0.6747,
+      "step": 9780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014767365219203044,
+      "loss": 0.6285,
+      "step": 9781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014766398106307813,
+      "loss": 0.6736,
+      "step": 9782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014765430935723278,
+      "loss": 0.6683,
+      "step": 9783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014764463707461138,
+      "loss": 0.6636,
+      "step": 9784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001476349642153311,
+      "loss": 0.6525,
+      "step": 9785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014762529077950892,
+      "loss": 0.6876,
+      "step": 9786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000147615616767262,
+      "loss": 0.6777,
+      "step": 9787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014760594217870737,
+      "loss": 0.6514,
+      "step": 9788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014759626701396216,
+      "loss": 0.7016,
+      "step": 9789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014758659127314346,
+      "loss": 0.6299,
+      "step": 9790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001475769149563684,
+      "loss": 0.7141,
+      "step": 9791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014756723806375405,
+      "loss": 0.6832,
+      "step": 9792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014755756059541756,
+      "loss": 0.7086,
+      "step": 9793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014754788255147606,
+      "loss": 0.7218,
+      "step": 9794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014753820393204666,
+      "loss": 0.6926,
+      "step": 9795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014752852473724656,
+      "loss": 0.6094,
+      "step": 9796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014751884496719283,
+      "loss": 0.7057,
+      "step": 9797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014750916462200273,
+      "loss": 0.6316,
+      "step": 9798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014749948370179335,
+      "loss": 0.7011,
+      "step": 9799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014748980220668186,
+      "loss": 0.7092,
+      "step": 9800
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 0.6857539415359497,
+      "eval_runtime": 851.8389,
+      "eval_samples_per_second": 28.245,
+      "eval_steps_per_second": 14.122,
+      "step": 9800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014748012013678547,
+      "loss": 0.6568,
+      "step": 9801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014747043749222135,
+      "loss": 0.6704,
+      "step": 9802
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001474607542731067,
+      "loss": 0.6512,
+      "step": 9803
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014745107047955873,
+      "loss": 0.7136,
+      "step": 9804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014744138611169462,
+      "loss": 0.6816,
+      "step": 9805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001474317011696316,
+      "loss": 0.7439,
+      "step": 9806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014742201565348686,
+      "loss": 0.7788,
+      "step": 9807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014741232956337768,
+      "loss": 0.6647,
+      "step": 9808
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014740264289942126,
+      "loss": 0.7018,
+      "step": 9809
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014739295566173482,
+      "loss": 0.5794,
+      "step": 9810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001473832678504357,
+      "loss": 0.7173,
+      "step": 9811
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014737357946564104,
+      "loss": 0.7222,
+      "step": 9812
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014736389050746813,
+      "loss": 0.6616,
+      "step": 9813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001473542009760343,
+      "loss": 0.696,
+      "step": 9814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014734451087145676,
+      "loss": 0.767,
+      "step": 9815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014733482019385283,
+      "loss": 0.7956,
+      "step": 9816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001473251289433398,
+      "loss": 0.6233,
+      "step": 9817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014731543712003493,
+      "loss": 0.7159,
+      "step": 9818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014730574472405556,
+      "loss": 0.6656,
+      "step": 9819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000147296051755519,
+      "loss": 0.6646,
+      "step": 9820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014728635821454255,
+      "loss": 0.7205,
+      "step": 9821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001472766641012435,
+      "loss": 0.7173,
+      "step": 9822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001472669694157393,
+      "loss": 0.8046,
+      "step": 9823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014725727415814714,
+      "loss": 0.6921,
+      "step": 9824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001472475783285844,
+      "loss": 0.6844,
+      "step": 9825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014723788192716857,
+      "loss": 0.6078,
+      "step": 9826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014722818495401682,
+      "loss": 0.6636,
+      "step": 9827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014721848740924663,
+      "loss": 0.6119,
+      "step": 9828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014720878929297534,
+      "loss": 0.7655,
+      "step": 9829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014719909060532032,
+      "loss": 0.7851,
+      "step": 9830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000147189391346399,
+      "loss": 0.6913,
+      "step": 9831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014717969151632873,
+      "loss": 0.7167,
+      "step": 9832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001471699911152269,
+      "loss": 0.7348,
+      "step": 9833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014716029014321096,
+      "loss": 0.6969,
+      "step": 9834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014715058860039827,
+      "loss": 0.7799,
+      "step": 9835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014714088648690632,
+      "loss": 0.7252,
+      "step": 9836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001471311838028525,
+      "loss": 0.6699,
+      "step": 9837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001471214805483542,
+      "loss": 0.5953,
+      "step": 9838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014711177672352895,
+      "loss": 0.6533,
+      "step": 9839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014710207232849416,
+      "loss": 0.742,
+      "step": 9840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014709236736336727,
+      "loss": 0.7229,
+      "step": 9841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001470826618282658,
+      "loss": 0.7986,
+      "step": 9842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014707295572330707,
+      "loss": 0.7496,
+      "step": 9843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014706324904860876,
+      "loss": 0.694,
+      "step": 9844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014705354180428822,
+      "loss": 0.6296,
+      "step": 9845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014704383399046295,
+      "loss": 0.6945,
+      "step": 9846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001470341256072505,
+      "loss": 0.6657,
+      "step": 9847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014702441665476833,
+      "loss": 0.6407,
+      "step": 9848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014701470713313398,
+      "loss": 0.7394,
+      "step": 9849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014700499704246492,
+      "loss": 0.724,
+      "step": 9850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014699528638287873,
+      "loss": 0.7018,
+      "step": 9851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001469855751544929,
+      "loss": 0.6749,
+      "step": 9852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000146975863357425,
+      "loss": 0.6401,
+      "step": 9853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014696615099179258,
+      "loss": 0.88,
+      "step": 9854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014695643805771318,
+      "loss": 0.6613,
+      "step": 9855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001469467245553043,
+      "loss": 0.6141,
+      "step": 9856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001469370104846836,
+      "loss": 0.7279,
+      "step": 9857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001469272958459686,
+      "loss": 0.8169,
+      "step": 9858
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001469175806392769,
+      "loss": 0.7355,
+      "step": 9859
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014690786486472608,
+      "loss": 0.7573,
+      "step": 9860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014689814852243368,
+      "loss": 0.6364,
+      "step": 9861
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014688843161251742,
+      "loss": 0.7509,
+      "step": 9862
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001468787141350948,
+      "loss": 0.7436,
+      "step": 9863
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014686899609028348,
+      "loss": 0.6355,
+      "step": 9864
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014685927747820109,
+      "loss": 0.7383,
+      "step": 9865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001468495582989652,
+      "loss": 0.755,
+      "step": 9866
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001468398385526935,
+      "loss": 0.7336,
+      "step": 9867
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001468301182395036,
+      "loss": 0.6611,
+      "step": 9868
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001468203973595132,
+      "loss": 0.6695,
+      "step": 9869
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014681067591283987,
+      "loss": 0.7671,
+      "step": 9870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014680095389960136,
+      "loss": 0.6638,
+      "step": 9871
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001467912313199153,
+      "loss": 0.8867,
+      "step": 9872
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014678150817389935,
+      "loss": 0.6875,
+      "step": 9873
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014677178446167118,
+      "loss": 0.6705,
+      "step": 9874
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014676206018334858,
+      "loss": 0.6732,
+      "step": 9875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001467523353390491,
+      "loss": 0.7969,
+      "step": 9876
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014674260992889054,
+      "loss": 0.6274,
+      "step": 9877
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014673288395299062,
+      "loss": 0.6493,
+      "step": 9878
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014672315741146697,
+      "loss": 0.7507,
+      "step": 9879
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001467134303044374,
+      "loss": 0.5905,
+      "step": 9880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001467037026320196,
+      "loss": 0.612,
+      "step": 9881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001466939743943313,
+      "loss": 0.8027,
+      "step": 9882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014668424559149027,
+      "loss": 0.6796,
+      "step": 9883
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014667451622361424,
+      "loss": 0.563,
+      "step": 9884
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000146664786290821,
+      "loss": 0.6848,
+      "step": 9885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014665505579322828,
+      "loss": 0.7368,
+      "step": 9886
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014664532473095387,
+      "loss": 0.6028,
+      "step": 9887
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001466355931041155,
+      "loss": 0.6735,
+      "step": 9888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014662586091283104,
+      "loss": 0.5974,
+      "step": 9889
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014661612815721823,
+      "loss": 0.6353,
+      "step": 9890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014660639483739485,
+      "loss": 0.7137,
+      "step": 9891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014659666095347877,
+      "loss": 0.6649,
+      "step": 9892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014658692650558773,
+      "loss": 0.7667,
+      "step": 9893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001465771914938396,
+      "loss": 0.654,
+      "step": 9894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001465674559183522,
+      "loss": 0.6646,
+      "step": 9895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014655771977924336,
+      "loss": 0.6994,
+      "step": 9896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001465479830766309,
+      "loss": 0.6971,
+      "step": 9897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014653824581063266,
+      "loss": 0.7171,
+      "step": 9898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014652850798136652,
+      "loss": 0.6812,
+      "step": 9899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014651876958895034,
+      "loss": 0.73,
+      "step": 9900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000146509030633502,
+      "loss": 0.7356,
+      "step": 9901
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001464992911151393,
+      "loss": 0.5871,
+      "step": 9902
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014648955103398024,
+      "loss": 0.596,
+      "step": 9903
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014647981039014258,
+      "loss": 0.5974,
+      "step": 9904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014647006918374432,
+      "loss": 0.6492,
+      "step": 9905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014646032741490333,
+      "loss": 0.6412,
+      "step": 9906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014645058508373746,
+      "loss": 0.6687,
+      "step": 9907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001464408421903647,
+      "loss": 0.7139,
+      "step": 9908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001464310987349029,
+      "loss": 0.7957,
+      "step": 9909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014642135471747006,
+      "loss": 0.6793,
+      "step": 9910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001464116101381841,
+      "loss": 0.6599,
+      "step": 9911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014640186499716292,
+      "loss": 0.6674,
+      "step": 9912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014639211929452453,
+      "loss": 0.7744,
+      "step": 9913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014638237303038682,
+      "loss": 0.6985,
+      "step": 9914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014637262620486777,
+      "loss": 0.7579,
+      "step": 9915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014636287881808542,
+      "loss": 0.7514,
+      "step": 9916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014635313087015762,
+      "loss": 0.7438,
+      "step": 9917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014634338236120247,
+      "loss": 0.6263,
+      "step": 9918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014633363329133792,
+      "loss": 0.6344,
+      "step": 9919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014632388366068191,
+      "loss": 0.7524,
+      "step": 9920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001463141334693525,
+      "loss": 0.7078,
+      "step": 9921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001463043827174677,
+      "loss": 0.6229,
+      "step": 9922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001462946314051455,
+      "loss": 0.7512,
+      "step": 9923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014628487953250395,
+      "loss": 0.7683,
+      "step": 9924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014627512709966108,
+      "loss": 0.7245,
+      "step": 9925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001462653741067349,
+      "loss": 0.5921,
+      "step": 9926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014625562055384347,
+      "loss": 0.6236,
+      "step": 9927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014624586644110485,
+      "loss": 0.7218,
+      "step": 9928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001462361117686371,
+      "loss": 0.646,
+      "step": 9929
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014622635653655826,
+      "loss": 0.7494,
+      "step": 9930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001462166007449864,
+      "loss": 0.5754,
+      "step": 9931
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014620684439403962,
+      "loss": 0.6508,
+      "step": 9932
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000146197087483836,
+      "loss": 0.7422,
+      "step": 9933
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014618733001449364,
+      "loss": 0.7126,
+      "step": 9934
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014617757198613054,
+      "loss": 0.6249,
+      "step": 9935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014616781339886497,
+      "loss": 0.7836,
+      "step": 9936
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014615805425281496,
+      "loss": 0.7326,
+      "step": 9937
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001461482945480986,
+      "loss": 0.6416,
+      "step": 9938
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014613853428483407,
+      "loss": 0.6943,
+      "step": 9939
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014612877346313944,
+      "loss": 0.7213,
+      "step": 9940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014611901208313292,
+      "loss": 0.7205,
+      "step": 9941
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001461092501449326,
+      "loss": 0.6667,
+      "step": 9942
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014609948764865663,
+      "loss": 0.6607,
+      "step": 9943
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014608972459442322,
+      "loss": 0.6305,
+      "step": 9944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001460799609823505,
+      "loss": 0.8489,
+      "step": 9945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014607019681255664,
+      "loss": 0.6974,
+      "step": 9946
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014606043208515983,
+      "loss": 0.7546,
+      "step": 9947
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014605066680027824,
+      "loss": 0.7161,
+      "step": 9948
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014604090095803008,
+      "loss": 0.7532,
+      "step": 9949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014603113455853356,
+      "loss": 0.7173,
+      "step": 9950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014602136760190684,
+      "loss": 0.743,
+      "step": 9951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014601160008826817,
+      "loss": 0.647,
+      "step": 9952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014600183201773575,
+      "loss": 0.7522,
+      "step": 9953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014599206339042787,
+      "loss": 0.6655,
+      "step": 9954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014598229420646267,
+      "loss": 0.6729,
+      "step": 9955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014597252446595845,
+      "loss": 0.6872,
+      "step": 9956
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014596275416903343,
+      "loss": 0.6423,
+      "step": 9957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014595298331580584,
+      "loss": 0.7598,
+      "step": 9958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000145943211906394,
+      "loss": 0.6018,
+      "step": 9959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014593343994091616,
+      "loss": 0.7755,
+      "step": 9960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014592366741949056,
+      "loss": 0.7812,
+      "step": 9961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001459138943422355,
+      "loss": 0.7187,
+      "step": 9962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014590412070926924,
+      "loss": 0.8222,
+      "step": 9963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014589434652071014,
+      "loss": 0.6048,
+      "step": 9964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014588457177667645,
+      "loss": 0.6662,
+      "step": 9965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001458747964772865,
+      "loss": 0.7448,
+      "step": 9966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014586502062265857,
+      "loss": 0.8046,
+      "step": 9967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014585524421291098,
+      "loss": 0.7104,
+      "step": 9968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014584546724816211,
+      "loss": 0.5959,
+      "step": 9969
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001458356897285303,
+      "loss": 0.693,
+      "step": 9970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014582591165413378,
+      "loss": 0.6459,
+      "step": 9971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014581613302509103,
+      "loss": 0.8143,
+      "step": 9972
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014580635384152027,
+      "loss": 0.6559,
+      "step": 9973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014579657410354003,
+      "loss": 0.8361,
+      "step": 9974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014578679381126853,
+      "loss": 0.6913,
+      "step": 9975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001457770129648242,
+      "loss": 0.8008,
+      "step": 9976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014576723156432543,
+      "loss": 0.6906,
+      "step": 9977
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014575744960989057,
+      "loss": 0.748,
+      "step": 9978
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014574766710163806,
+      "loss": 0.6598,
+      "step": 9979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001457378840396863,
+      "loss": 0.8124,
+      "step": 9980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014572810042415364,
+      "loss": 0.589,
+      "step": 9981
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014571831625515854,
+      "loss": 0.6428,
+      "step": 9982
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014570853153281942,
+      "loss": 0.6114,
+      "step": 9983
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001456987462572547,
+      "loss": 0.6842,
+      "step": 9984
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014568896042858282,
+      "loss": 0.7202,
+      "step": 9985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001456791740469222,
+      "loss": 0.7453,
+      "step": 9986
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014566938711239134,
+      "loss": 0.804,
+      "step": 9987
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014565959962510865,
+      "loss": 0.7355,
+      "step": 9988
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014564981158519258,
+      "loss": 0.7099,
+      "step": 9989
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014564002299276163,
+      "loss": 0.7864,
+      "step": 9990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014563023384793423,
+      "loss": 0.6847,
+      "step": 9991
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014562044415082897,
+      "loss": 0.5688,
+      "step": 9992
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001456106539015642,
+      "loss": 0.7289,
+      "step": 9993
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001456008631002585,
+      "loss": 0.6088,
+      "step": 9994
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014559107174703034,
+      "loss": 0.8093,
+      "step": 9995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014558127984199825,
+      "loss": 0.7421,
+      "step": 9996
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001455714873852807,
+      "loss": 0.6818,
+      "step": 9997
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014556169437699628,
+      "loss": 0.6531,
+      "step": 9998
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014555190081726345,
+      "loss": 0.622,
+      "step": 9999
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014554210670620078,
+      "loss": 0.7396,
+      "step": 10000
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 0.677370548248291,
+      "eval_runtime": 845.2459,
+      "eval_samples_per_second": 28.465,
+      "eval_steps_per_second": 14.233,
+      "step": 10000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014553231204392683,
+      "loss": 0.7027,
+      "step": 10001
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001455225168305601,
+      "loss": 0.5885,
+      "step": 10002
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014551272106621917,
+      "loss": 0.7296,
+      "step": 10003
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014550292475102263,
+      "loss": 0.7367,
+      "step": 10004
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000145493127885089,
+      "loss": 0.6733,
+      "step": 10005
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014548333046853688,
+      "loss": 0.6584,
+      "step": 10006
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014547353250148482,
+      "loss": 0.5784,
+      "step": 10007
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014546373398405143,
+      "loss": 0.6649,
+      "step": 10008
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014545393491635533,
+      "loss": 0.5912,
+      "step": 10009
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001454441352985151,
+      "loss": 0.705,
+      "step": 10010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014543433513064936,
+      "loss": 0.7307,
+      "step": 10011
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001454245344128767,
+      "loss": 0.5677,
+      "step": 10012
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014541473314531574,
+      "loss": 0.677,
+      "step": 10013
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014540493132808515,
+      "loss": 0.697,
+      "step": 10014
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014539512896130348,
+      "loss": 0.6958,
+      "step": 10015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014538532604508946,
+      "loss": 0.5727,
+      "step": 10016
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014537552257956173,
+      "loss": 0.6506,
+      "step": 10017
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001453657185648389,
+      "loss": 0.7106,
+      "step": 10018
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014535591400103966,
+      "loss": 0.5867,
+      "step": 10019
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014534610888828264,
+      "loss": 0.6386,
+      "step": 10020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014533630322668658,
+      "loss": 0.6723,
+      "step": 10021
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001453264970163701,
+      "loss": 0.6509,
+      "step": 10022
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014531669025745193,
+      "loss": 0.73,
+      "step": 10023
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014530688295005071,
+      "loss": 0.7416,
+      "step": 10024
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014529707509428522,
+      "loss": 0.5682,
+      "step": 10025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014528726669027413,
+      "loss": 0.6722,
+      "step": 10026
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001452774577381361,
+      "loss": 0.7332,
+      "step": 10027
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014526764823798996,
+      "loss": 0.6177,
+      "step": 10028
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014525783818995433,
+      "loss": 0.7572,
+      "step": 10029
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014524802759414802,
+      "loss": 0.7057,
+      "step": 10030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014523821645068974,
+      "loss": 0.7497,
+      "step": 10031
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014522840475969824,
+      "loss": 0.7119,
+      "step": 10032
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014521859252129228,
+      "loss": 0.6106,
+      "step": 10033
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001452087797355906,
+      "loss": 0.706,
+      "step": 10034
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014519896640271198,
+      "loss": 0.7554,
+      "step": 10035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014518915252277522,
+      "loss": 0.6598,
+      "step": 10036
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014517933809589907,
+      "loss": 0.7119,
+      "step": 10037
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014516952312220233,
+      "loss": 0.6331,
+      "step": 10038
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001451597076018038,
+      "loss": 0.5961,
+      "step": 10039
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014514989153482225,
+      "loss": 0.5399,
+      "step": 10040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001451400749213765,
+      "loss": 0.691,
+      "step": 10041
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001451302577615854,
+      "loss": 0.5674,
+      "step": 10042
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014512044005556774,
+      "loss": 0.7024,
+      "step": 10043
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014511062180344234,
+      "loss": 0.7091,
+      "step": 10044
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014510080300532803,
+      "loss": 0.7308,
+      "step": 10045
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014509098366134366,
+      "loss": 0.8364,
+      "step": 10046
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014508116377160812,
+      "loss": 0.652,
+      "step": 10047
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014507134333624017,
+      "loss": 0.6153,
+      "step": 10048
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014506152235535874,
+      "loss": 0.8215,
+      "step": 10049
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001450517008290827,
+      "loss": 0.7976,
+      "step": 10050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014504187875753088,
+      "loss": 0.6351,
+      "step": 10051
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001450320561408222,
+      "loss": 0.8004,
+      "step": 10052
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001450222329790755,
+      "loss": 0.7525,
+      "step": 10053
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014501240927240975,
+      "loss": 0.6772,
+      "step": 10054
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014500258502094372,
+      "loss": 0.7161,
+      "step": 10055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014499276022479645,
+      "loss": 0.6868,
+      "step": 10056
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014498293488408683,
+      "loss": 0.6504,
+      "step": 10057
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001449731089989337,
+      "loss": 0.7134,
+      "step": 10058
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014496328256945605,
+      "loss": 0.7553,
+      "step": 10059
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001449534555957728,
+      "loss": 0.6751,
+      "step": 10060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001449436280780029,
+      "loss": 0.6664,
+      "step": 10061
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001449338000162653,
+      "loss": 0.6935,
+      "step": 10062
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014492397141067887,
+      "loss": 0.6526,
+      "step": 10063
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014491414226136269,
+      "loss": 0.7883,
+      "step": 10064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014490431256843566,
+      "loss": 0.713,
+      "step": 10065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014489448233201673,
+      "loss": 0.698,
+      "step": 10066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014488465155222495,
+      "loss": 0.5783,
+      "step": 10067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014487482022917925,
+      "loss": 0.6506,
+      "step": 10068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014486498836299862,
+      "loss": 0.5535,
+      "step": 10069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001448551559538021,
+      "loss": 0.7649,
+      "step": 10070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014484532300170866,
+      "loss": 0.6422,
+      "step": 10071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014483548950683733,
+      "loss": 0.626,
+      "step": 10072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001448256554693071,
+      "loss": 0.6279,
+      "step": 10073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014481582088923705,
+      "loss": 0.7194,
+      "step": 10074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014480598576674614,
+      "loss": 0.6138,
+      "step": 10075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014479615010195345,
+      "loss": 0.7637,
+      "step": 10076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014478631389497804,
+      "loss": 0.6884,
+      "step": 10077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014477647714593891,
+      "loss": 0.6258,
+      "step": 10078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014476663985495516,
+      "loss": 0.7844,
+      "step": 10079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014475680202214586,
+      "loss": 0.8542,
+      "step": 10080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014474696364763002,
+      "loss": 0.6206,
+      "step": 10081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001447371247315268,
+      "loss": 0.6456,
+      "step": 10082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014472728527395521,
+      "loss": 0.7203,
+      "step": 10083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001447174452750344,
+      "loss": 0.6351,
+      "step": 10084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014470760473488342,
+      "loss": 0.7154,
+      "step": 10085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014469776365362138,
+      "loss": 0.6543,
+      "step": 10086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014468792203136744,
+      "loss": 0.5489,
+      "step": 10087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014467807986824066,
+      "loss": 0.685,
+      "step": 10088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014466823716436018,
+      "loss": 0.7486,
+      "step": 10089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014465839391984513,
+      "loss": 0.6565,
+      "step": 10090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014464855013481462,
+      "loss": 0.5951,
+      "step": 10091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014463870580938787,
+      "loss": 0.6675,
+      "step": 10092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014462886094368394,
+      "loss": 0.674,
+      "step": 10093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014461901553782205,
+      "loss": 0.6247,
+      "step": 10094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014460916959192131,
+      "loss": 0.571,
+      "step": 10095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014459932310610093,
+      "loss": 0.6161,
+      "step": 10096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014458947608048009,
+      "loss": 0.7652,
+      "step": 10097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014457962851517793,
+      "loss": 0.7784,
+      "step": 10098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014456978041031365,
+      "loss": 0.7848,
+      "step": 10099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001445599317660065,
+      "loss": 0.6115,
+      "step": 10100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014455008258237558,
+      "loss": 0.7835,
+      "step": 10101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014454023285954018,
+      "loss": 0.613,
+      "step": 10102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014453038259761952,
+      "loss": 0.6302,
+      "step": 10103
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014452053179673274,
+      "loss": 0.6543,
+      "step": 10104
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014451068045699916,
+      "loss": 0.7752,
+      "step": 10105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014450082857853794,
+      "loss": 0.6517,
+      "step": 10106
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014449097616146838,
+      "loss": 0.6701,
+      "step": 10107
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001444811232059097,
+      "loss": 0.7142,
+      "step": 10108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014447126971198112,
+      "loss": 0.6843,
+      "step": 10109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014446141567980197,
+      "loss": 0.6652,
+      "step": 10110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014445156110949145,
+      "loss": 0.6622,
+      "step": 10111
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014444170600116888,
+      "loss": 0.6847,
+      "step": 10112
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014443185035495354,
+      "loss": 0.669,
+      "step": 10113
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014442199417096467,
+      "loss": 0.7622,
+      "step": 10114
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001444121374493216,
+      "loss": 0.6323,
+      "step": 10115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014440228019014365,
+      "loss": 0.7226,
+      "step": 10116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014439242239355004,
+      "loss": 0.6721,
+      "step": 10117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001443825640596602,
+      "loss": 0.6207,
+      "step": 10118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014437270518859334,
+      "loss": 0.7,
+      "step": 10119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014436284578046887,
+      "loss": 0.5758,
+      "step": 10120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014435298583540606,
+      "loss": 0.6719,
+      "step": 10121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014434312535352426,
+      "loss": 0.7493,
+      "step": 10122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014433326433494287,
+      "loss": 0.7906,
+      "step": 10123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014432340277978115,
+      "loss": 0.6143,
+      "step": 10124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014431354068815855,
+      "loss": 0.7772,
+      "step": 10125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014430367806019437,
+      "loss": 0.678,
+      "step": 10126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000144293814896008,
+      "loss": 0.6429,
+      "step": 10127
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014428395119571883,
+      "loss": 0.6763,
+      "step": 10128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001442740869594462,
+      "loss": 0.6435,
+      "step": 10129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014426422218730958,
+      "loss": 0.6407,
+      "step": 10130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001442543568794283,
+      "loss": 0.6748,
+      "step": 10131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014424449103592177,
+      "loss": 0.6054,
+      "step": 10132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014423462465690942,
+      "loss": 0.7312,
+      "step": 10133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014422475774251067,
+      "loss": 0.6166,
+      "step": 10134
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014421489029284488,
+      "loss": 0.6743,
+      "step": 10135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001442050223080316,
+      "loss": 0.6214,
+      "step": 10136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014419515378819015,
+      "loss": 0.681,
+      "step": 10137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014418528473344005,
+      "loss": 0.643,
+      "step": 10138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001441754151439007,
+      "loss": 0.7097,
+      "step": 10139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014416554501969154,
+      "loss": 0.6511,
+      "step": 10140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001441556743609321,
+      "loss": 0.6543,
+      "step": 10141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014414580316774177,
+      "loss": 0.6812,
+      "step": 10142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001441359314402401,
+      "loss": 0.6509,
+      "step": 10143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014412605917854654,
+      "loss": 0.6827,
+      "step": 10144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014411618638278056,
+      "loss": 0.5518,
+      "step": 10145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014410631305306166,
+      "loss": 0.7032,
+      "step": 10146
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014409643918950933,
+      "loss": 0.8308,
+      "step": 10147
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014408656479224311,
+      "loss": 0.743,
+      "step": 10148
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014407668986138252,
+      "loss": 0.6751,
+      "step": 10149
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014406681439704702,
+      "loss": 0.7356,
+      "step": 10150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014405693839935618,
+      "loss": 0.7034,
+      "step": 10151
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014404706186842953,
+      "loss": 0.7471,
+      "step": 10152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001440371848043866,
+      "loss": 0.6406,
+      "step": 10153
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014402730720734693,
+      "loss": 0.627,
+      "step": 10154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014401742907743008,
+      "loss": 0.7247,
+      "step": 10155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001440075504147556,
+      "loss": 0.6833,
+      "step": 10156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014399767121944313,
+      "loss": 0.7595,
+      "step": 10157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014398779149161212,
+      "loss": 0.733,
+      "step": 10158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001439779112313822,
+      "loss": 0.6526,
+      "step": 10159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014396803043887295,
+      "loss": 0.6521,
+      "step": 10160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014395814911420397,
+      "loss": 0.6743,
+      "step": 10161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014394826725749486,
+      "loss": 0.6755,
+      "step": 10162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014393838486886521,
+      "loss": 0.722,
+      "step": 10163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014392850194843465,
+      "loss": 0.7346,
+      "step": 10164
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014391861849632277,
+      "loss": 0.774,
+      "step": 10165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014390873451264924,
+      "loss": 0.7192,
+      "step": 10166
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014389884999753362,
+      "loss": 0.6827,
+      "step": 10167
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001438889649510956,
+      "loss": 0.6476,
+      "step": 10168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001438790793734548,
+      "loss": 0.7033,
+      "step": 10169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014386919326473083,
+      "loss": 0.6428,
+      "step": 10170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014385930662504342,
+      "loss": 0.6877,
+      "step": 10171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001438494194545122,
+      "loss": 0.7192,
+      "step": 10172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014383953175325681,
+      "loss": 0.6533,
+      "step": 10173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014382964352139697,
+      "loss": 0.6675,
+      "step": 10174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014381975475905235,
+      "loss": 0.705,
+      "step": 10175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001438098654663426,
+      "loss": 0.6948,
+      "step": 10176
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001437999756433875,
+      "loss": 0.6943,
+      "step": 10177
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001437900852903066,
+      "loss": 0.6754,
+      "step": 10178
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014378019440721974,
+      "loss": 0.5775,
+      "step": 10179
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001437703029942466,
+      "loss": 0.6232,
+      "step": 10180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014376041105150687,
+      "loss": 0.6992,
+      "step": 10181
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014375051857912032,
+      "loss": 0.6313,
+      "step": 10182
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001437406255772066,
+      "loss": 0.6978,
+      "step": 10183
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014373073204588556,
+      "loss": 0.5897,
+      "step": 10184
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014372083798527686,
+      "loss": 0.6608,
+      "step": 10185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014371094339550027,
+      "loss": 0.6403,
+      "step": 10186
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014370104827667558,
+      "loss": 0.7513,
+      "step": 10187
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001436911526289225,
+      "loss": 0.7305,
+      "step": 10188
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001436812564523609,
+      "loss": 0.6607,
+      "step": 10189
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014367135974711044,
+      "loss": 0.6051,
+      "step": 10190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001436614625132909,
+      "loss": 0.7397,
+      "step": 10191
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001436515647510222,
+      "loss": 0.6031,
+      "step": 10192
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000143641666460424,
+      "loss": 0.677,
+      "step": 10193
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001436317676416162,
+      "loss": 0.6864,
+      "step": 10194
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014362186829471857,
+      "loss": 0.6347,
+      "step": 10195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001436119684198509,
+      "loss": 0.6943,
+      "step": 10196
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014360206801713302,
+      "loss": 0.606,
+      "step": 10197
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014359216708668482,
+      "loss": 0.7592,
+      "step": 10198
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014358226562862604,
+      "loss": 0.7176,
+      "step": 10199
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014357236364307657,
+      "loss": 0.5967,
+      "step": 10200
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 0.6702423095703125,
+      "eval_runtime": 850.3575,
+      "eval_samples_per_second": 28.294,
+      "eval_steps_per_second": 14.147,
+      "step": 10200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014356246113015627,
+      "loss": 0.7325,
+      "step": 10201
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014355255808998498,
+      "loss": 0.7146,
+      "step": 10202
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014354265452268256,
+      "loss": 0.7795,
+      "step": 10203
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014353275042836884,
+      "loss": 0.5844,
+      "step": 10204
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014352284580716377,
+      "loss": 0.6284,
+      "step": 10205
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014351294065918714,
+      "loss": 0.6294,
+      "step": 10206
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014350303498455894,
+      "loss": 0.7694,
+      "step": 10207
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014349312878339895,
+      "loss": 0.6569,
+      "step": 10208
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014348322205582714,
+      "loss": 0.6025,
+      "step": 10209
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001434733148019634,
+      "loss": 0.6832,
+      "step": 10210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001434634070219276,
+      "loss": 0.7089,
+      "step": 10211
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001434534987158397,
+      "loss": 0.6066,
+      "step": 10212
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014344358988381965,
+      "loss": 0.7098,
+      "step": 10213
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001434336805259873,
+      "loss": 0.6479,
+      "step": 10214
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014342377064246268,
+      "loss": 0.6679,
+      "step": 10215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014341386023336566,
+      "loss": 0.5871,
+      "step": 10216
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014340394929881621,
+      "loss": 0.5967,
+      "step": 10217
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001433940378389343,
+      "loss": 0.6574,
+      "step": 10218
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014338412585383986,
+      "loss": 0.7416,
+      "step": 10219
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001433742133436529,
+      "loss": 0.7615,
+      "step": 10220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014336430030849337,
+      "loss": 0.7,
+      "step": 10221
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014335438674848125,
+      "loss": 0.7454,
+      "step": 10222
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014334447266373655,
+      "loss": 0.6757,
+      "step": 10223
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014333455805437924,
+      "loss": 0.6931,
+      "step": 10224
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001433246429205293,
+      "loss": 0.7224,
+      "step": 10225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014331472726230674,
+      "loss": 0.6548,
+      "step": 10226
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014330481107983163,
+      "loss": 0.5683,
+      "step": 10227
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014329489437322397,
+      "loss": 0.5982,
+      "step": 10228
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001432849771426037,
+      "loss": 0.7397,
+      "step": 10229
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000143275059388091,
+      "loss": 0.7089,
+      "step": 10230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014326514110980576,
+      "loss": 0.6457,
+      "step": 10231
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014325522230786811,
+      "loss": 0.7372,
+      "step": 10232
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014324530298239808,
+      "loss": 0.6505,
+      "step": 10233
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001432353831335157,
+      "loss": 0.5751,
+      "step": 10234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014322546276134115,
+      "loss": 0.6447,
+      "step": 10235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014321554186599435,
+      "loss": 0.6125,
+      "step": 10236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014320562044759544,
+      "loss": 0.6685,
+      "step": 10237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014319569850626448,
+      "loss": 0.6571,
+      "step": 10238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001431857760421216,
+      "loss": 0.7286,
+      "step": 10239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014317585305528686,
+      "loss": 0.7018,
+      "step": 10240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001431659295458804,
+      "loss": 0.7548,
+      "step": 10241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014315600551402229,
+      "loss": 0.6865,
+      "step": 10242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014314608095983264,
+      "loss": 0.6319,
+      "step": 10243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014313615588343158,
+      "loss": 0.6397,
+      "step": 10244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014312623028493927,
+      "loss": 0.7469,
+      "step": 10245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014311630416447582,
+      "loss": 0.6344,
+      "step": 10246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014310637752216131,
+      "loss": 0.7747,
+      "step": 10247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014309645035811598,
+      "loss": 0.6646,
+      "step": 10248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014308652267245995,
+      "loss": 0.666,
+      "step": 10249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014307659446531333,
+      "loss": 0.6488,
+      "step": 10250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014306666573679635,
+      "loss": 0.6693,
+      "step": 10251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014305673648702913,
+      "loss": 0.8382,
+      "step": 10252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001430468067161319,
+      "loss": 0.6615,
+      "step": 10253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014303687642422484,
+      "loss": 0.6626,
+      "step": 10254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014302694561142804,
+      "loss": 0.7158,
+      "step": 10255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014301701427786183,
+      "loss": 0.6788,
+      "step": 10256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001430070824236463,
+      "loss": 0.6496,
+      "step": 10257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014299715004890176,
+      "loss": 0.6726,
+      "step": 10258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014298721715374837,
+      "loss": 0.6161,
+      "step": 10259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001429772837383063,
+      "loss": 0.6315,
+      "step": 10260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001429673498026959,
+      "loss": 0.724,
+      "step": 10261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014295741534703732,
+      "loss": 0.5452,
+      "step": 10262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014294748037145083,
+      "loss": 0.6594,
+      "step": 10263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014293754487605664,
+      "loss": 0.6562,
+      "step": 10264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014292760886097503,
+      "loss": 0.7095,
+      "step": 10265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014291767232632626,
+      "loss": 0.6648,
+      "step": 10266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001429077352722306,
+      "loss": 0.6941,
+      "step": 10267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014289779769880828,
+      "loss": 0.6807,
+      "step": 10268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014288785960617966,
+      "loss": 0.6006,
+      "step": 10269
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014287792099446493,
+      "loss": 0.6659,
+      "step": 10270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014286798186378445,
+      "loss": 0.6713,
+      "step": 10271
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001428580422142585,
+      "loss": 0.695,
+      "step": 10272
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014284810204600733,
+      "loss": 0.7382,
+      "step": 10273
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014283816135915133,
+      "loss": 0.6469,
+      "step": 10274
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014282822015381077,
+      "loss": 0.6867,
+      "step": 10275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014281827843010597,
+      "loss": 0.6883,
+      "step": 10276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001428083361881573,
+      "loss": 0.5845,
+      "step": 10277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014279839342808503,
+      "loss": 0.6867,
+      "step": 10278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014278845015000956,
+      "loss": 0.6259,
+      "step": 10279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014277850635405117,
+      "loss": 0.5727,
+      "step": 10280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014276856204033028,
+      "loss": 0.6999,
+      "step": 10281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014275861720896725,
+      "loss": 0.6081,
+      "step": 10282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014274867186008235,
+      "loss": 0.6685,
+      "step": 10283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001427387259937961,
+      "loss": 0.6839,
+      "step": 10284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014272877961022876,
+      "loss": 0.7359,
+      "step": 10285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014271883270950073,
+      "loss": 0.6383,
+      "step": 10286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014270888529173244,
+      "loss": 0.6642,
+      "step": 10287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014269893735704425,
+      "loss": 0.6724,
+      "step": 10288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014268898890555662,
+      "loss": 0.6289,
+      "step": 10289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014267903993738992,
+      "loss": 0.7313,
+      "step": 10290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014266909045266454,
+      "loss": 0.6744,
+      "step": 10291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014265914045150097,
+      "loss": 0.6451,
+      "step": 10292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014264918993401956,
+      "loss": 0.6981,
+      "step": 10293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014263923890034082,
+      "loss": 0.7257,
+      "step": 10294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001426292873505851,
+      "loss": 0.6393,
+      "step": 10295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014261933528487294,
+      "loss": 0.6896,
+      "step": 10296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014260938270332476,
+      "loss": 0.7898,
+      "step": 10297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000142599429606061,
+      "loss": 0.6488,
+      "step": 10298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014258947599320214,
+      "loss": 0.7036,
+      "step": 10299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014257952186486864,
+      "loss": 0.7424,
+      "step": 10300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000142569567221181,
+      "loss": 0.6084,
+      "step": 10301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014255961206225968,
+      "loss": 0.6669,
+      "step": 10302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001425496563882252,
+      "loss": 0.6212,
+      "step": 10303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014253970019919803,
+      "loss": 0.648,
+      "step": 10304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001425297434952987,
+      "loss": 0.7031,
+      "step": 10305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014251978627664768,
+      "loss": 0.7148,
+      "step": 10306
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001425098285433655,
+      "loss": 0.5999,
+      "step": 10307
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001424998702955727,
+      "loss": 0.6985,
+      "step": 10308
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014248991153338985,
+      "loss": 0.6811,
+      "step": 10309
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014247995225693738,
+      "loss": 0.7722,
+      "step": 10310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001424699924663359,
+      "loss": 0.7148,
+      "step": 10311
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001424600321617059,
+      "loss": 0.7351,
+      "step": 10312
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014245007134316802,
+      "loss": 0.7626,
+      "step": 10313
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014244011001084275,
+      "loss": 0.7309,
+      "step": 10314
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001424301481648507,
+      "loss": 0.7305,
+      "step": 10315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001424201858053124,
+      "loss": 0.6737,
+      "step": 10316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014241022293234843,
+      "loss": 0.5923,
+      "step": 10317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014240025954607942,
+      "loss": 0.6275,
+      "step": 10318
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001423902956466259,
+      "loss": 0.7041,
+      "step": 10319
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014238033123410853,
+      "loss": 0.6638,
+      "step": 10320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014237036630864784,
+      "loss": 0.5912,
+      "step": 10321
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014236040087036453,
+      "loss": 0.6137,
+      "step": 10322
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014235043491937915,
+      "loss": 0.6412,
+      "step": 10323
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014234046845581228,
+      "loss": 0.6623,
+      "step": 10324
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014233050147978467,
+      "loss": 0.6477,
+      "step": 10325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014232053399141682,
+      "loss": 0.7267,
+      "step": 10326
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001423105659908295,
+      "loss": 0.6456,
+      "step": 10327
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014230059747814328,
+      "loss": 0.6635,
+      "step": 10328
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014229062845347879,
+      "loss": 0.7643,
+      "step": 10329
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014228065891695675,
+      "loss": 0.6536,
+      "step": 10330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014227068886869777,
+      "loss": 0.6589,
+      "step": 10331
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014226071830882256,
+      "loss": 0.6833,
+      "step": 10332
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001422507472374518,
+      "loss": 0.614,
+      "step": 10333
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014224077565470612,
+      "loss": 0.574,
+      "step": 10334
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001422308035607063,
+      "loss": 0.5725,
+      "step": 10335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014222083095557296,
+      "loss": 0.7411,
+      "step": 10336
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001422108578394268,
+      "loss": 0.6386,
+      "step": 10337
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001422008842123886,
+      "loss": 0.7668,
+      "step": 10338
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014219091007457898,
+      "loss": 0.6037,
+      "step": 10339
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014218093542611872,
+      "loss": 0.6847,
+      "step": 10340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014217096026712854,
+      "loss": 0.6333,
+      "step": 10341
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014216098459772917,
+      "loss": 0.619,
+      "step": 10342
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014215100841804133,
+      "loss": 0.7566,
+      "step": 10343
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001421410317281858,
+      "loss": 0.706,
+      "step": 10344
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001421310545282833,
+      "loss": 0.6917,
+      "step": 10345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001421210768184546,
+      "loss": 0.6429,
+      "step": 10346
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014211109859882046,
+      "loss": 0.6911,
+      "step": 10347
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014210111986950164,
+      "loss": 0.6674,
+      "step": 10348
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014209114063061894,
+      "loss": 0.6679,
+      "step": 10349
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014208116088229315,
+      "loss": 0.6049,
+      "step": 10350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000142071180624645,
+      "loss": 0.6207,
+      "step": 10351
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014206119985779535,
+      "loss": 0.7124,
+      "step": 10352
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014205121858186496,
+      "loss": 0.6605,
+      "step": 10353
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014204123679697467,
+      "loss": 0.7107,
+      "step": 10354
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014203125450324523,
+      "loss": 0.7486,
+      "step": 10355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014202127170079756,
+      "loss": 0.7435,
+      "step": 10356
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014201128838975237,
+      "loss": 0.6639,
+      "step": 10357
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014200130457023058,
+      "loss": 0.6571,
+      "step": 10358
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000141991320242353,
+      "loss": 0.6362,
+      "step": 10359
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014198133540624047,
+      "loss": 0.7177,
+      "step": 10360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014197135006201381,
+      "loss": 0.6104,
+      "step": 10361
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001419613642097939,
+      "loss": 0.6078,
+      "step": 10362
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014195137784970164,
+      "loss": 0.7583,
+      "step": 10363
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014194139098185784,
+      "loss": 0.6072,
+      "step": 10364
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014193140360638343,
+      "loss": 0.6552,
+      "step": 10365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014192141572339924,
+      "loss": 0.6267,
+      "step": 10366
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014191142733302616,
+      "loss": 0.6481,
+      "step": 10367
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014190143843538513,
+      "loss": 0.7046,
+      "step": 10368
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014189144903059702,
+      "loss": 0.6868,
+      "step": 10369
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001418814591187827,
+      "loss": 0.5921,
+      "step": 10370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014187146870006315,
+      "loss": 0.776,
+      "step": 10371
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014186147777455922,
+      "loss": 0.753,
+      "step": 10372
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001418514863423919,
+      "loss": 0.6278,
+      "step": 10373
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014184149440368206,
+      "loss": 0.6616,
+      "step": 10374
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014183150195855066,
+      "loss": 0.5971,
+      "step": 10375
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014182150900711866,
+      "loss": 0.738,
+      "step": 10376
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000141811515549507,
+      "loss": 0.7135,
+      "step": 10377
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001418015215858366,
+      "loss": 0.6449,
+      "step": 10378
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014179152711622846,
+      "loss": 0.6896,
+      "step": 10379
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001417815321408035,
+      "loss": 0.7141,
+      "step": 10380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014177153665968277,
+      "loss": 0.6708,
+      "step": 10381
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014176154067298716,
+      "loss": 0.7056,
+      "step": 10382
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014175154418083773,
+      "loss": 0.6339,
+      "step": 10383
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014174154718335545,
+      "loss": 0.7029,
+      "step": 10384
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014173154968066126,
+      "loss": 0.6692,
+      "step": 10385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014172155167287625,
+      "loss": 0.6182,
+      "step": 10386
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001417115531601214,
+      "loss": 0.6473,
+      "step": 10387
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014170155414251767,
+      "loss": 0.6792,
+      "step": 10388
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014169155462018614,
+      "loss": 0.7396,
+      "step": 10389
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014168155459324782,
+      "loss": 0.6213,
+      "step": 10390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014167155406182374,
+      "loss": 0.7027,
+      "step": 10391
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014166155302603497,
+      "loss": 0.6883,
+      "step": 10392
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001416515514860025,
+      "loss": 0.6963,
+      "step": 10393
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014164154944184746,
+      "loss": 0.6115,
+      "step": 10394
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014163154689369082,
+      "loss": 0.6915,
+      "step": 10395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014162154384165368,
+      "loss": 0.7394,
+      "step": 10396
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014161154028585716,
+      "loss": 0.6681,
+      "step": 10397
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014160153622642225,
+      "loss": 0.6392,
+      "step": 10398
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001415915316634701,
+      "loss": 0.8041,
+      "step": 10399
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014158152659712176,
+      "loss": 0.6984,
+      "step": 10400
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 0.6635603308677673,
+      "eval_runtime": 850.2066,
+      "eval_samples_per_second": 28.299,
+      "eval_steps_per_second": 14.15,
+      "step": 10400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014157152102749836,
+      "loss": 0.5837,
+      "step": 10401
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014156151495472096,
+      "loss": 0.6337,
+      "step": 10402
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014155150837891068,
+      "loss": 0.7091,
+      "step": 10403
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014154150130018866,
+      "loss": 0.6451,
+      "step": 10404
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014153149371867597,
+      "loss": 0.5755,
+      "step": 10405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014152148563449378,
+      "loss": 0.6791,
+      "step": 10406
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014151147704776322,
+      "loss": 0.6107,
+      "step": 10407
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014150146795860536,
+      "loss": 0.6137,
+      "step": 10408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014149145836714143,
+      "loss": 0.6205,
+      "step": 10409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014148144827349257,
+      "loss": 0.7692,
+      "step": 10410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014147143767777987,
+      "loss": 0.7522,
+      "step": 10411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014146142658012456,
+      "loss": 0.6281,
+      "step": 10412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014145141498064773,
+      "loss": 0.7569,
+      "step": 10413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014144140287947066,
+      "loss": 0.6681,
+      "step": 10414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014143139027671442,
+      "loss": 0.6543,
+      "step": 10415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001414213771725003,
+      "loss": 0.7398,
+      "step": 10416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014141136356694942,
+      "loss": 0.6389,
+      "step": 10417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000141401349460183,
+      "loss": 0.5827,
+      "step": 10418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001413913348523222,
+      "loss": 0.6517,
+      "step": 10419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014138131974348833,
+      "loss": 0.5731,
+      "step": 10420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001413713041338025,
+      "loss": 0.693,
+      "step": 10421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014136128802338602,
+      "loss": 0.7598,
+      "step": 10422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001413512714123601,
+      "loss": 0.7195,
+      "step": 10423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014134125430084588,
+      "loss": 0.5762,
+      "step": 10424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014133123668896473,
+      "loss": 0.639,
+      "step": 10425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014132121857683783,
+      "loss": 0.643,
+      "step": 10426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014131119996458645,
+      "loss": 0.6576,
+      "step": 10427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014130118085233182,
+      "loss": 0.5875,
+      "step": 10428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014129116124019523,
+      "loss": 0.624,
+      "step": 10429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014128114112829798,
+      "loss": 0.6194,
+      "step": 10430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014127112051676125,
+      "loss": 0.6184,
+      "step": 10431
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014126109940570643,
+      "loss": 0.6833,
+      "step": 10432
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014125107779525476,
+      "loss": 0.7258,
+      "step": 10433
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014124105568552754,
+      "loss": 0.6466,
+      "step": 10434
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014123103307664607,
+      "loss": 0.7442,
+      "step": 10435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014122100996873167,
+      "loss": 0.6979,
+      "step": 10436
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014121098636190563,
+      "loss": 0.6285,
+      "step": 10437
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014120096225628926,
+      "loss": 0.6171,
+      "step": 10438
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014119093765200394,
+      "loss": 0.731,
+      "step": 10439
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014118091254917094,
+      "loss": 0.7445,
+      "step": 10440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014117088694791163,
+      "loss": 0.6908,
+      "step": 10441
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014116086084834737,
+      "loss": 0.6255,
+      "step": 10442
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014115083425059947,
+      "loss": 0.7598,
+      "step": 10443
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001411408071547893,
+      "loss": 0.6679,
+      "step": 10444
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001411307795610382,
+      "loss": 0.7466,
+      "step": 10445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014112075146946758,
+      "loss": 0.7547,
+      "step": 10446
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001411107228801988,
+      "loss": 0.6546,
+      "step": 10447
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001411006937933532,
+      "loss": 0.6265,
+      "step": 10448
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014109066420905223,
+      "loss": 0.5531,
+      "step": 10449
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014108063412741727,
+      "loss": 0.6329,
+      "step": 10450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014107060354856966,
+      "loss": 0.6176,
+      "step": 10451
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014106057247263086,
+      "loss": 0.7714,
+      "step": 10452
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014105054089972224,
+      "loss": 0.6158,
+      "step": 10453
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014104050882996524,
+      "loss": 0.661,
+      "step": 10454
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014103047626348129,
+      "loss": 0.7138,
+      "step": 10455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001410204432003918,
+      "loss": 0.6241,
+      "step": 10456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001410104096408182,
+      "loss": 0.6243,
+      "step": 10457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014100037558488195,
+      "loss": 0.6904,
+      "step": 10458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014099034103270448,
+      "loss": 0.6566,
+      "step": 10459
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014098030598440724,
+      "loss": 0.6502,
+      "step": 10460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014097027044011168,
+      "loss": 0.6495,
+      "step": 10461
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001409602343999393,
+      "loss": 0.651,
+      "step": 10462
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014095019786401154,
+      "loss": 0.6304,
+      "step": 10463
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014094016083244983,
+      "loss": 0.6358,
+      "step": 10464
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014093012330537575,
+      "loss": 0.7215,
+      "step": 10465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014092008528291075,
+      "loss": 0.6758,
+      "step": 10466
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014091004676517626,
+      "loss": 0.6501,
+      "step": 10467
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001409000077522939,
+      "loss": 0.5798,
+      "step": 10468
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014088996824438509,
+      "loss": 0.6001,
+      "step": 10469
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001408799282415713,
+      "loss": 0.7123,
+      "step": 10470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001408698877439742,
+      "loss": 0.6093,
+      "step": 10471
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014085984675171515,
+      "loss": 0.6789,
+      "step": 10472
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014084980526491578,
+      "loss": 0.61,
+      "step": 10473
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001408397632836976,
+      "loss": 0.6528,
+      "step": 10474
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014082972080818213,
+      "loss": 0.6588,
+      "step": 10475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014081967783849094,
+      "loss": 0.7558,
+      "step": 10476
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014080963437474556,
+      "loss": 0.6373,
+      "step": 10477
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001407995904170676,
+      "loss": 0.6025,
+      "step": 10478
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001407895459655786,
+      "loss": 0.7021,
+      "step": 10479
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014077950102040008,
+      "loss": 0.5845,
+      "step": 10480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014076945558165368,
+      "loss": 0.8409,
+      "step": 10481
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014075940964946099,
+      "loss": 0.642,
+      "step": 10482
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014074936322394353,
+      "loss": 0.6826,
+      "step": 10483
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014073931630522298,
+      "loss": 0.617,
+      "step": 10484
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014072926889342085,
+      "loss": 0.8105,
+      "step": 10485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014071922098865883,
+      "loss": 0.6284,
+      "step": 10486
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014070917259105853,
+      "loss": 0.6645,
+      "step": 10487
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001406991237007415,
+      "loss": 0.7117,
+      "step": 10488
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014068907431782942,
+      "loss": 0.7766,
+      "step": 10489
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014067902444244386,
+      "loss": 0.5836,
+      "step": 10490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014066897407470657,
+      "loss": 0.6281,
+      "step": 10491
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001406589232147391,
+      "loss": 0.6532,
+      "step": 10492
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014064887186266315,
+      "loss": 0.669,
+      "step": 10493
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001406388200186003,
+      "loss": 0.6911,
+      "step": 10494
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014062876768267227,
+      "loss": 0.6142,
+      "step": 10495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014061871485500076,
+      "loss": 0.7009,
+      "step": 10496
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014060866153570736,
+      "loss": 0.6035,
+      "step": 10497
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014059860772491377,
+      "loss": 0.7028,
+      "step": 10498
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014058855342274175,
+      "loss": 0.6686,
+      "step": 10499
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001405784986293129,
+      "loss": 0.6229,
+      "step": 10500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014056844334474895,
+      "loss": 0.611,
+      "step": 10501
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001405583875691716,
+      "loss": 0.6461,
+      "step": 10502
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014054833130270258,
+      "loss": 0.6485,
+      "step": 10503
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014053827454546357,
+      "loss": 0.7223,
+      "step": 10504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014052821729757632,
+      "loss": 0.751,
+      "step": 10505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014051815955916253,
+      "loss": 0.6913,
+      "step": 10506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014050810133034393,
+      "loss": 0.7054,
+      "step": 10507
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014049804261124228,
+      "loss": 0.6676,
+      "step": 10508
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014048798340197933,
+      "loss": 0.6665,
+      "step": 10509
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001404779237026768,
+      "loss": 0.7871,
+      "step": 10510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014046786351345648,
+      "loss": 0.7229,
+      "step": 10511
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014045780283444012,
+      "loss": 0.6547,
+      "step": 10512
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014044774166574946,
+      "loss": 0.6599,
+      "step": 10513
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001404376800075063,
+      "loss": 0.5673,
+      "step": 10514
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001404276178598324,
+      "loss": 0.5519,
+      "step": 10515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001404175552228496,
+      "loss": 0.6474,
+      "step": 10516
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014040749209667962,
+      "loss": 0.6395,
+      "step": 10517
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001403974284814443,
+      "loss": 0.665,
+      "step": 10518
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014038736437726543,
+      "loss": 0.7206,
+      "step": 10519
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014037729978426484,
+      "loss": 0.7856,
+      "step": 10520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001403672347025643,
+      "loss": 0.5989,
+      "step": 10521
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014035716913228568,
+      "loss": 0.5498,
+      "step": 10522
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014034710307355076,
+      "loss": 0.668,
+      "step": 10523
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001403370365264814,
+      "loss": 0.6136,
+      "step": 10524
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014032696949119945,
+      "loss": 0.7532,
+      "step": 10525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014031690196782674,
+      "loss": 0.6583,
+      "step": 10526
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014030683395648512,
+      "loss": 0.5832,
+      "step": 10527
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014029676545729644,
+      "loss": 0.7194,
+      "step": 10528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001402866964703826,
+      "loss": 0.6824,
+      "step": 10529
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001402766269958654,
+      "loss": 0.7969,
+      "step": 10530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014026655703386676,
+      "loss": 0.6305,
+      "step": 10531
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001402564865845086,
+      "loss": 0.5681,
+      "step": 10532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001402464156479127,
+      "loss": 0.7382,
+      "step": 10533
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014023634422420105,
+      "loss": 0.6407,
+      "step": 10534
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001402262723134955,
+      "loss": 0.5764,
+      "step": 10535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014021619991591794,
+      "loss": 0.6959,
+      "step": 10536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014020612703159033,
+      "loss": 0.7174,
+      "step": 10537
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014019605366063454,
+      "loss": 0.6825,
+      "step": 10538
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014018597980317252,
+      "loss": 0.5834,
+      "step": 10539
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001401759054593262,
+      "loss": 0.629,
+      "step": 10540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014016583062921746,
+      "loss": 0.5603,
+      "step": 10541
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014015575531296833,
+      "loss": 0.7312,
+      "step": 10542
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014014567951070066,
+      "loss": 0.6409,
+      "step": 10543
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014013560322253646,
+      "loss": 0.582,
+      "step": 10544
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014012552644859766,
+      "loss": 0.7249,
+      "step": 10545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014011544918900623,
+      "loss": 0.6212,
+      "step": 10546
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014010537144388416,
+      "loss": 0.6275,
+      "step": 10547
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001400952932133534,
+      "loss": 0.7215,
+      "step": 10548
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014008521449753594,
+      "loss": 0.7019,
+      "step": 10549
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014007513529655378,
+      "loss": 0.6549,
+      "step": 10550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014006505561052889,
+      "loss": 0.6741,
+      "step": 10551
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014005497543958324,
+      "loss": 0.6738,
+      "step": 10552
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014004489478383888,
+      "loss": 0.671,
+      "step": 10553
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014003481364341779,
+      "loss": 0.8451,
+      "step": 10554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014002473201844205,
+      "loss": 0.8018,
+      "step": 10555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014001464990903362,
+      "loss": 0.7343,
+      "step": 10556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001400045673153145,
+      "loss": 0.7271,
+      "step": 10557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013999448423740677,
+      "loss": 0.7159,
+      "step": 10558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013998440067543246,
+      "loss": 0.7361,
+      "step": 10559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013997431662951366,
+      "loss": 0.6325,
+      "step": 10560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013996423209977233,
+      "loss": 0.5891,
+      "step": 10561
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001399541470863306,
+      "loss": 0.6759,
+      "step": 10562
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013994406158931053,
+      "loss": 0.6338,
+      "step": 10563
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001399339756088341,
+      "loss": 0.59,
+      "step": 10564
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013992388914502347,
+      "loss": 0.6773,
+      "step": 10565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001399138021980007,
+      "loss": 0.7049,
+      "step": 10566
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001399037147678879,
+      "loss": 0.6081,
+      "step": 10567
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013989362685480709,
+      "loss": 0.6145,
+      "step": 10568
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013988353845888042,
+      "loss": 0.674,
+      "step": 10569
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013987344958023,
+      "loss": 0.5691,
+      "step": 10570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013986336021897794,
+      "loss": 0.6894,
+      "step": 10571
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001398532703752463,
+      "loss": 0.6649,
+      "step": 10572
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013984318004915726,
+      "loss": 0.7298,
+      "step": 10573
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013983308924083292,
+      "loss": 0.476,
+      "step": 10574
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013982299795039545,
+      "loss": 0.6634,
+      "step": 10575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013981290617796693,
+      "loss": 0.6301,
+      "step": 10576
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013980281392366953,
+      "loss": 0.8111,
+      "step": 10577
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001397927211876254,
+      "loss": 0.7083,
+      "step": 10578
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013978262796995669,
+      "loss": 0.6054,
+      "step": 10579
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013977253427078557,
+      "loss": 0.5536,
+      "step": 10580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013976244009023423,
+      "loss": 0.7607,
+      "step": 10581
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013975234542842478,
+      "loss": 0.7953,
+      "step": 10582
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013974225028547948,
+      "loss": 0.6751,
+      "step": 10583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013973215466152043,
+      "loss": 0.5998,
+      "step": 10584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013972205855666988,
+      "loss": 0.6469,
+      "step": 10585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013971196197105004,
+      "loss": 0.6536,
+      "step": 10586
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013970186490478304,
+      "loss": 0.5962,
+      "step": 10587
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013969176735799117,
+      "loss": 0.7225,
+      "step": 10588
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013968166933079655,
+      "loss": 0.6759,
+      "step": 10589
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013967157082332148,
+      "loss": 0.6594,
+      "step": 10590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013966147183568819,
+      "loss": 0.7251,
+      "step": 10591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013965137236801885,
+      "loss": 0.7509,
+      "step": 10592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013964127242043574,
+      "loss": 0.6564,
+      "step": 10593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001396311719930611,
+      "loss": 0.6722,
+      "step": 10594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013962107108601713,
+      "loss": 0.6514,
+      "step": 10595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013961096969942617,
+      "loss": 0.6438,
+      "step": 10596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001396008678334104,
+      "loss": 0.6992,
+      "step": 10597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013959076548809217,
+      "loss": 0.7596,
+      "step": 10598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013958066266359366,
+      "loss": 0.7018,
+      "step": 10599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001395705593600372,
+      "loss": 0.5621,
+      "step": 10600
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 0.6578021049499512,
+      "eval_runtime": 843.9845,
+      "eval_samples_per_second": 28.508,
+      "eval_steps_per_second": 14.254,
+      "step": 10600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013956045557754506,
+      "loss": 0.6549,
+      "step": 10601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013955035131623956,
+      "loss": 0.581,
+      "step": 10602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013954024657624294,
+      "loss": 0.7946,
+      "step": 10603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013953014135767756,
+      "loss": 0.6834,
+      "step": 10604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013952003566066568,
+      "loss": 0.6753,
+      "step": 10605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013950992948532964,
+      "loss": 0.6835,
+      "step": 10606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013949982283179176,
+      "loss": 0.6241,
+      "step": 10607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001394897157001743,
+      "loss": 0.6383,
+      "step": 10608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001394796080905997,
+      "loss": 0.7453,
+      "step": 10609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013946950000319022,
+      "loss": 0.717,
+      "step": 10610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013945939143806822,
+      "loss": 0.6875,
+      "step": 10611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013944928239535603,
+      "loss": 0.6341,
+      "step": 10612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013943917287517605,
+      "loss": 0.6492,
+      "step": 10613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001394290628776506,
+      "loss": 0.586,
+      "step": 10614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013941895240290203,
+      "loss": 0.5919,
+      "step": 10615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013940884145105276,
+      "loss": 0.593,
+      "step": 10616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013939873002222515,
+      "loss": 0.6594,
+      "step": 10617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013938861811654154,
+      "loss": 0.6819,
+      "step": 10618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001393785057341244,
+      "loss": 0.7073,
+      "step": 10619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013936839287509605,
+      "loss": 0.7105,
+      "step": 10620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013935827953957887,
+      "loss": 0.5861,
+      "step": 10621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013934816572769537,
+      "loss": 0.6653,
+      "step": 10622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013933805143956786,
+      "loss": 0.5596,
+      "step": 10623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001393279366753188,
+      "loss": 0.6198,
+      "step": 10624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013931782143507067,
+      "loss": 0.7456,
+      "step": 10625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013930770571894576,
+      "loss": 0.7637,
+      "step": 10626
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013929758952706663,
+      "loss": 0.6337,
+      "step": 10627
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013928747285955564,
+      "loss": 0.6896,
+      "step": 10628
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001392773557165353,
+      "loss": 0.6926,
+      "step": 10629
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013926723809812799,
+      "loss": 0.6438,
+      "step": 10630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001392571200044562,
+      "loss": 0.6101,
+      "step": 10631
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013924700143564243,
+      "loss": 0.6755,
+      "step": 10632
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013923688239180908,
+      "loss": 0.7177,
+      "step": 10633
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001392267628730787,
+      "loss": 0.6807,
+      "step": 10634
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001392166428795737,
+      "loss": 0.6384,
+      "step": 10635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001392065224114166,
+      "loss": 0.6589,
+      "step": 10636
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013919640146872988,
+      "loss": 0.625,
+      "step": 10637
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013918628005163603,
+      "loss": 0.6727,
+      "step": 10638
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001391761581602576,
+      "loss": 0.7117,
+      "step": 10639
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013916603579471705,
+      "loss": 0.725,
+      "step": 10640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013915591295513688,
+      "loss": 0.6119,
+      "step": 10641
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001391457896416397,
+      "loss": 0.6674,
+      "step": 10642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001391356658543479,
+      "loss": 0.7437,
+      "step": 10643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013912554159338413,
+      "loss": 0.715,
+      "step": 10644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001391154168588709,
+      "loss": 0.7019,
+      "step": 10645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013910529165093072,
+      "loss": 0.5532,
+      "step": 10646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013909516596968615,
+      "loss": 0.6841,
+      "step": 10647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013908503981525973,
+      "loss": 0.7936,
+      "step": 10648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001390749131877741,
+      "loss": 0.7252,
+      "step": 10649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013906478608735173,
+      "loss": 0.6412,
+      "step": 10650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001390546585141152,
+      "loss": 0.5916,
+      "step": 10651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013904453046818715,
+      "loss": 0.6851,
+      "step": 10652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013903440194969012,
+      "loss": 0.6893,
+      "step": 10653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013902427295874669,
+      "loss": 0.5087,
+      "step": 10654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001390141434954795,
+      "loss": 0.7776,
+      "step": 10655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001390040135600111,
+      "loss": 0.6081,
+      "step": 10656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013899388315246413,
+      "loss": 0.6125,
+      "step": 10657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001389837522729612,
+      "loss": 0.6991,
+      "step": 10658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013897362092162488,
+      "loss": 0.7126,
+      "step": 10659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013896348909857789,
+      "loss": 0.6596,
+      "step": 10660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013895335680394274,
+      "loss": 0.6631,
+      "step": 10661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013894322403784217,
+      "loss": 0.6242,
+      "step": 10662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013893309080039877,
+      "loss": 0.6121,
+      "step": 10663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013892295709173514,
+      "loss": 0.663,
+      "step": 10664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013891282291197404,
+      "loss": 0.6596,
+      "step": 10665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013890268826123803,
+      "loss": 0.8141,
+      "step": 10666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001388925531396498,
+      "loss": 0.6635,
+      "step": 10667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013888241754733208,
+      "loss": 0.5234,
+      "step": 10668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013887228148440746,
+      "loss": 0.7124,
+      "step": 10669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013886214495099867,
+      "loss": 0.702,
+      "step": 10670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001388520079472284,
+      "loss": 0.6649,
+      "step": 10671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013884187047321928,
+      "loss": 0.7092,
+      "step": 10672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013883173252909406,
+      "loss": 0.6368,
+      "step": 10673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013882159411497544,
+      "loss": 0.6431,
+      "step": 10674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013881145523098612,
+      "loss": 0.6531,
+      "step": 10675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013880131587724884,
+      "loss": 0.6409,
+      "step": 10676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013879117605388624,
+      "loss": 0.6691,
+      "step": 10677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013878103576102116,
+      "loss": 0.6781,
+      "step": 10678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013877089499877624,
+      "loss": 0.7155,
+      "step": 10679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013876075376727427,
+      "loss": 0.7081,
+      "step": 10680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013875061206663797,
+      "loss": 0.6641,
+      "step": 10681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013874046989699004,
+      "loss": 0.6948,
+      "step": 10682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013873032725845335,
+      "loss": 0.6196,
+      "step": 10683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013872018415115055,
+      "loss": 0.7594,
+      "step": 10684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001387100405752045,
+      "loss": 0.6382,
+      "step": 10685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013869989653073788,
+      "loss": 0.761,
+      "step": 10686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001386897520178735,
+      "loss": 0.6632,
+      "step": 10687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001386796070367342,
+      "loss": 0.6897,
+      "step": 10688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001386694615874427,
+      "loss": 0.5915,
+      "step": 10689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013865931567012178,
+      "loss": 0.552,
+      "step": 10690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001386491692848943,
+      "loss": 0.6533,
+      "step": 10691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013863902243188304,
+      "loss": 0.7614,
+      "step": 10692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013862887511121075,
+      "loss": 0.7054,
+      "step": 10693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013861872732300038,
+      "loss": 0.6741,
+      "step": 10694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013860857906737464,
+      "loss": 0.7068,
+      "step": 10695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013859843034445644,
+      "loss": 0.6642,
+      "step": 10696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013858828115436849,
+      "loss": 0.6156,
+      "step": 10697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013857813149723374,
+      "loss": 0.6671,
+      "step": 10698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013856798137317503,
+      "loss": 0.6524,
+      "step": 10699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013855783078231513,
+      "loss": 0.6618,
+      "step": 10700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000138547679724777,
+      "loss": 0.5977,
+      "step": 10701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013853752820068338,
+      "loss": 0.7023,
+      "step": 10702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013852737621015726,
+      "loss": 0.6678,
+      "step": 10703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013851722375332146,
+      "loss": 0.704,
+      "step": 10704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001385070708302988,
+      "loss": 0.5976,
+      "step": 10705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001384969174412123,
+      "loss": 0.6237,
+      "step": 10706
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001384867635861847,
+      "loss": 0.7033,
+      "step": 10707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013847660926533903,
+      "loss": 0.6126,
+      "step": 10708
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013846645447879809,
+      "loss": 0.7251,
+      "step": 10709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013845629922668482,
+      "loss": 0.6179,
+      "step": 10710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013844614350912217,
+      "loss": 0.6076,
+      "step": 10711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000138435987326233,
+      "loss": 0.5565,
+      "step": 10712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013842583067814025,
+      "loss": 0.6944,
+      "step": 10713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001384156735649669,
+      "loss": 0.6738,
+      "step": 10714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001384055159868358,
+      "loss": 0.6772,
+      "step": 10715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013839535794386994,
+      "loss": 0.6315,
+      "step": 10716
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013838519943619226,
+      "loss": 0.7298,
+      "step": 10717
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013837504046392574,
+      "loss": 0.6046,
+      "step": 10718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001383648810271933,
+      "loss": 0.6918,
+      "step": 10719
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001383547211261179,
+      "loss": 0.6589,
+      "step": 10720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001383445607608225,
+      "loss": 0.5979,
+      "step": 10721
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013833439993143014,
+      "loss": 0.5757,
+      "step": 10722
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001383242386380637,
+      "loss": 0.7073,
+      "step": 10723
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013831407688084628,
+      "loss": 0.7813,
+      "step": 10724
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001383039146599008,
+      "loss": 0.6039,
+      "step": 10725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013829375197535023,
+      "loss": 0.6376,
+      "step": 10726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013828358882731764,
+      "loss": 0.6685,
+      "step": 10727
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013827342521592598,
+      "loss": 0.6277,
+      "step": 10728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013826326114129835,
+      "loss": 0.5895,
+      "step": 10729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013825309660355767,
+      "loss": 0.6916,
+      "step": 10730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000138242931602827,
+      "loss": 0.5499,
+      "step": 10731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001382327661392294,
+      "loss": 0.6165,
+      "step": 10732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013822260021288787,
+      "loss": 0.6807,
+      "step": 10733
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013821243382392548,
+      "loss": 0.6667,
+      "step": 10734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013820226697246524,
+      "loss": 0.7005,
+      "step": 10735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013819209965863023,
+      "loss": 0.6124,
+      "step": 10736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001381819318825435,
+      "loss": 0.5554,
+      "step": 10737
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013817176364432814,
+      "loss": 0.6546,
+      "step": 10738
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013816159494410717,
+      "loss": 0.7382,
+      "step": 10739
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013815142578200372,
+      "loss": 0.6735,
+      "step": 10740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001381412561581408,
+      "loss": 0.7037,
+      "step": 10741
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013813108607264155,
+      "loss": 0.6938,
+      "step": 10742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013812091552562906,
+      "loss": 0.6935,
+      "step": 10743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013811074451722642,
+      "loss": 0.7539,
+      "step": 10744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013810057304755675,
+      "loss": 0.7056,
+      "step": 10745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013809040111674312,
+      "loss": 0.6994,
+      "step": 10746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013808022872490868,
+      "loss": 0.6335,
+      "step": 10747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001380700558721765,
+      "loss": 0.6717,
+      "step": 10748
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013805988255866976,
+      "loss": 0.7022,
+      "step": 10749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013804970878451158,
+      "loss": 0.6511,
+      "step": 10750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013803953454982504,
+      "loss": 0.7224,
+      "step": 10751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001380293598547334,
+      "loss": 0.6319,
+      "step": 10752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013801918469935967,
+      "loss": 0.6813,
+      "step": 10753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013800900908382711,
+      "loss": 0.6875,
+      "step": 10754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013799883300825885,
+      "loss": 0.6607,
+      "step": 10755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013798865647277801,
+      "loss": 0.4941,
+      "step": 10756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013797847947750783,
+      "loss": 0.6494,
+      "step": 10757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001379683020225714,
+      "loss": 0.6085,
+      "step": 10758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013795812410809198,
+      "loss": 0.6874,
+      "step": 10759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013794794573419273,
+      "loss": 0.6396,
+      "step": 10760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001379377669009968,
+      "loss": 0.809,
+      "step": 10761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013792758760862748,
+      "loss": 0.6165,
+      "step": 10762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001379174078572079,
+      "loss": 0.6588,
+      "step": 10763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013790722764686132,
+      "loss": 0.757,
+      "step": 10764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013789704697771086,
+      "loss": 0.6914,
+      "step": 10765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013788686584987985,
+      "loss": 0.5749,
+      "step": 10766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013787668426349146,
+      "loss": 0.5882,
+      "step": 10767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001378665022186689,
+      "loss": 0.6946,
+      "step": 10768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001378563197155355,
+      "loss": 0.6365,
+      "step": 10769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013784613675421438,
+      "loss": 0.7456,
+      "step": 10770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013783595333482886,
+      "loss": 0.6699,
+      "step": 10771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001378257694575022,
+      "loss": 0.7494,
+      "step": 10772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001378155851223576,
+      "loss": 0.7356,
+      "step": 10773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001378054003295184,
+      "loss": 0.693,
+      "step": 10774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013779521507910784,
+      "loss": 0.4936,
+      "step": 10775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013778502937124916,
+      "loss": 0.7314,
+      "step": 10776
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001377748432060657,
+      "loss": 0.6423,
+      "step": 10777
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001377646565836807,
+      "loss": 0.6756,
+      "step": 10778
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013775446950421746,
+      "loss": 0.6578,
+      "step": 10779
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001377442819677993,
+      "loss": 0.7308,
+      "step": 10780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013773409397454948,
+      "loss": 0.7069,
+      "step": 10781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013772390552459137,
+      "loss": 0.63,
+      "step": 10782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013771371661804827,
+      "loss": 0.6438,
+      "step": 10783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013770352725504345,
+      "loss": 0.6319,
+      "step": 10784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013769333743570027,
+      "loss": 0.6758,
+      "step": 10785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013768314716014205,
+      "loss": 0.6904,
+      "step": 10786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013767295642849216,
+      "loss": 0.6036,
+      "step": 10787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001376627652408739,
+      "loss": 0.5758,
+      "step": 10788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013765257359741063,
+      "loss": 0.6767,
+      "step": 10789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013764238149822572,
+      "loss": 0.5813,
+      "step": 10790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013763218894344252,
+      "loss": 0.5672,
+      "step": 10791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013762199593318435,
+      "loss": 0.6048,
+      "step": 10792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013761180246757467,
+      "loss": 0.6912,
+      "step": 10793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001376016085467368,
+      "loss": 0.6611,
+      "step": 10794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013759141417079407,
+      "loss": 0.5468,
+      "step": 10795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013758121933986998,
+      "loss": 0.7432,
+      "step": 10796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001375710240540878,
+      "loss": 0.6864,
+      "step": 10797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013756082831357106,
+      "loss": 0.662,
+      "step": 10798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013755063211844305,
+      "loss": 0.6628,
+      "step": 10799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001375404354688272,
+      "loss": 0.5342,
+      "step": 10800
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 0.6511815190315247,
+      "eval_runtime": 853.6038,
+      "eval_samples_per_second": 28.186,
+      "eval_steps_per_second": 14.093,
+      "step": 10800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013753023836484698,
+      "loss": 0.5743,
+      "step": 10801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013752004080662574,
+      "loss": 0.6017,
+      "step": 10802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013750984279428693,
+      "loss": 0.5459,
+      "step": 10803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013749964432795402,
+      "loss": 0.7155,
+      "step": 10804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001374894454077504,
+      "loss": 0.5783,
+      "step": 10805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001374792460337995,
+      "loss": 0.6034,
+      "step": 10806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013746904620622478,
+      "loss": 0.7507,
+      "step": 10807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013745884592514975,
+      "loss": 0.59,
+      "step": 10808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013744864519069777,
+      "loss": 0.6245,
+      "step": 10809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013743844400299237,
+      "loss": 0.6181,
+      "step": 10810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013742824236215702,
+      "loss": 0.6026,
+      "step": 10811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013741804026831512,
+      "loss": 0.8695,
+      "step": 10812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013740783772159024,
+      "loss": 0.561,
+      "step": 10813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013739763472210583,
+      "loss": 0.6328,
+      "step": 10814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013738743126998535,
+      "loss": 0.6862,
+      "step": 10815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013737722736535234,
+      "loss": 0.6893,
+      "step": 10816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013736702300833027,
+      "loss": 0.7435,
+      "step": 10817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001373568181990427,
+      "loss": 0.6686,
+      "step": 10818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013734661293761307,
+      "loss": 0.5861,
+      "step": 10819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013733640722416493,
+      "loss": 0.6506,
+      "step": 10820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013732620105882178,
+      "loss": 0.6896,
+      "step": 10821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013731599444170723,
+      "loss": 0.746,
+      "step": 10822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001373057873729447,
+      "loss": 0.7014,
+      "step": 10823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001372955798526578,
+      "loss": 0.7084,
+      "step": 10824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001372853718809701,
+      "loss": 0.5948,
+      "step": 10825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001372751634580051,
+      "loss": 0.6555,
+      "step": 10826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013726495458388633,
+      "loss": 0.5797,
+      "step": 10827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001372547452587374,
+      "loss": 0.6679,
+      "step": 10828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013724453548268187,
+      "loss": 0.6849,
+      "step": 10829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013723432525584333,
+      "loss": 0.5753,
+      "step": 10830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013722411457834532,
+      "loss": 0.6299,
+      "step": 10831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013721390345031143,
+      "loss": 0.6814,
+      "step": 10832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013720369187186527,
+      "loss": 0.5771,
+      "step": 10833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013719347984313043,
+      "loss": 0.6336,
+      "step": 10834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013718326736423048,
+      "loss": 0.6808,
+      "step": 10835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013717305443528908,
+      "loss": 0.646,
+      "step": 10836
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013716284105642977,
+      "loss": 0.5791,
+      "step": 10837
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001371526272277762,
+      "loss": 0.5944,
+      "step": 10838
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000137142412949452,
+      "loss": 0.6615,
+      "step": 10839
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013713219822158082,
+      "loss": 0.6157,
+      "step": 10840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013712198304428622,
+      "loss": 0.6615,
+      "step": 10841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013711176741769193,
+      "loss": 0.6259,
+      "step": 10842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001371015513419215,
+      "loss": 0.6377,
+      "step": 10843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013709133481709863,
+      "loss": 0.6841,
+      "step": 10844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013708111784334698,
+      "loss": 0.638,
+      "step": 10845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013707090042079022,
+      "loss": 0.6424,
+      "step": 10846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013706068254955195,
+      "loss": 0.6895,
+      "step": 10847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013705046422975587,
+      "loss": 0.6726,
+      "step": 10848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013704024546152573,
+      "loss": 0.6414,
+      "step": 10849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013703002624498506,
+      "loss": 0.6968,
+      "step": 10850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001370198065802577,
+      "loss": 0.6097,
+      "step": 10851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013700958646746728,
+      "loss": 0.6477,
+      "step": 10852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013699936590673745,
+      "loss": 0.6646,
+      "step": 10853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000136989144898192,
+      "loss": 0.719,
+      "step": 10854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001369789234419546,
+      "loss": 0.6055,
+      "step": 10855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013696870153814892,
+      "loss": 0.6715,
+      "step": 10856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013695847918689872,
+      "loss": 0.7469,
+      "step": 10857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013694825638832774,
+      "loss": 0.5763,
+      "step": 10858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013693803314255968,
+      "loss": 0.7482,
+      "step": 10859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013692780944971834,
+      "loss": 0.6657,
+      "step": 10860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013691758530992733,
+      "loss": 0.652,
+      "step": 10861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001369073607233105,
+      "loss": 0.6585,
+      "step": 10862
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001368971356899916,
+      "loss": 0.614,
+      "step": 10863
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013688691021009435,
+      "loss": 0.6724,
+      "step": 10864
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013687668428374254,
+      "loss": 0.64,
+      "step": 10865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001368664579110599,
+      "loss": 0.6112,
+      "step": 10866
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013685623109217025,
+      "loss": 0.5538,
+      "step": 10867
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001368460038271973,
+      "loss": 0.7234,
+      "step": 10868
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013683577611626492,
+      "loss": 0.5732,
+      "step": 10869
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013682554795949687,
+      "loss": 0.6226,
+      "step": 10870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001368153193570169,
+      "loss": 0.7298,
+      "step": 10871
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013680509030894887,
+      "loss": 0.6987,
+      "step": 10872
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013679486081541654,
+      "loss": 0.6565,
+      "step": 10873
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013678463087654375,
+      "loss": 0.7317,
+      "step": 10874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013677440049245432,
+      "loss": 0.6609,
+      "step": 10875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000136764169663272,
+      "loss": 0.8067,
+      "step": 10876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013675393838912073,
+      "loss": 0.5844,
+      "step": 10877
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013674370667012432,
+      "loss": 0.6765,
+      "step": 10878
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001367334745064065,
+      "loss": 0.6911,
+      "step": 10879
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013672324189809124,
+      "loss": 0.7735,
+      "step": 10880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001367130088453023,
+      "loss": 0.6432,
+      "step": 10881
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013670277534816357,
+      "loss": 0.6019,
+      "step": 10882
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013669254140679895,
+      "loss": 0.6713,
+      "step": 10883
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013668230702133224,
+      "loss": 0.7212,
+      "step": 10884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013667207219188735,
+      "loss": 0.6273,
+      "step": 10885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013666183691858814,
+      "loss": 0.7508,
+      "step": 10886
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013665160120155848,
+      "loss": 0.6594,
+      "step": 10887
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001366413650409223,
+      "loss": 0.6762,
+      "step": 10888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013663112843680342,
+      "loss": 0.6649,
+      "step": 10889
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001366208913893258,
+      "loss": 0.6901,
+      "step": 10890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013661065389861335,
+      "loss": 0.6636,
+      "step": 10891
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001366004159647899,
+      "loss": 0.7089,
+      "step": 10892
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013659017758797942,
+      "loss": 0.6974,
+      "step": 10893
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001365799387683058,
+      "loss": 0.7031,
+      "step": 10894
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013656969950589303,
+      "loss": 0.7077,
+      "step": 10895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000136559459800865,
+      "loss": 0.6231,
+      "step": 10896
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013654921965334558,
+      "loss": 0.6192,
+      "step": 10897
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013653897906345882,
+      "loss": 0.5696,
+      "step": 10898
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001365287380313286,
+      "loss": 0.6497,
+      "step": 10899
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013651849655707887,
+      "loss": 0.6454,
+      "step": 10900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001365082546408336,
+      "loss": 0.5663,
+      "step": 10901
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013649801228271673,
+      "loss": 0.6258,
+      "step": 10902
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013648776948285232,
+      "loss": 0.5775,
+      "step": 10903
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013647752624136418,
+      "loss": 0.7235,
+      "step": 10904
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013646728255837642,
+      "loss": 0.6446,
+      "step": 10905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013645703843401298,
+      "loss": 0.7375,
+      "step": 10906
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013644679386839786,
+      "loss": 0.5939,
+      "step": 10907
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013643654886165502,
+      "loss": 0.6708,
+      "step": 10908
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013642630341390848,
+      "loss": 0.6056,
+      "step": 10909
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013641605752528224,
+      "loss": 0.8037,
+      "step": 10910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013640581119590033,
+      "loss": 0.9612,
+      "step": 10911
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001363955644258867,
+      "loss": 0.6012,
+      "step": 10912
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013638531721536548,
+      "loss": 0.5524,
+      "step": 10913
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001363750695644606,
+      "loss": 0.6157,
+      "step": 10914
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013636482147329614,
+      "loss": 0.5995,
+      "step": 10915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001363545729419961,
+      "loss": 0.6567,
+      "step": 10916
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013634432397068456,
+      "loss": 0.7566,
+      "step": 10917
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013633407455948554,
+      "loss": 0.8045,
+      "step": 10918
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013632382470852306,
+      "loss": 0.5734,
+      "step": 10919
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013631357441792126,
+      "loss": 0.6211,
+      "step": 10920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013630332368780418,
+      "loss": 0.5544,
+      "step": 10921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013629307251829584,
+      "loss": 0.6377,
+      "step": 10922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013628282090952037,
+      "loss": 0.6181,
+      "step": 10923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013627256886160178,
+      "loss": 0.6992,
+      "step": 10924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001362623163746642,
+      "loss": 0.614,
+      "step": 10925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013625206344883176,
+      "loss": 0.69,
+      "step": 10926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013624181008422843,
+      "loss": 0.6654,
+      "step": 10927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013623155628097842,
+      "loss": 0.6381,
+      "step": 10928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013622130203920586,
+      "loss": 0.6497,
+      "step": 10929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013621104735903473,
+      "loss": 0.6798,
+      "step": 10930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013620079224058924,
+      "loss": 0.6646,
+      "step": 10931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013619053668399348,
+      "loss": 0.6734,
+      "step": 10932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013618028068937163,
+      "loss": 0.6316,
+      "step": 10933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013617002425684774,
+      "loss": 0.5408,
+      "step": 10934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000136159767386546,
+      "loss": 0.7465,
+      "step": 10935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013614951007859054,
+      "loss": 0.5869,
+      "step": 10936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001361392523331055,
+      "loss": 0.6778,
+      "step": 10937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013612899415021504,
+      "loss": 0.5731,
+      "step": 10938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001361187355300433,
+      "loss": 0.7628,
+      "step": 10939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013610847647271447,
+      "loss": 0.8206,
+      "step": 10940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001360982169783527,
+      "loss": 0.6318,
+      "step": 10941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013608795704708222,
+      "loss": 0.5885,
+      "step": 10942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001360776966790271,
+      "loss": 0.7254,
+      "step": 10943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013606743587431163,
+      "loss": 0.6799,
+      "step": 10944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013605717463305994,
+      "loss": 0.7245,
+      "step": 10945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013604691295539625,
+      "loss": 0.5627,
+      "step": 10946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013603665084144475,
+      "loss": 0.5515,
+      "step": 10947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013602638829132962,
+      "loss": 0.7184,
+      "step": 10948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013601612530517512,
+      "loss": 0.7322,
+      "step": 10949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013600586188310546,
+      "loss": 0.7702,
+      "step": 10950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013599559802524483,
+      "loss": 0.6297,
+      "step": 10951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001359853337317175,
+      "loss": 0.6663,
+      "step": 10952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013597506900264766,
+      "loss": 0.737,
+      "step": 10953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013596480383815958,
+      "loss": 0.6102,
+      "step": 10954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013595453823837746,
+      "loss": 0.6661,
+      "step": 10955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013594427220342558,
+      "loss": 0.5659,
+      "step": 10956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001359340057334282,
+      "loss": 0.6931,
+      "step": 10957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013592373882850955,
+      "loss": 0.6076,
+      "step": 10958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013591347148879393,
+      "loss": 0.8014,
+      "step": 10959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001359032037144056,
+      "loss": 0.586,
+      "step": 10960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013589293550546878,
+      "loss": 0.6807,
+      "step": 10961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001358826668621078,
+      "loss": 0.617,
+      "step": 10962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013587239778444695,
+      "loss": 0.6467,
+      "step": 10963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013586212827261052,
+      "loss": 0.6484,
+      "step": 10964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001358518583267228,
+      "loss": 0.6245,
+      "step": 10965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013584158794690805,
+      "loss": 0.6334,
+      "step": 10966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013583131713329065,
+      "loss": 0.6634,
+      "step": 10967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013582104588599485,
+      "loss": 0.6693,
+      "step": 10968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000135810774205145,
+      "loss": 0.6184,
+      "step": 10969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013580050209086538,
+      "loss": 0.7628,
+      "step": 10970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013579022954328035,
+      "loss": 0.6369,
+      "step": 10971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013577995656251425,
+      "loss": 0.6468,
+      "step": 10972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013576968314869142,
+      "loss": 0.6741,
+      "step": 10973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013575940930193615,
+      "loss": 0.64,
+      "step": 10974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013574913502237284,
+      "loss": 0.5764,
+      "step": 10975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013573886031012584,
+      "loss": 0.6445,
+      "step": 10976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001357285851653195,
+      "loss": 0.7284,
+      "step": 10977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013571830958807816,
+      "loss": 0.6241,
+      "step": 10978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013570803357852623,
+      "loss": 0.596,
+      "step": 10979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013569775713678805,
+      "loss": 0.7216,
+      "step": 10980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013568748026298802,
+      "loss": 0.7748,
+      "step": 10981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013567720295725052,
+      "loss": 0.6683,
+      "step": 10982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001356669252196999,
+      "loss": 0.7483,
+      "step": 10983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001356566470504606,
+      "loss": 0.6015,
+      "step": 10984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013564636844965705,
+      "loss": 0.6226,
+      "step": 10985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001356360894174136,
+      "loss": 0.6544,
+      "step": 10986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013562580995385467,
+      "loss": 0.6579,
+      "step": 10987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013561553005910466,
+      "loss": 0.6662,
+      "step": 10988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013560524973328805,
+      "loss": 0.6466,
+      "step": 10989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013559496897652923,
+      "loss": 0.6434,
+      "step": 10990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013558468778895262,
+      "loss": 0.7586,
+      "step": 10991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013557440617068266,
+      "loss": 0.5554,
+      "step": 10992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013556412412184382,
+      "loss": 0.6896,
+      "step": 10993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013555384164256048,
+      "loss": 0.6273,
+      "step": 10994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013554355873295719,
+      "loss": 0.6408,
+      "step": 10995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013553327539315833,
+      "loss": 0.6583,
+      "step": 10996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001355229916232884,
+      "loss": 0.6927,
+      "step": 10997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001355127074234719,
+      "loss": 0.6719,
+      "step": 10998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001355024227938332,
+      "loss": 0.7247,
+      "step": 10999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013549213773449682,
+      "loss": 0.6695,
+      "step": 11000
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 0.6442050933837891,
+      "eval_runtime": 852.0931,
+      "eval_samples_per_second": 28.236,
+      "eval_steps_per_second": 14.118,
+      "step": 11000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013548185224558732,
+      "loss": 0.6086,
+      "step": 11001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001354715663272291,
+      "loss": 0.6735,
+      "step": 11002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001354612799795467,
+      "loss": 0.691,
+      "step": 11003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001354509932026646,
+      "loss": 0.6743,
+      "step": 11004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001354407059967073,
+      "loss": 0.5789,
+      "step": 11005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001354304183617993,
+      "loss": 0.6134,
+      "step": 11006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013542013029806517,
+      "loss": 0.6056,
+      "step": 11007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001354098418056294,
+      "loss": 0.6424,
+      "step": 11008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013539955288461645,
+      "loss": 0.6593,
+      "step": 11009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013538926353515095,
+      "loss": 0.5748,
+      "step": 11010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001353789737573574,
+      "loss": 0.6499,
+      "step": 11011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013536868355136029,
+      "loss": 0.6527,
+      "step": 11012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013535839291728424,
+      "loss": 0.5745,
+      "step": 11013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013534810185525378,
+      "loss": 0.7692,
+      "step": 11014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013533781036539344,
+      "loss": 0.7237,
+      "step": 11015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001353275184478278,
+      "loss": 0.597,
+      "step": 11016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013531722610268142,
+      "loss": 0.644,
+      "step": 11017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013530693333007888,
+      "loss": 0.6905,
+      "step": 11018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013529664013014475,
+      "loss": 0.6512,
+      "step": 11019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013528634650300365,
+      "loss": 0.6617,
+      "step": 11020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001352760524487801,
+      "loss": 0.7171,
+      "step": 11021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001352657579675987,
+      "loss": 0.6487,
+      "step": 11022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013525546305958412,
+      "loss": 0.5423,
+      "step": 11023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001352451677248609,
+      "loss": 0.7198,
+      "step": 11024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013523487196355363,
+      "loss": 0.6688,
+      "step": 11025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000135224575775787,
+      "loss": 0.5894,
+      "step": 11026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013521427916168553,
+      "loss": 0.6257,
+      "step": 11027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013520398212137395,
+      "loss": 0.569,
+      "step": 11028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013519368465497684,
+      "loss": 0.733,
+      "step": 11029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001351833867626188,
+      "loss": 0.6805,
+      "step": 11030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001351730884444245,
+      "loss": 0.6974,
+      "step": 11031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013516278970051858,
+      "loss": 0.6697,
+      "step": 11032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013515249053102572,
+      "loss": 0.6966,
+      "step": 11033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001351421909360705,
+      "loss": 0.6394,
+      "step": 11034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013513189091577768,
+      "loss": 0.5858,
+      "step": 11035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013512159047027184,
+      "loss": 0.5966,
+      "step": 11036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013511128959967764,
+      "loss": 0.7447,
+      "step": 11037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013510098830411986,
+      "loss": 0.6936,
+      "step": 11038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013509068658372307,
+      "loss": 0.5813,
+      "step": 11039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000135080384438612,
+      "loss": 0.7598,
+      "step": 11040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013507008186891138,
+      "loss": 0.5955,
+      "step": 11041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013505977887474582,
+      "loss": 0.7338,
+      "step": 11042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001350494754562401,
+      "loss": 0.7513,
+      "step": 11043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013503917161351888,
+      "loss": 0.6046,
+      "step": 11044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001350288673467069,
+      "loss": 0.6042,
+      "step": 11045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001350185626559288,
+      "loss": 0.6375,
+      "step": 11046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001350082575413094,
+      "loss": 0.6923,
+      "step": 11047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013499795200297337,
+      "loss": 0.6033,
+      "step": 11048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001349876460410455,
+      "loss": 0.7349,
+      "step": 11049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001349773396556504,
+      "loss": 0.6357,
+      "step": 11050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013496703284691299,
+      "loss": 0.6083,
+      "step": 11051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013495672561495784,
+      "loss": 0.5898,
+      "step": 11052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013494641795990986,
+      "loss": 0.5945,
+      "step": 11053
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001349361098818937,
+      "loss": 0.6284,
+      "step": 11054
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001349258013810341,
+      "loss": 0.7007,
+      "step": 11055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013491549245745596,
+      "loss": 0.7467,
+      "step": 11056
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013490518311128392,
+      "loss": 0.5883,
+      "step": 11057
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013489487334264285,
+      "loss": 0.6393,
+      "step": 11058
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013488456315165746,
+      "loss": 0.7139,
+      "step": 11059
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013487425253845255,
+      "loss": 0.661,
+      "step": 11060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000134863941503153,
+      "loss": 0.5379,
+      "step": 11061
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013485363004588348,
+      "loss": 0.6755,
+      "step": 11062
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013484331816676886,
+      "loss": 0.5951,
+      "step": 11063
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000134833005865934,
+      "loss": 0.6275,
+      "step": 11064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013482269314350358,
+      "loss": 0.5795,
+      "step": 11065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013481237999960253,
+      "loss": 0.7113,
+      "step": 11066
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013480206643435566,
+      "loss": 0.6449,
+      "step": 11067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001347917524478877,
+      "loss": 0.5849,
+      "step": 11068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013478143804032363,
+      "loss": 0.6934,
+      "step": 11069
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013477112321178819,
+      "loss": 0.6747,
+      "step": 11070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013476080796240625,
+      "loss": 0.6908,
+      "step": 11071
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013475049229230268,
+      "loss": 0.6524,
+      "step": 11072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013474017620160227,
+      "loss": 0.6145,
+      "step": 11073
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013472985969042997,
+      "loss": 0.6347,
+      "step": 11074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001347195427589106,
+      "loss": 0.6016,
+      "step": 11075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000134709225407169,
+      "loss": 0.5822,
+      "step": 11076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013469890763533008,
+      "loss": 0.6161,
+      "step": 11077
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001346885894435187,
+      "loss": 0.7422,
+      "step": 11078
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013467827083185976,
+      "loss": 0.6494,
+      "step": 11079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013466795180047816,
+      "loss": 0.6456,
+      "step": 11080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013465763234949874,
+      "loss": 0.6764,
+      "step": 11081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001346473124790465,
+      "loss": 0.7612,
+      "step": 11082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013463699218924623,
+      "loss": 0.6054,
+      "step": 11083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013462667148022293,
+      "loss": 0.5807,
+      "step": 11084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013461635035210144,
+      "loss": 0.6857,
+      "step": 11085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013460602880500675,
+      "loss": 0.499,
+      "step": 11086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013459570683906376,
+      "loss": 0.6672,
+      "step": 11087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013458538445439738,
+      "loss": 0.6939,
+      "step": 11088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013457506165113253,
+      "loss": 0.6628,
+      "step": 11089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013456473842939423,
+      "loss": 0.7422,
+      "step": 11090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013455441478930734,
+      "loss": 0.5736,
+      "step": 11091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013454409073099688,
+      "loss": 0.5644,
+      "step": 11092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013453376625458777,
+      "loss": 0.6676,
+      "step": 11093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013452344136020492,
+      "loss": 0.5862,
+      "step": 11094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001345131160479734,
+      "loss": 0.7554,
+      "step": 11095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013450279031801808,
+      "loss": 0.6798,
+      "step": 11096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013449246417046403,
+      "loss": 0.6389,
+      "step": 11097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013448213760543616,
+      "loss": 0.6132,
+      "step": 11098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001344718106230595,
+      "loss": 0.7538,
+      "step": 11099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013446148322345905,
+      "loss": 0.6324,
+      "step": 11100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001344511554067597,
+      "loss": 0.6349,
+      "step": 11101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013444082717308658,
+      "loss": 0.5985,
+      "step": 11102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013443049852256462,
+      "loss": 0.5276,
+      "step": 11103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013442016945531887,
+      "loss": 0.6612,
+      "step": 11104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013440983997147434,
+      "loss": 0.6812,
+      "step": 11105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013439951007115602,
+      "loss": 0.6317,
+      "step": 11106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013438917975448897,
+      "loss": 0.6138,
+      "step": 11107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013437884902159822,
+      "loss": 0.675,
+      "step": 11108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013436851787260875,
+      "loss": 0.7025,
+      "step": 11109
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001343581863076457,
+      "loss": 0.6827,
+      "step": 11110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013434785432683404,
+      "loss": 0.622,
+      "step": 11111
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013433752193029886,
+      "loss": 0.6603,
+      "step": 11112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013432718911816522,
+      "loss": 0.7103,
+      "step": 11113
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013431685589055813,
+      "loss": 0.5913,
+      "step": 11114
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001343065222476027,
+      "loss": 0.6538,
+      "step": 11115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013429618818942398,
+      "loss": 0.6655,
+      "step": 11116
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013428585371614706,
+      "loss": 0.7163,
+      "step": 11117
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013427551882789704,
+      "loss": 0.6766,
+      "step": 11118
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013426518352479898,
+      "loss": 0.7671,
+      "step": 11119
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013425484780697796,
+      "loss": 0.6135,
+      "step": 11120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001342445116745591,
+      "loss": 0.6451,
+      "step": 11121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013423417512766752,
+      "loss": 0.4985,
+      "step": 11122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001342238381664283,
+      "loss": 0.6011,
+      "step": 11123
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013421350079096653,
+      "loss": 0.6167,
+      "step": 11124
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013420316300140738,
+      "loss": 0.6214,
+      "step": 11125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001341928247978759,
+      "loss": 0.6331,
+      "step": 11126
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001341824861804973,
+      "loss": 0.6655,
+      "step": 11127
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013417214714939667,
+      "loss": 0.7573,
+      "step": 11128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013416180770469912,
+      "loss": 0.6086,
+      "step": 11129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013415146784652985,
+      "loss": 0.6787,
+      "step": 11130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013414112757501398,
+      "loss": 0.6613,
+      "step": 11131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013413078689027662,
+      "loss": 0.6941,
+      "step": 11132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000134120445792443,
+      "loss": 0.6062,
+      "step": 11133
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001341101042816382,
+      "loss": 0.6193,
+      "step": 11134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013409976235798752,
+      "loss": 0.6729,
+      "step": 11135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013408942002161598,
+      "loss": 0.5718,
+      "step": 11136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013407907727264884,
+      "loss": 0.6674,
+      "step": 11137
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013406873411121126,
+      "loss": 0.6748,
+      "step": 11138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001340583905374284,
+      "loss": 0.5862,
+      "step": 11139
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013404804655142556,
+      "loss": 0.6316,
+      "step": 11140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001340377021533278,
+      "loss": 0.5865,
+      "step": 11141
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001340273573432604,
+      "loss": 0.5353,
+      "step": 11142
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013401701212134855,
+      "loss": 0.6597,
+      "step": 11143
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013400666648771744,
+      "loss": 0.6646,
+      "step": 11144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001339963204424923,
+      "loss": 0.5856,
+      "step": 11145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013398597398579838,
+      "loss": 0.6321,
+      "step": 11146
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001339756271177609,
+      "loss": 0.5508,
+      "step": 11147
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013396527983850504,
+      "loss": 0.6509,
+      "step": 11148
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013395493214815612,
+      "loss": 0.6681,
+      "step": 11149
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013394458404683928,
+      "loss": 0.7149,
+      "step": 11150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013393423553467986,
+      "loss": 0.6444,
+      "step": 11151
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013392388661180303,
+      "loss": 0.7019,
+      "step": 11152
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013391353727833412,
+      "loss": 0.6274,
+      "step": 11153
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013390318753439835,
+      "loss": 0.6751,
+      "step": 11154
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013389283738012098,
+      "loss": 0.6218,
+      "step": 11155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013388248681562735,
+      "loss": 0.6524,
+      "step": 11156
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013387213584104265,
+      "loss": 0.7107,
+      "step": 11157
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013386178445649218,
+      "loss": 0.6832,
+      "step": 11158
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013385143266210129,
+      "loss": 0.6858,
+      "step": 11159
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013384108045799516,
+      "loss": 0.6295,
+      "step": 11160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013383072784429917,
+      "loss": 0.6686,
+      "step": 11161
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013382037482113863,
+      "loss": 0.6077,
+      "step": 11162
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001338100213886388,
+      "loss": 0.5816,
+      "step": 11163
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000133799667546925,
+      "loss": 0.7054,
+      "step": 11164
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013378931329612254,
+      "loss": 0.6247,
+      "step": 11165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013377895863635677,
+      "loss": 0.5402,
+      "step": 11166
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000133768603567753,
+      "loss": 0.7166,
+      "step": 11167
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013375824809043658,
+      "loss": 0.7161,
+      "step": 11168
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001337478922045328,
+      "loss": 0.621,
+      "step": 11169
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013373753591016706,
+      "loss": 0.6304,
+      "step": 11170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013372717920746464,
+      "loss": 0.6366,
+      "step": 11171
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013371682209655091,
+      "loss": 0.5521,
+      "step": 11172
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013370646457755127,
+      "loss": 0.6921,
+      "step": 11173
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013369610665059104,
+      "loss": 0.5471,
+      "step": 11174
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001336857483157956,
+      "loss": 0.6324,
+      "step": 11175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001336753895732903,
+      "loss": 0.6651,
+      "step": 11176
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013366503042320057,
+      "loss": 0.5856,
+      "step": 11177
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013365467086565172,
+      "loss": 0.6213,
+      "step": 11178
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013364431090076917,
+      "loss": 0.6984,
+      "step": 11179
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013363395052867828,
+      "loss": 0.5836,
+      "step": 11180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001336235897495045,
+      "loss": 0.592,
+      "step": 11181
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001336132285633732,
+      "loss": 0.716,
+      "step": 11182
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013360286697040977,
+      "loss": 0.7155,
+      "step": 11183
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013359250497073968,
+      "loss": 0.5978,
+      "step": 11184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013358214256448827,
+      "loss": 0.7168,
+      "step": 11185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000133571779751781,
+      "loss": 0.6418,
+      "step": 11186
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013356141653274328,
+      "loss": 0.5901,
+      "step": 11187
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013355105290750055,
+      "loss": 0.6301,
+      "step": 11188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013354068887617824,
+      "loss": 0.5892,
+      "step": 11189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013353032443890175,
+      "loss": 0.5796,
+      "step": 11190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001335199595957966,
+      "loss": 0.6551,
+      "step": 11191
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001335095943469882,
+      "loss": 0.5785,
+      "step": 11192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000133499228692602,
+      "loss": 0.5931,
+      "step": 11193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013348886263276345,
+      "loss": 0.6994,
+      "step": 11194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013347849616759805,
+      "loss": 0.585,
+      "step": 11195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013346812929723124,
+      "loss": 0.4925,
+      "step": 11196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013345776202178852,
+      "loss": 0.6686,
+      "step": 11197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013344739434139534,
+      "loss": 0.579,
+      "step": 11198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013343702625617717,
+      "loss": 0.5839,
+      "step": 11199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013342665776625956,
+      "loss": 0.7242,
+      "step": 11200
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 0.6374056339263916,
+      "eval_runtime": 852.285,
+      "eval_samples_per_second": 28.23,
+      "eval_steps_per_second": 14.115,
+      "step": 11200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013341628887176792,
+      "loss": 0.7801,
+      "step": 11201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001334059195728278,
+      "loss": 0.6669,
+      "step": 11202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013339554986956472,
+      "loss": 0.6118,
+      "step": 11203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013338517976210415,
+      "loss": 0.644,
+      "step": 11204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013337480925057165,
+      "loss": 0.6646,
+      "step": 11205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013336443833509266,
+      "loss": 0.6458,
+      "step": 11206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001333540670157928,
+      "loss": 0.6919,
+      "step": 11207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013334369529279747,
+      "loss": 0.7153,
+      "step": 11208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013333332316623232,
+      "loss": 0.6403,
+      "step": 11209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013332295063622285,
+      "loss": 0.6941,
+      "step": 11210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001333125777028946,
+      "loss": 0.6679,
+      "step": 11211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001333022043663731,
+      "loss": 0.686,
+      "step": 11212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013329183062678394,
+      "loss": 0.6407,
+      "step": 11213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013328145648425262,
+      "loss": 0.6012,
+      "step": 11214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013327108193890474,
+      "loss": 0.6209,
+      "step": 11215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013326070699086587,
+      "loss": 0.5179,
+      "step": 11216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013325033164026157,
+      "loss": 0.623,
+      "step": 11217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013323995588721742,
+      "loss": 0.6343,
+      "step": 11218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000133229579731859,
+      "loss": 0.6329,
+      "step": 11219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001332192031743119,
+      "loss": 0.5912,
+      "step": 11220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013320882621470166,
+      "loss": 0.7006,
+      "step": 11221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000133198448853154,
+      "loss": 0.7227,
+      "step": 11222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001331880710897944,
+      "loss": 0.5645,
+      "step": 11223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013317769292474853,
+      "loss": 0.6439,
+      "step": 11224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013316731435814196,
+      "loss": 0.6271,
+      "step": 11225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013315693539010033,
+      "loss": 0.7359,
+      "step": 11226
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013314655602074925,
+      "loss": 0.6651,
+      "step": 11227
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013313617625021438,
+      "loss": 0.5168,
+      "step": 11228
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001331257960786213,
+      "loss": 0.6951,
+      "step": 11229
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013311541550609565,
+      "loss": 0.5918,
+      "step": 11230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001331050345327631,
+      "loss": 0.6733,
+      "step": 11231
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013309465315874928,
+      "loss": 0.6109,
+      "step": 11232
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013308427138417983,
+      "loss": 0.6867,
+      "step": 11233
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001330738892091804,
+      "loss": 0.6791,
+      "step": 11234
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001330635066338767,
+      "loss": 0.5694,
+      "step": 11235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013305312365839432,
+      "loss": 0.7648,
+      "step": 11236
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013304274028285896,
+      "loss": 0.693,
+      "step": 11237
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013303235650739633,
+      "loss": 0.5712,
+      "step": 11238
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013302197233213203,
+      "loss": 0.657,
+      "step": 11239
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001330115877571918,
+      "loss": 0.5694,
+      "step": 11240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013300120278270133,
+      "loss": 0.7009,
+      "step": 11241
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001329908174087863,
+      "loss": 0.5963,
+      "step": 11242
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013298043163557238,
+      "loss": 0.6046,
+      "step": 11243
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001329700454631853,
+      "loss": 0.68,
+      "step": 11244
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001329596588917508,
+      "loss": 0.6677,
+      "step": 11245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013294927192139454,
+      "loss": 0.7795,
+      "step": 11246
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013293888455224225,
+      "loss": 0.6129,
+      "step": 11247
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013292849678441968,
+      "loss": 0.7453,
+      "step": 11248
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001329181086180525,
+      "loss": 0.7247,
+      "step": 11249
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013290772005326648,
+      "loss": 0.5656,
+      "step": 11250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013289733109018737,
+      "loss": 0.6698,
+      "step": 11251
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013288694172894087,
+      "loss": 0.6365,
+      "step": 11252
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001328765519696528,
+      "loss": 0.5995,
+      "step": 11253
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001328661618124488,
+      "loss": 0.6196,
+      "step": 11254
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013285577125745469,
+      "loss": 0.674,
+      "step": 11255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013284538030479626,
+      "loss": 0.5887,
+      "step": 11256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001328349889545992,
+      "loss": 0.6918,
+      "step": 11257
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013282459720698935,
+      "loss": 0.6421,
+      "step": 11258
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013281420506209242,
+      "loss": 0.6473,
+      "step": 11259
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013280381252003427,
+      "loss": 0.6182,
+      "step": 11260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013279341958094063,
+      "loss": 0.5741,
+      "step": 11261
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013278302624493725,
+      "loss": 0.5426,
+      "step": 11262
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013277263251215002,
+      "loss": 0.7047,
+      "step": 11263
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001327622383827047,
+      "loss": 0.6088,
+      "step": 11264
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013275184385672705,
+      "loss": 0.6454,
+      "step": 11265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013274144893434293,
+      "loss": 0.6904,
+      "step": 11266
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013273105361567811,
+      "loss": 0.7282,
+      "step": 11267
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013272065790085846,
+      "loss": 0.6071,
+      "step": 11268
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013271026179000978,
+      "loss": 0.635,
+      "step": 11269
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001326998652832579,
+      "loss": 0.6295,
+      "step": 11270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013268946838072861,
+      "loss": 0.6634,
+      "step": 11271
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001326790710825478,
+      "loss": 0.5732,
+      "step": 11272
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001326686733888413,
+      "loss": 0.5911,
+      "step": 11273
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013265827529973496,
+      "loss": 0.6109,
+      "step": 11274
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013264787681535462,
+      "loss": 0.6665,
+      "step": 11275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013263747793582615,
+      "loss": 0.7301,
+      "step": 11276
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001326270786612754,
+      "loss": 0.7024,
+      "step": 11277
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001326166789918282,
+      "loss": 0.6324,
+      "step": 11278
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013260627892761048,
+      "loss": 0.7911,
+      "step": 11279
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001325958784687481,
+      "loss": 0.5941,
+      "step": 11280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013258547761536694,
+      "loss": 0.6864,
+      "step": 11281
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001325750763675929,
+      "loss": 0.7167,
+      "step": 11282
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001325646747255518,
+      "loss": 0.6585,
+      "step": 11283
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001325542726893696,
+      "loss": 0.6754,
+      "step": 11284
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013254387025917217,
+      "loss": 0.6961,
+      "step": 11285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013253346743508546,
+      "loss": 0.595,
+      "step": 11286
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013252306421723532,
+      "loss": 0.6587,
+      "step": 11287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001325126606057477,
+      "loss": 0.6662,
+      "step": 11288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013250225660074852,
+      "loss": 0.6143,
+      "step": 11289
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013249185220236365,
+      "loss": 0.7721,
+      "step": 11290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001324814474107191,
+      "loss": 0.6494,
+      "step": 11291
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013247104222594076,
+      "loss": 0.7284,
+      "step": 11292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013246063664815454,
+      "loss": 0.5996,
+      "step": 11293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001324502306774864,
+      "loss": 0.6436,
+      "step": 11294
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013243982431406233,
+      "loss": 0.618,
+      "step": 11295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013242941755800823,
+      "loss": 0.7047,
+      "step": 11296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001324190104094501,
+      "loss": 0.7427,
+      "step": 11297
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013240860286851386,
+      "loss": 0.5826,
+      "step": 11298
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001323981949353255,
+      "loss": 0.6126,
+      "step": 11299
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000132387786610011,
+      "loss": 0.7227,
+      "step": 11300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013237737789269628,
+      "loss": 0.5737,
+      "step": 11301
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001323669687835074,
+      "loss": 0.5828,
+      "step": 11302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013235655928257027,
+      "loss": 0.6262,
+      "step": 11303
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013234614939001092,
+      "loss": 0.5518,
+      "step": 11304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013233573910595534,
+      "loss": 0.691,
+      "step": 11305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013232532843052955,
+      "loss": 0.6065,
+      "step": 11306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001323149173638595,
+      "loss": 0.7285,
+      "step": 11307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013230450590607124,
+      "loss": 0.6789,
+      "step": 11308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001322940940572908,
+      "loss": 0.6037,
+      "step": 11309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013228368181764414,
+      "loss": 0.6181,
+      "step": 11310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013227326918725728,
+      "loss": 0.6232,
+      "step": 11311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013226285616625634,
+      "loss": 0.7693,
+      "step": 11312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013225244275476728,
+      "loss": 0.5949,
+      "step": 11313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013224202895291613,
+      "loss": 0.665,
+      "step": 11314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013223161476082898,
+      "loss": 0.6703,
+      "step": 11315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013222120017863183,
+      "loss": 0.6821,
+      "step": 11316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013221078520645076,
+      "loss": 0.7087,
+      "step": 11317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001322003698444118,
+      "loss": 0.533,
+      "step": 11318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013218995409264105,
+      "loss": 0.5971,
+      "step": 11319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013217953795126457,
+      "loss": 0.5416,
+      "step": 11320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013216912142040836,
+      "loss": 0.7266,
+      "step": 11321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013215870450019858,
+      "loss": 0.6345,
+      "step": 11322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013214828719076125,
+      "loss": 0.5775,
+      "step": 11323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001321378694922225,
+      "loss": 0.6681,
+      "step": 11324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001321274514047084,
+      "loss": 0.6622,
+      "step": 11325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013211703292834504,
+      "loss": 0.6531,
+      "step": 11326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013210661406325852,
+      "loss": 0.6228,
+      "step": 11327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013209619480957497,
+      "loss": 0.705,
+      "step": 11328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001320857751674204,
+      "loss": 0.6245,
+      "step": 11329
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013207535513692106,
+      "loss": 0.6626,
+      "step": 11330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013206493471820297,
+      "loss": 0.5634,
+      "step": 11331
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001320545139113923,
+      "loss": 0.6656,
+      "step": 11332
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013204409271661515,
+      "loss": 0.7024,
+      "step": 11333
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013203367113399765,
+      "loss": 0.6158,
+      "step": 11334
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013202324916366598,
+      "loss": 0.6899,
+      "step": 11335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001320128268057462,
+      "loss": 0.6332,
+      "step": 11336
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013200240406036455,
+      "loss": 0.7467,
+      "step": 11337
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013199198092764714,
+      "loss": 0.6645,
+      "step": 11338
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013198155740772006,
+      "loss": 0.6823,
+      "step": 11339
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013197113350070955,
+      "loss": 0.6101,
+      "step": 11340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013196070920674176,
+      "loss": 0.5912,
+      "step": 11341
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013195028452594285,
+      "loss": 0.8002,
+      "step": 11342
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013193985945843899,
+      "loss": 0.7414,
+      "step": 11343
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013192943400435636,
+      "loss": 0.7201,
+      "step": 11344
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013191900816382114,
+      "loss": 0.6733,
+      "step": 11345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013190858193695953,
+      "loss": 0.5808,
+      "step": 11346
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013189815532389768,
+      "loss": 0.6601,
+      "step": 11347
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013188772832476188,
+      "loss": 0.7691,
+      "step": 11348
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001318773009396782,
+      "loss": 0.7604,
+      "step": 11349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013186687316877298,
+      "loss": 0.6013,
+      "step": 11350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013185644501217234,
+      "loss": 0.6249,
+      "step": 11351
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013184601647000252,
+      "loss": 0.5621,
+      "step": 11352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013183558754238975,
+      "loss": 0.6838,
+      "step": 11353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013182515822946023,
+      "loss": 0.5586,
+      "step": 11354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013181472853134024,
+      "loss": 0.6896,
+      "step": 11355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000131804298448156,
+      "loss": 0.6683,
+      "step": 11356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013179386798003366,
+      "loss": 0.637,
+      "step": 11357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013178343712709958,
+      "loss": 0.6491,
+      "step": 11358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013177300588947994,
+      "loss": 0.7019,
+      "step": 11359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013176257426730106,
+      "loss": 0.5581,
+      "step": 11360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001317521422606891,
+      "loss": 0.6537,
+      "step": 11361
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013174170986977039,
+      "loss": 0.5318,
+      "step": 11362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001317312770946712,
+      "loss": 0.697,
+      "step": 11363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001317208439355178,
+      "loss": 0.6295,
+      "step": 11364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001317104103924364,
+      "loss": 0.6477,
+      "step": 11365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013169997646555334,
+      "loss": 0.5782,
+      "step": 11366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013168954215499488,
+      "loss": 0.689,
+      "step": 11367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013167910746088735,
+      "loss": 0.5647,
+      "step": 11368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013166867238335704,
+      "loss": 0.7875,
+      "step": 11369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001316582369225302,
+      "loss": 0.6251,
+      "step": 11370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001316478010785332,
+      "loss": 0.6394,
+      "step": 11371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013163736485149224,
+      "loss": 0.791,
+      "step": 11372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013162692824153374,
+      "loss": 0.5917,
+      "step": 11373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013161649124878398,
+      "loss": 0.5618,
+      "step": 11374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001316060538733693,
+      "loss": 0.5927,
+      "step": 11375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000131595616115416,
+      "loss": 0.6097,
+      "step": 11376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013158517797505042,
+      "loss": 0.6631,
+      "step": 11377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013157473945239891,
+      "loss": 0.6553,
+      "step": 11378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001315643005475878,
+      "loss": 0.6839,
+      "step": 11379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001315538612607434,
+      "loss": 0.6144,
+      "step": 11380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013154342159199214,
+      "loss": 0.696,
+      "step": 11381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013153298154146027,
+      "loss": 0.6623,
+      "step": 11382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013152254110927427,
+      "loss": 0.5647,
+      "step": 11383
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013151210029556043,
+      "loss": 0.6192,
+      "step": 11384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001315016591004451,
+      "loss": 0.5826,
+      "step": 11385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013149121752405472,
+      "loss": 0.638,
+      "step": 11386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013148077556651564,
+      "loss": 0.6684,
+      "step": 11387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013147033322795422,
+      "loss": 0.6095,
+      "step": 11388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013145989050849686,
+      "loss": 0.5903,
+      "step": 11389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013144944740826995,
+      "loss": 0.6333,
+      "step": 11390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013143900392739992,
+      "loss": 0.6461,
+      "step": 11391
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001314285600660131,
+      "loss": 0.6333,
+      "step": 11392
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013141811582423598,
+      "loss": 0.6614,
+      "step": 11393
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001314076712021949,
+      "loss": 0.5388,
+      "step": 11394
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013139722620001632,
+      "loss": 0.5171,
+      "step": 11395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013138678081782662,
+      "loss": 0.6908,
+      "step": 11396
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013137633505575226,
+      "loss": 0.6191,
+      "step": 11397
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013136588891391965,
+      "loss": 0.5424,
+      "step": 11398
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013135544239245523,
+      "loss": 0.6873,
+      "step": 11399
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013134499549148542,
+      "loss": 0.6046,
+      "step": 11400
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 0.631673276424408,
+      "eval_runtime": 853.1984,
+      "eval_samples_per_second": 28.2,
+      "eval_steps_per_second": 14.1,
+      "step": 11400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001313345482111367,
+      "loss": 0.5184,
+      "step": 11401
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013132410055153551,
+      "loss": 0.6168,
+      "step": 11402
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013131365251280825,
+      "loss": 0.6808,
+      "step": 11403
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013130320409508142,
+      "loss": 0.7064,
+      "step": 11404
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013129275529848145,
+      "loss": 0.6126,
+      "step": 11405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013128230612313488,
+      "loss": 0.8312,
+      "step": 11406
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001312718565691681,
+      "loss": 0.5802,
+      "step": 11407
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001312614066367076,
+      "loss": 0.5865,
+      "step": 11408
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013125095632587991,
+      "loss": 0.6936,
+      "step": 11409
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013124050563681142,
+      "loss": 0.6424,
+      "step": 11410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013123005456962874,
+      "loss": 0.6347,
+      "step": 11411
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013121960312445827,
+      "loss": 0.6711,
+      "step": 11412
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001312091513014265,
+      "loss": 0.5502,
+      "step": 11413
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013119869910066,
+      "loss": 0.6172,
+      "step": 11414
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013118824652228524,
+      "loss": 0.6488,
+      "step": 11415
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013117779356642872,
+      "loss": 0.7205,
+      "step": 11416
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013116734023321697,
+      "loss": 0.7244,
+      "step": 11417
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001311568865227765,
+      "loss": 0.6659,
+      "step": 11418
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013114643243523387,
+      "loss": 0.593,
+      "step": 11419
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013113597797071558,
+      "loss": 0.7032,
+      "step": 11420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013112552312934814,
+      "loss": 0.6631,
+      "step": 11421
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001311150679112581,
+      "loss": 0.6642,
+      "step": 11422
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013110461231657206,
+      "loss": 0.6784,
+      "step": 11423
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013109415634541651,
+      "loss": 0.5878,
+      "step": 11424
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013108369999791802,
+      "loss": 0.6259,
+      "step": 11425
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013107324327420313,
+      "loss": 0.6255,
+      "step": 11426
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013106278617439844,
+      "loss": 0.6287,
+      "step": 11427
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013105232869863048,
+      "loss": 0.5817,
+      "step": 11428
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013104187084702585,
+      "loss": 0.693,
+      "step": 11429
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013103141261971106,
+      "loss": 0.499,
+      "step": 11430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013102095401681276,
+      "loss": 0.6957,
+      "step": 11431
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013101049503845753,
+      "loss": 0.6315,
+      "step": 11432
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001310000356847719,
+      "loss": 0.5844,
+      "step": 11433
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013098957595588252,
+      "loss": 0.583,
+      "step": 11434
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013097911585191599,
+      "loss": 0.692,
+      "step": 11435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013096865537299887,
+      "loss": 0.6532,
+      "step": 11436
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001309581945192578,
+      "loss": 0.7083,
+      "step": 11437
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013094773329081936,
+      "loss": 0.5713,
+      "step": 11438
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013093727168781019,
+      "loss": 0.596,
+      "step": 11439
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001309268097103569,
+      "loss": 0.725,
+      "step": 11440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001309163473585861,
+      "loss": 0.7261,
+      "step": 11441
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013090588463262449,
+      "loss": 0.6321,
+      "step": 11442
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001308954215325986,
+      "loss": 0.612,
+      "step": 11443
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013088495805863517,
+      "loss": 0.5898,
+      "step": 11444
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013087449421086077,
+      "loss": 0.6432,
+      "step": 11445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001308640299894021,
+      "loss": 0.6127,
+      "step": 11446
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013085356539438574,
+      "loss": 0.6597,
+      "step": 11447
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001308431004259384,
+      "loss": 0.7923,
+      "step": 11448
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013083263508418673,
+      "loss": 0.682,
+      "step": 11449
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013082216936925742,
+      "loss": 0.674,
+      "step": 11450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013081170328127706,
+      "loss": 0.6417,
+      "step": 11451
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013080123682037247,
+      "loss": 0.5965,
+      "step": 11452
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013079076998667017,
+      "loss": 0.6042,
+      "step": 11453
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013078030278029695,
+      "loss": 0.6585,
+      "step": 11454
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013076983520137945,
+      "loss": 0.613,
+      "step": 11455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013075936725004435,
+      "loss": 0.6726,
+      "step": 11456
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013074889892641839,
+      "loss": 0.5182,
+      "step": 11457
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013073843023062826,
+      "loss": 0.6443,
+      "step": 11458
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013072796116280066,
+      "loss": 0.6266,
+      "step": 11459
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013071749172306228,
+      "loss": 0.7219,
+      "step": 11460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013070702191153986,
+      "loss": 0.6613,
+      "step": 11461
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013069655172836015,
+      "loss": 0.661,
+      "step": 11462
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001306860811736498,
+      "loss": 0.5848,
+      "step": 11463
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013067561024753557,
+      "loss": 0.6264,
+      "step": 11464
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013066513895014425,
+      "loss": 0.6466,
+      "step": 11465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013065466728160252,
+      "loss": 0.6638,
+      "step": 11466
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013064419524203707,
+      "loss": 0.5579,
+      "step": 11467
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013063372283157477,
+      "loss": 0.5693,
+      "step": 11468
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013062325005034226,
+      "loss": 0.7087,
+      "step": 11469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013061277689846638,
+      "loss": 0.6276,
+      "step": 11470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013060230337607384,
+      "loss": 0.5665,
+      "step": 11471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013059182948329143,
+      "loss": 0.6896,
+      "step": 11472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013058135522024588,
+      "loss": 0.6126,
+      "step": 11473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013057088058706398,
+      "loss": 0.6002,
+      "step": 11474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013056040558387258,
+      "loss": 0.577,
+      "step": 11475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013054993021079838,
+      "loss": 0.7014,
+      "step": 11476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013053945446796813,
+      "loss": 0.743,
+      "step": 11477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013052897835550876,
+      "loss": 0.6461,
+      "step": 11478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013051850187354694,
+      "loss": 0.6521,
+      "step": 11479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001305080250222095,
+      "loss": 0.5367,
+      "step": 11480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001304975478016233,
+      "loss": 0.6151,
+      "step": 11481
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013048707021191507,
+      "loss": 0.5946,
+      "step": 11482
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001304765922532117,
+      "loss": 0.5952,
+      "step": 11483
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013046611392563995,
+      "loss": 0.6612,
+      "step": 11484
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013045563522932666,
+      "loss": 0.6344,
+      "step": 11485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001304451561643987,
+      "loss": 0.6319,
+      "step": 11486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013043467673098282,
+      "loss": 0.6325,
+      "step": 11487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001304241969292059,
+      "loss": 0.6501,
+      "step": 11488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013041371675919482,
+      "loss": 0.5632,
+      "step": 11489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013040323622107636,
+      "loss": 0.5497,
+      "step": 11490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001303927553149774,
+      "loss": 0.6274,
+      "step": 11491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013038227404102475,
+      "loss": 0.58,
+      "step": 11492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013037179239934533,
+      "loss": 0.6526,
+      "step": 11493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000130361310390066,
+      "loss": 0.6432,
+      "step": 11494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013035082801331356,
+      "loss": 0.6152,
+      "step": 11495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000130340345269215,
+      "loss": 0.7372,
+      "step": 11496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001303298621578971,
+      "loss": 0.5873,
+      "step": 11497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013031937867948674,
+      "loss": 0.5673,
+      "step": 11498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013030889483411082,
+      "loss": 0.5941,
+      "step": 11499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013029841062189626,
+      "loss": 0.5688,
+      "step": 11500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013028792604296993,
+      "loss": 0.6683,
+      "step": 11501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013027744109745873,
+      "loss": 0.5838,
+      "step": 11502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013026695578548957,
+      "loss": 0.6662,
+      "step": 11503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013025647010718938,
+      "loss": 0.5978,
+      "step": 11504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000130245984062685,
+      "loss": 0.622,
+      "step": 11505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013023549765210342,
+      "loss": 0.6082,
+      "step": 11506
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013022501087557147,
+      "loss": 0.5669,
+      "step": 11507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013021452373321622,
+      "loss": 0.6375,
+      "step": 11508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013020403622516447,
+      "loss": 0.6409,
+      "step": 11509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013019354835154317,
+      "loss": 0.5636,
+      "step": 11510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013018306011247934,
+      "loss": 0.6435,
+      "step": 11511
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013017257150809984,
+      "loss": 0.547,
+      "step": 11512
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013016208253853164,
+      "loss": 0.6961,
+      "step": 11513
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013015159320390173,
+      "loss": 0.6966,
+      "step": 11514
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000130141103504337,
+      "loss": 0.6502,
+      "step": 11515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013013061343996444,
+      "loss": 0.5432,
+      "step": 11516
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013012012301091106,
+      "loss": 0.5796,
+      "step": 11517
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013010963221730373,
+      "loss": 0.5896,
+      "step": 11518
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013009914105926952,
+      "loss": 0.6366,
+      "step": 11519
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013008864953693535,
+      "loss": 0.6126,
+      "step": 11520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013007815765042824,
+      "loss": 0.655,
+      "step": 11521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013006766539987516,
+      "loss": 0.6651,
+      "step": 11522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013005717278540308,
+      "loss": 0.6158,
+      "step": 11523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013004667980713906,
+      "loss": 0.6469,
+      "step": 11524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013003618646520999,
+      "loss": 0.6656,
+      "step": 11525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013002569275974297,
+      "loss": 0.5822,
+      "step": 11526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000130015198690865,
+      "loss": 0.5693,
+      "step": 11527
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013000470425870303,
+      "loss": 0.5523,
+      "step": 11528
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012999420946338414,
+      "loss": 0.5455,
+      "step": 11529
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012998371430503536,
+      "loss": 0.6098,
+      "step": 11530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012997321878378362,
+      "loss": 0.6411,
+      "step": 11531
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012996272289975607,
+      "loss": 0.6658,
+      "step": 11532
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012995222665307968,
+      "loss": 0.5471,
+      "step": 11533
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012994173004388153,
+      "loss": 0.6004,
+      "step": 11534
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012993123307228861,
+      "loss": 0.6538,
+      "step": 11535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012992073573842798,
+      "loss": 0.591,
+      "step": 11536
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012991023804242673,
+      "loss": 0.656,
+      "step": 11537
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012989973998441188,
+      "loss": 0.6815,
+      "step": 11538
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012988924156451053,
+      "loss": 0.5704,
+      "step": 11539
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001298787427828497,
+      "loss": 0.5181,
+      "step": 11540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001298682436395565,
+      "loss": 0.5742,
+      "step": 11541
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012985774413475798,
+      "loss": 0.6443,
+      "step": 11542
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001298472442685812,
+      "loss": 0.6681,
+      "step": 11543
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012983674404115332,
+      "loss": 0.6312,
+      "step": 11544
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012982624345260138,
+      "loss": 0.6705,
+      "step": 11545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012981574250305243,
+      "loss": 0.6073,
+      "step": 11546
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001298052411926336,
+      "loss": 0.7288,
+      "step": 11547
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012979473952147205,
+      "loss": 0.6357,
+      "step": 11548
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001297842374896948,
+      "loss": 0.7242,
+      "step": 11549
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000129773735097429,
+      "loss": 0.5339,
+      "step": 11550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012976323234480173,
+      "loss": 0.5739,
+      "step": 11551
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012975272923194017,
+      "loss": 0.7602,
+      "step": 11552
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001297422257589714,
+      "loss": 0.7276,
+      "step": 11553
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012973172192602252,
+      "loss": 0.6653,
+      "step": 11554
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012972121773322074,
+      "loss": 0.6663,
+      "step": 11555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012971071318069313,
+      "loss": 0.6595,
+      "step": 11556
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012970020826856687,
+      "loss": 0.6703,
+      "step": 11557
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012968970299696907,
+      "loss": 0.6798,
+      "step": 11558
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012967919736602687,
+      "loss": 0.5983,
+      "step": 11559
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001296686913758675,
+      "loss": 0.5341,
+      "step": 11560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000129658185026618,
+      "loss": 0.6447,
+      "step": 11561
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012964767831840566,
+      "loss": 0.6185,
+      "step": 11562
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012963717125135754,
+      "loss": 0.6915,
+      "step": 11563
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012962666382560086,
+      "loss": 0.6137,
+      "step": 11564
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012961615604126282,
+      "loss": 0.728,
+      "step": 11565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012960564789847055,
+      "loss": 0.5862,
+      "step": 11566
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012959513939735123,
+      "loss": 0.6421,
+      "step": 11567
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001295846305380321,
+      "loss": 0.5996,
+      "step": 11568
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001295741213206403,
+      "loss": 0.5883,
+      "step": 11569
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012956361174530306,
+      "loss": 0.6556,
+      "step": 11570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001295531018121476,
+      "loss": 0.7775,
+      "step": 11571
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012954259152130106,
+      "loss": 0.6061,
+      "step": 11572
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001295320808728907,
+      "loss": 0.5818,
+      "step": 11573
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001295215698670437,
+      "loss": 0.6534,
+      "step": 11574
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001295110585038873,
+      "loss": 0.5504,
+      "step": 11575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001295005467835487,
+      "loss": 0.826,
+      "step": 11576
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012949003470615517,
+      "loss": 0.6803,
+      "step": 11577
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012947952227183393,
+      "loss": 0.6144,
+      "step": 11578
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012946900948071217,
+      "loss": 0.6087,
+      "step": 11579
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012945849633291717,
+      "loss": 0.5962,
+      "step": 11580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012944798282857618,
+      "loss": 0.6291,
+      "step": 11581
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012943746896781642,
+      "loss": 0.6646,
+      "step": 11582
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012942695475076516,
+      "loss": 0.6007,
+      "step": 11583
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012941644017754964,
+      "loss": 0.5857,
+      "step": 11584
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012940592524829717,
+      "loss": 0.6475,
+      "step": 11585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012939540996313498,
+      "loss": 0.5779,
+      "step": 11586
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001293848943221903,
+      "loss": 0.6245,
+      "step": 11587
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012937437832559047,
+      "loss": 0.5916,
+      "step": 11588
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012936386197346273,
+      "loss": 0.5899,
+      "step": 11589
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001293533452659344,
+      "loss": 0.6973,
+      "step": 11590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012934282820313274,
+      "loss": 0.7097,
+      "step": 11591
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012933231078518504,
+      "loss": 0.6649,
+      "step": 11592
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001293217930122186,
+      "loss": 0.725,
+      "step": 11593
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012931127488436073,
+      "loss": 0.689,
+      "step": 11594
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012930075640173874,
+      "loss": 0.6003,
+      "step": 11595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001292902375644799,
+      "loss": 0.6283,
+      "step": 11596
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012927971837271154,
+      "loss": 0.6426,
+      "step": 11597
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012926919882656098,
+      "loss": 0.6043,
+      "step": 11598
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001292586789261556,
+      "loss": 0.5017,
+      "step": 11599
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012924815867162263,
+      "loss": 0.5994,
+      "step": 11600
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 0.625977098941803,
+      "eval_runtime": 845.1258,
+      "eval_samples_per_second": 28.469,
+      "eval_steps_per_second": 14.235,
+      "step": 11600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012923763806308943,
+      "loss": 0.5533,
+      "step": 11601
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012922711710068339,
+      "loss": 0.6431,
+      "step": 11602
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012921659578453177,
+      "loss": 0.5781,
+      "step": 11603
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012920607411476197,
+      "loss": 0.686,
+      "step": 11604
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001291955520915013,
+      "loss": 0.6166,
+      "step": 11605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012918502971487716,
+      "loss": 0.634,
+      "step": 11606
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012917450698501684,
+      "loss": 0.639,
+      "step": 11607
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012916398390204776,
+      "loss": 0.6864,
+      "step": 11608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012915346046609723,
+      "loss": 0.5509,
+      "step": 11609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012914293667729265,
+      "loss": 0.7335,
+      "step": 11610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001291324125357614,
+      "loss": 0.6161,
+      "step": 11611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012912188804163084,
+      "loss": 0.6815,
+      "step": 11612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001291113631950284,
+      "loss": 0.6279,
+      "step": 11613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012910083799608137,
+      "loss": 0.5692,
+      "step": 11614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012909031244491721,
+      "loss": 0.7074,
+      "step": 11615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001290797865416633,
+      "loss": 0.5977,
+      "step": 11616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012906926028644704,
+      "loss": 0.6098,
+      "step": 11617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012905873367939584,
+      "loss": 0.7594,
+      "step": 11618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012904820672063707,
+      "loss": 0.786,
+      "step": 11619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012903767941029818,
+      "loss": 0.7231,
+      "step": 11620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012902715174850657,
+      "loss": 0.6286,
+      "step": 11621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012901662373538967,
+      "loss": 0.6796,
+      "step": 11622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012900609537107492,
+      "loss": 0.6456,
+      "step": 11623
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012899556665568968,
+      "loss": 0.7094,
+      "step": 11624
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012898503758936145,
+      "loss": 0.5757,
+      "step": 11625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012897450817221765,
+      "loss": 0.6447,
+      "step": 11626
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012896397840438573,
+      "loss": 0.6164,
+      "step": 11627
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012895344828599308,
+      "loss": 0.6478,
+      "step": 11628
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012894291781716724,
+      "loss": 0.6347,
+      "step": 11629
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001289323869980356,
+      "loss": 0.6303,
+      "step": 11630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001289218558287256,
+      "loss": 0.5964,
+      "step": 11631
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012891132430936474,
+      "loss": 0.7391,
+      "step": 11632
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012890079244008048,
+      "loss": 0.6849,
+      "step": 11633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012889026022100031,
+      "loss": 0.615,
+      "step": 11634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012887972765225172,
+      "loss": 0.6069,
+      "step": 11635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001288691947339621,
+      "loss": 0.5857,
+      "step": 11636
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012885866146625902,
+      "loss": 0.6011,
+      "step": 11637
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012884812784926995,
+      "loss": 0.6984,
+      "step": 11638
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012883759388312234,
+      "loss": 0.6032,
+      "step": 11639
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001288270595679437,
+      "loss": 0.6697,
+      "step": 11640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001288165249038616,
+      "loss": 0.5896,
+      "step": 11641
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012880598989100346,
+      "loss": 0.7559,
+      "step": 11642
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001287954545294968,
+      "loss": 0.627,
+      "step": 11643
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001287849188194692,
+      "loss": 0.7209,
+      "step": 11644
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012877438276104807,
+      "loss": 0.6647,
+      "step": 11645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012876384635436103,
+      "loss": 0.6493,
+      "step": 11646
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012875330959953555,
+      "loss": 0.6437,
+      "step": 11647
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001287427724966992,
+      "loss": 0.5756,
+      "step": 11648
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012873223504597948,
+      "loss": 0.5591,
+      "step": 11649
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012872169724750395,
+      "loss": 0.6704,
+      "step": 11650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012871115910140007,
+      "loss": 0.6054,
+      "step": 11651
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012870062060779553,
+      "loss": 0.625,
+      "step": 11652
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012869008176681778,
+      "loss": 0.6321,
+      "step": 11653
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012867954257859442,
+      "loss": 0.6671,
+      "step": 11654
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000128669003043253,
+      "loss": 0.6854,
+      "step": 11655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012865846316092105,
+      "loss": 0.5638,
+      "step": 11656
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012864792293172618,
+      "loss": 0.6368,
+      "step": 11657
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012863738235579595,
+      "loss": 0.6041,
+      "step": 11658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001286268414332579,
+      "loss": 0.7248,
+      "step": 11659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001286163001642397,
+      "loss": 0.6086,
+      "step": 11660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012860575854886882,
+      "loss": 0.6643,
+      "step": 11661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012859521658727295,
+      "loss": 0.5846,
+      "step": 11662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012858467427957957,
+      "loss": 0.582,
+      "step": 11663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012857413162591642,
+      "loss": 0.5627,
+      "step": 11664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012856358862641104,
+      "loss": 0.6788,
+      "step": 11665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012855304528119093,
+      "loss": 0.5588,
+      "step": 11666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012854250159038388,
+      "loss": 0.6507,
+      "step": 11667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012853195755411738,
+      "loss": 0.5834,
+      "step": 11668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012852141317251908,
+      "loss": 0.6264,
+      "step": 11669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001285108684457166,
+      "loss": 0.5051,
+      "step": 11670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012850032337383756,
+      "loss": 0.61,
+      "step": 11671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012848977795700965,
+      "loss": 0.5761,
+      "step": 11672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012847923219536043,
+      "loss": 0.6,
+      "step": 11673
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012846868608901757,
+      "loss": 0.5512,
+      "step": 11674
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012845813963810872,
+      "loss": 0.6123,
+      "step": 11675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001284475928427615,
+      "loss": 0.5972,
+      "step": 11676
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012843704570310356,
+      "loss": 0.6249,
+      "step": 11677
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012842649821926262,
+      "loss": 0.6327,
+      "step": 11678
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012841595039136627,
+      "loss": 0.5815,
+      "step": 11679
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001284054022195422,
+      "loss": 0.6027,
+      "step": 11680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001283948537039181,
+      "loss": 0.5887,
+      "step": 11681
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012838430484462157,
+      "loss": 0.6316,
+      "step": 11682
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012837375564178033,
+      "loss": 0.6703,
+      "step": 11683
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001283632060955221,
+      "loss": 0.5896,
+      "step": 11684
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012835265620597452,
+      "loss": 0.6264,
+      "step": 11685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012834210597326528,
+      "loss": 0.6439,
+      "step": 11686
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012833155539752208,
+      "loss": 0.6029,
+      "step": 11687
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012832100447887263,
+      "loss": 0.702,
+      "step": 11688
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001283104532174446,
+      "loss": 0.5852,
+      "step": 11689
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012829990161336574,
+      "loss": 0.7052,
+      "step": 11690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012828934966676374,
+      "loss": 0.5876,
+      "step": 11691
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012827879737776628,
+      "loss": 0.6967,
+      "step": 11692
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001282682447465011,
+      "loss": 0.587,
+      "step": 11693
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012825769177309595,
+      "loss": 0.5877,
+      "step": 11694
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012824713845767852,
+      "loss": 0.7481,
+      "step": 11695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012823658480037658,
+      "loss": 0.5849,
+      "step": 11696
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001282260308013178,
+      "loss": 0.5948,
+      "step": 11697
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012821547646063,
+      "loss": 0.7382,
+      "step": 11698
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012820492177844086,
+      "loss": 0.6981,
+      "step": 11699
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012819436675487813,
+      "loss": 0.6492,
+      "step": 11700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012818381139006958,
+      "loss": 0.5744,
+      "step": 11701
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012817325568414297,
+      "loss": 0.6555,
+      "step": 11702
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012816269963722605,
+      "loss": 0.6674,
+      "step": 11703
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012815214324944662,
+      "loss": 0.6628,
+      "step": 11704
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012814158652093235,
+      "loss": 0.6957,
+      "step": 11705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012813102945181116,
+      "loss": 0.5857,
+      "step": 11706
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012812047204221067,
+      "loss": 0.6307,
+      "step": 11707
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012810991429225877,
+      "loss": 0.7212,
+      "step": 11708
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012809935620208316,
+      "loss": 0.6234,
+      "step": 11709
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012808879777181172,
+      "loss": 0.6508,
+      "step": 11710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012807823900157214,
+      "loss": 0.6476,
+      "step": 11711
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001280676798914923,
+      "loss": 0.6009,
+      "step": 11712
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012805712044169999,
+      "loss": 0.6514,
+      "step": 11713
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000128046560652323,
+      "loss": 0.564,
+      "step": 11714
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012803600052348907,
+      "loss": 0.7176,
+      "step": 11715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012802544005532612,
+      "loss": 0.6356,
+      "step": 11716
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012801487924796192,
+      "loss": 0.5956,
+      "step": 11717
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012800431810152432,
+      "loss": 0.4823,
+      "step": 11718
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001279937566161411,
+      "loss": 0.5061,
+      "step": 11719
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001279831947919401,
+      "loss": 0.635,
+      "step": 11720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012797263262904918,
+      "loss": 0.5406,
+      "step": 11721
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001279620701275961,
+      "loss": 0.6158,
+      "step": 11722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012795150728770884,
+      "loss": 0.6209,
+      "step": 11723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012794094410951515,
+      "loss": 0.5964,
+      "step": 11724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012793038059314286,
+      "loss": 0.7087,
+      "step": 11725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001279198167387199,
+      "loss": 0.6442,
+      "step": 11726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012790925254637406,
+      "loss": 0.5706,
+      "step": 11727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012789868801623325,
+      "loss": 0.6567,
+      "step": 11728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012788812314842531,
+      "loss": 0.6043,
+      "step": 11729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001278775579430781,
+      "loss": 0.7412,
+      "step": 11730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012786699240031955,
+      "loss": 0.6014,
+      "step": 11731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012785642652027746,
+      "loss": 0.5672,
+      "step": 11732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012784586030307974,
+      "loss": 0.5831,
+      "step": 11733
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012783529374885434,
+      "loss": 0.5646,
+      "step": 11734
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012782472685772905,
+      "loss": 0.6954,
+      "step": 11735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012781415962983186,
+      "loss": 0.5923,
+      "step": 11736
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001278035920652906,
+      "loss": 0.6253,
+      "step": 11737
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012779302416423316,
+      "loss": 0.6195,
+      "step": 11738
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012778245592678754,
+      "loss": 0.603,
+      "step": 11739
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012777188735308155,
+      "loss": 0.6344,
+      "step": 11740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012776131844324315,
+      "loss": 0.6088,
+      "step": 11741
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012775074919740026,
+      "loss": 0.6045,
+      "step": 11742
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001277401796156808,
+      "loss": 0.7368,
+      "step": 11743
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012772960969821272,
+      "loss": 0.6131,
+      "step": 11744
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012771903944512389,
+      "loss": 0.5964,
+      "step": 11745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001277084688565423,
+      "loss": 0.5088,
+      "step": 11746
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012769789793259587,
+      "loss": 0.5719,
+      "step": 11747
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012768732667341255,
+      "loss": 0.6618,
+      "step": 11748
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012767675507912028,
+      "loss": 0.6626,
+      "step": 11749
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000127666183149847,
+      "loss": 0.6252,
+      "step": 11750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001276556108857207,
+      "loss": 0.6678,
+      "step": 11751
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012764503828686932,
+      "loss": 0.6235,
+      "step": 11752
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012763446535342081,
+      "loss": 0.603,
+      "step": 11753
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012762389208550317,
+      "loss": 0.6063,
+      "step": 11754
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012761331848324433,
+      "loss": 0.6149,
+      "step": 11755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012760274454677231,
+      "loss": 0.6775,
+      "step": 11756
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012759217027621505,
+      "loss": 0.708,
+      "step": 11757
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012758159567170056,
+      "loss": 0.7341,
+      "step": 11758
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012757102073335683,
+      "loss": 0.6403,
+      "step": 11759
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012756044546131183,
+      "loss": 0.6718,
+      "step": 11760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012754986985569357,
+      "loss": 0.5658,
+      "step": 11761
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012753929391663007,
+      "loss": 0.5794,
+      "step": 11762
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012752871764424927,
+      "loss": 0.6267,
+      "step": 11763
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012751814103867924,
+      "loss": 0.6805,
+      "step": 11764
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012750756410004802,
+      "loss": 0.6819,
+      "step": 11765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001274969868284835,
+      "loss": 0.6389,
+      "step": 11766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012748640922411383,
+      "loss": 0.6588,
+      "step": 11767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012747583128706698,
+      "loss": 0.6359,
+      "step": 11768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012746525301747096,
+      "loss": 0.6415,
+      "step": 11769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012745467441545384,
+      "loss": 0.562,
+      "step": 11770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012744409548114363,
+      "loss": 0.6395,
+      "step": 11771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012743351621466838,
+      "loss": 0.729,
+      "step": 11772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012742293661615616,
+      "loss": 0.575,
+      "step": 11773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000127412356685735,
+      "loss": 0.6139,
+      "step": 11774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012740177642353294,
+      "loss": 0.6012,
+      "step": 11775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000127391195829678,
+      "loss": 0.5416,
+      "step": 11776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012738061490429834,
+      "loss": 0.6807,
+      "step": 11777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001273700336475219,
+      "loss": 0.6184,
+      "step": 11778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012735945205947689,
+      "loss": 0.6882,
+      "step": 11779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012734887014029126,
+      "loss": 0.5809,
+      "step": 11780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012733828789009315,
+      "loss": 0.6143,
+      "step": 11781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012732770530901063,
+      "loss": 0.5634,
+      "step": 11782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012731712239717177,
+      "loss": 0.6376,
+      "step": 11783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012730653915470468,
+      "loss": 0.8312,
+      "step": 11784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012729595558173744,
+      "loss": 0.5902,
+      "step": 11785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001272853716783981,
+      "loss": 0.5872,
+      "step": 11786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001272747874448149,
+      "loss": 0.6306,
+      "step": 11787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001272642028811158,
+      "loss": 0.539,
+      "step": 11788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012725361798742894,
+      "loss": 0.6256,
+      "step": 11789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001272430327638825,
+      "loss": 0.5282,
+      "step": 11790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012723244721060453,
+      "loss": 0.6855,
+      "step": 11791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012722186132772316,
+      "loss": 0.5884,
+      "step": 11792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012721127511536654,
+      "loss": 0.6142,
+      "step": 11793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012720068857366277,
+      "loss": 0.6421,
+      "step": 11794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012719010170274001,
+      "loss": 0.6708,
+      "step": 11795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012717951450272636,
+      "loss": 0.6356,
+      "step": 11796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012716892697375002,
+      "loss": 0.5996,
+      "step": 11797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001271583391159391,
+      "loss": 0.7152,
+      "step": 11798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001271477509294217,
+      "loss": 0.6229,
+      "step": 11799
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012713716241432606,
+      "loss": 0.6133,
+      "step": 11800
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 0.6201381087303162,
+      "eval_runtime": 855.5515,
+      "eval_samples_per_second": 28.122,
+      "eval_steps_per_second": 14.061,
+      "step": 11800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012712657357078026,
+      "loss": 0.6682,
+      "step": 11801
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001271159843989125,
+      "loss": 0.6554,
+      "step": 11802
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012710539489885095,
+      "loss": 0.6375,
+      "step": 11803
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012709480507072378,
+      "loss": 0.522,
+      "step": 11804
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012708421491465914,
+      "loss": 0.5729,
+      "step": 11805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012707362443078522,
+      "loss": 0.6785,
+      "step": 11806
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001270630336192302,
+      "loss": 0.6401,
+      "step": 11807
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012705244248012223,
+      "loss": 0.5239,
+      "step": 11808
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012704185101358953,
+      "loss": 0.6145,
+      "step": 11809
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012703125921976032,
+      "loss": 0.6104,
+      "step": 11810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012702066709876275,
+      "loss": 0.6834,
+      "step": 11811
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012701007465072502,
+      "loss": 0.6125,
+      "step": 11812
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012699948187577537,
+      "loss": 0.6663,
+      "step": 11813
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000126988888774042,
+      "loss": 0.6569,
+      "step": 11814
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012697829534565307,
+      "loss": 0.4991,
+      "step": 11815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001269677015907369,
+      "loss": 0.6604,
+      "step": 11816
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012695710750942158,
+      "loss": 0.6275,
+      "step": 11817
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012694651310183542,
+      "loss": 0.704,
+      "step": 11818
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012693591836810666,
+      "loss": 0.644,
+      "step": 11819
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012692532330836346,
+      "loss": 0.6825,
+      "step": 11820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001269147279227341,
+      "loss": 0.5614,
+      "step": 11821
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001269041322113468,
+      "loss": 0.6022,
+      "step": 11822
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012689353617432984,
+      "loss": 0.6344,
+      "step": 11823
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012688293981181144,
+      "loss": 0.5861,
+      "step": 11824
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012687234312391986,
+      "loss": 0.6762,
+      "step": 11825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012686174611078338,
+      "loss": 0.6856,
+      "step": 11826
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012685114877253014,
+      "loss": 0.6706,
+      "step": 11827
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012684055110928857,
+      "loss": 0.63,
+      "step": 11828
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012682995312118684,
+      "loss": 0.6337,
+      "step": 11829
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012681935480835324,
+      "loss": 0.5112,
+      "step": 11830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012680875617091604,
+      "loss": 0.6969,
+      "step": 11831
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012679815720900353,
+      "loss": 0.5941,
+      "step": 11832
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012678755792274398,
+      "loss": 0.6199,
+      "step": 11833
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001267769583122657,
+      "loss": 0.5727,
+      "step": 11834
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012676635837769693,
+      "loss": 0.5298,
+      "step": 11835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012675575811916604,
+      "loss": 0.6886,
+      "step": 11836
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012674515753680128,
+      "loss": 0.6039,
+      "step": 11837
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012673455663073094,
+      "loss": 0.5519,
+      "step": 11838
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012672395540108336,
+      "loss": 0.5896,
+      "step": 11839
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012671335384798686,
+      "loss": 0.6864,
+      "step": 11840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001267027519715697,
+      "loss": 0.6686,
+      "step": 11841
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012669214977196021,
+      "loss": 0.7217,
+      "step": 11842
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012668154724928678,
+      "loss": 0.681,
+      "step": 11843
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012667094440367766,
+      "loss": 0.5605,
+      "step": 11844
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001266603412352612,
+      "loss": 0.6423,
+      "step": 11845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012664973774416575,
+      "loss": 0.4969,
+      "step": 11846
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012663913393051962,
+      "loss": 0.698,
+      "step": 11847
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012662852979445122,
+      "loss": 0.5479,
+      "step": 11848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001266179253360888,
+      "loss": 0.5324,
+      "step": 11849
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012660732055556074,
+      "loss": 0.6484,
+      "step": 11850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012659671545299546,
+      "loss": 0.7175,
+      "step": 11851
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012658611002852124,
+      "loss": 0.708,
+      "step": 11852
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012657550428226644,
+      "loss": 0.5909,
+      "step": 11853
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001265648982143595,
+      "loss": 0.6749,
+      "step": 11854
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001265542918249287,
+      "loss": 0.6469,
+      "step": 11855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012654368511410245,
+      "loss": 0.6713,
+      "step": 11856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012653307808200914,
+      "loss": 0.6258,
+      "step": 11857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012652247072877714,
+      "loss": 0.6278,
+      "step": 11858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012651186305453484,
+      "loss": 0.6213,
+      "step": 11859
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001265012550594106,
+      "loss": 0.6391,
+      "step": 11860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012649064674353285,
+      "loss": 0.5395,
+      "step": 11861
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012648003810702998,
+      "loss": 0.5962,
+      "step": 11862
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012646942915003036,
+      "loss": 0.5523,
+      "step": 11863
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012645881987266238,
+      "loss": 0.5626,
+      "step": 11864
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001264482102750545,
+      "loss": 0.5768,
+      "step": 11865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012643760035733516,
+      "loss": 0.5549,
+      "step": 11866
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001264269901196327,
+      "loss": 0.6206,
+      "step": 11867
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012641637956207555,
+      "loss": 0.7204,
+      "step": 11868
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012640576868479218,
+      "loss": 0.577,
+      "step": 11869
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012639515748791096,
+      "loss": 0.7766,
+      "step": 11870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012638454597156036,
+      "loss": 0.6625,
+      "step": 11871
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001263739341358688,
+      "loss": 0.6213,
+      "step": 11872
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001263633219809647,
+      "loss": 0.6435,
+      "step": 11873
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012635270950697654,
+      "loss": 0.6757,
+      "step": 11874
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012634209671403277,
+      "loss": 0.6127,
+      "step": 11875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012633148360226179,
+      "loss": 0.6273,
+      "step": 11876
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001263208701717921,
+      "loss": 0.6009,
+      "step": 11877
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012631025642275212,
+      "loss": 0.6339,
+      "step": 11878
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012629964235527037,
+      "loss": 0.5839,
+      "step": 11879
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012628902796947526,
+      "loss": 0.5718,
+      "step": 11880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012627841326549527,
+      "loss": 0.5649,
+      "step": 11881
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012626779824345888,
+      "loss": 0.5272,
+      "step": 11882
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001262571829034946,
+      "loss": 0.7037,
+      "step": 11883
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012624656724573083,
+      "loss": 0.532,
+      "step": 11884
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012623595127029614,
+      "loss": 0.6628,
+      "step": 11885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012622533497731896,
+      "loss": 0.7012,
+      "step": 11886
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012621471836692785,
+      "loss": 0.5853,
+      "step": 11887
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012620410143925125,
+      "loss": 0.688,
+      "step": 11888
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012619348419441764,
+      "loss": 0.6186,
+      "step": 11889
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012618286663255556,
+      "loss": 0.6142,
+      "step": 11890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012617224875379352,
+      "loss": 0.5961,
+      "step": 11891
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012616163055826005,
+      "loss": 0.5851,
+      "step": 11892
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012615101204608363,
+      "loss": 0.6786,
+      "step": 11893
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012614039321739276,
+      "loss": 0.5971,
+      "step": 11894
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012612977407231606,
+      "loss": 0.7272,
+      "step": 11895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012611915461098193,
+      "loss": 0.7683,
+      "step": 11896
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000126108534833519,
+      "loss": 0.7599,
+      "step": 11897
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012609791474005574,
+      "loss": 0.6882,
+      "step": 11898
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001260872943307207,
+      "loss": 0.6152,
+      "step": 11899
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012607667360564251,
+      "loss": 0.5467,
+      "step": 11900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012606605256494958,
+      "loss": 0.6553,
+      "step": 11901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012605543120877054,
+      "loss": 0.6865,
+      "step": 11902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012604480953723395,
+      "loss": 0.6375,
+      "step": 11903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012603418755046829,
+      "loss": 0.5428,
+      "step": 11904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012602356524860225,
+      "loss": 0.5658,
+      "step": 11905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012601294263176426,
+      "loss": 0.6673,
+      "step": 11906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012600231970008298,
+      "loss": 0.6209,
+      "step": 11907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012599169645368698,
+      "loss": 0.6114,
+      "step": 11908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012598107289270478,
+      "loss": 0.5839,
+      "step": 11909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012597044901726498,
+      "loss": 0.6081,
+      "step": 11910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012595982482749619,
+      "loss": 0.5573,
+      "step": 11911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012594920032352698,
+      "loss": 0.6433,
+      "step": 11912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012593857550548596,
+      "loss": 0.545,
+      "step": 11913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001259279503735017,
+      "loss": 0.6805,
+      "step": 11914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001259173249277028,
+      "loss": 0.6484,
+      "step": 11915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001259066991682179,
+      "loss": 0.619,
+      "step": 11916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012589607309517557,
+      "loss": 0.5523,
+      "step": 11917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012588544670870443,
+      "loss": 0.6111,
+      "step": 11918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012587482000893306,
+      "loss": 0.6498,
+      "step": 11919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001258641929959902,
+      "loss": 0.6465,
+      "step": 11920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012585356567000432,
+      "loss": 0.6674,
+      "step": 11921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012584293803110413,
+      "loss": 0.6107,
+      "step": 11922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012583231007941823,
+      "loss": 0.6186,
+      "step": 11923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001258216818150753,
+      "loss": 0.6281,
+      "step": 11924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012581105323820392,
+      "loss": 0.6652,
+      "step": 11925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001258004243489328,
+      "loss": 0.64,
+      "step": 11926
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001257897951473905,
+      "loss": 0.6551,
+      "step": 11927
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012577916563370572,
+      "loss": 0.6252,
+      "step": 11928
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012576853580800707,
+      "loss": 0.5853,
+      "step": 11929
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012575790567042328,
+      "loss": 0.6206,
+      "step": 11930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012574727522108296,
+      "loss": 0.5971,
+      "step": 11931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012573664446011477,
+      "loss": 0.6503,
+      "step": 11932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001257260133876474,
+      "loss": 0.6198,
+      "step": 11933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001257153820038095,
+      "loss": 0.6381,
+      "step": 11934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012570475030872978,
+      "loss": 0.5921,
+      "step": 11935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012569411830253688,
+      "loss": 0.7033,
+      "step": 11936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012568348598535948,
+      "loss": 0.6358,
+      "step": 11937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012567285335732633,
+      "loss": 0.4866,
+      "step": 11938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012566222041856607,
+      "loss": 0.6699,
+      "step": 11939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012565158716920736,
+      "loss": 0.5805,
+      "step": 11940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012564095360937896,
+      "loss": 0.6639,
+      "step": 11941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001256303197392095,
+      "loss": 0.5917,
+      "step": 11942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012561968555882782,
+      "loss": 0.6131,
+      "step": 11943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001256090510683625,
+      "loss": 0.659,
+      "step": 11944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012559841626794227,
+      "loss": 0.7402,
+      "step": 11945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012558778115769593,
+      "loss": 0.6691,
+      "step": 11946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001255771457377521,
+      "loss": 0.6218,
+      "step": 11947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012556651000823955,
+      "loss": 0.5883,
+      "step": 11948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012555587396928704,
+      "loss": 0.7763,
+      "step": 11949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012554523762102322,
+      "loss": 0.6724,
+      "step": 11950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001255346009635769,
+      "loss": 0.7635,
+      "step": 11951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012552396399707677,
+      "loss": 0.6494,
+      "step": 11952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012551332672165158,
+      "loss": 0.5703,
+      "step": 11953
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012550268913743016,
+      "loss": 0.6403,
+      "step": 11954
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001254920512445411,
+      "loss": 0.6557,
+      "step": 11955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012548141304311332,
+      "loss": 0.6157,
+      "step": 11956
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012547077453327544,
+      "loss": 0.5837,
+      "step": 11957
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012546013571515633,
+      "loss": 0.6942,
+      "step": 11958
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012544949658888468,
+      "loss": 0.6532,
+      "step": 11959
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012543885715458926,
+      "loss": 0.6142,
+      "step": 11960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012542821741239895,
+      "loss": 0.6334,
+      "step": 11961
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012541757736244237,
+      "loss": 0.5638,
+      "step": 11962
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001254069370048484,
+      "loss": 0.6278,
+      "step": 11963
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012539629633974582,
+      "loss": 0.5811,
+      "step": 11964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012538565536726335,
+      "loss": 0.5942,
+      "step": 11965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001253750140875299,
+      "loss": 0.5912,
+      "step": 11966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012536437250067415,
+      "loss": 0.5668,
+      "step": 11967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001253537306068249,
+      "loss": 0.5121,
+      "step": 11968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012534308840611105,
+      "loss": 0.7151,
+      "step": 11969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012533244589866136,
+      "loss": 0.5875,
+      "step": 11970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012532180308460458,
+      "loss": 0.6231,
+      "step": 11971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012531115996406963,
+      "loss": 0.5663,
+      "step": 11972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001253005165371852,
+      "loss": 0.7118,
+      "step": 11973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012528987280408026,
+      "loss": 0.5964,
+      "step": 11974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012527922876488353,
+      "loss": 0.6339,
+      "step": 11975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012526858441972387,
+      "loss": 0.6001,
+      "step": 11976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012525793976873008,
+      "loss": 0.677,
+      "step": 11977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012524729481203103,
+      "loss": 0.6094,
+      "step": 11978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012523664954975557,
+      "loss": 0.5969,
+      "step": 11979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012522600398203251,
+      "loss": 0.6851,
+      "step": 11980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012521535810899073,
+      "loss": 0.6595,
+      "step": 11981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012520471193075904,
+      "loss": 0.5682,
+      "step": 11982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012519406544746632,
+      "loss": 0.6583,
+      "step": 11983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012518341865924143,
+      "loss": 0.5764,
+      "step": 11984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012517277156621324,
+      "loss": 0.5931,
+      "step": 11985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012516212416851058,
+      "loss": 0.6247,
+      "step": 11986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012515147646626235,
+      "loss": 0.625,
+      "step": 11987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001251408284595974,
+      "loss": 0.6841,
+      "step": 11988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012513018014864463,
+      "loss": 0.6495,
+      "step": 11989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012511953153353293,
+      "loss": 0.6404,
+      "step": 11990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012510888261439113,
+      "loss": 0.6964,
+      "step": 11991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012509823339134819,
+      "loss": 0.5784,
+      "step": 11992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001250875838645329,
+      "loss": 0.6223,
+      "step": 11993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012507693403407427,
+      "loss": 0.6672,
+      "step": 11994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012506628390010117,
+      "loss": 0.5308,
+      "step": 11995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001250556334627424,
+      "loss": 0.5331,
+      "step": 11996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012504498272212703,
+      "loss": 0.5174,
+      "step": 11997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012503433167838378,
+      "loss": 0.6877,
+      "step": 11998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012502368033164176,
+      "loss": 0.6103,
+      "step": 11999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012501302868202975,
+      "loss": 0.6707,
+      "step": 12000
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 0.6138113737106323,
+      "eval_runtime": 844.262,
+      "eval_samples_per_second": 28.498,
+      "eval_steps_per_second": 14.249,
+      "step": 12000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012500237672967673,
+      "loss": 0.6065,
+      "step": 12001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012499172447471158,
+      "loss": 0.6357,
+      "step": 12002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012498107191726327,
+      "loss": 0.5459,
+      "step": 12003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001249704190574607,
+      "loss": 0.5087,
+      "step": 12004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012495976589543286,
+      "loss": 0.5447,
+      "step": 12005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012494911243130863,
+      "loss": 0.5493,
+      "step": 12006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012493845866521697,
+      "loss": 0.7247,
+      "step": 12007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012492780459728685,
+      "loss": 0.5404,
+      "step": 12008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012491715022764716,
+      "loss": 0.5848,
+      "step": 12009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012490649555642693,
+      "loss": 0.5461,
+      "step": 12010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012489584058375503,
+      "loss": 0.6583,
+      "step": 12011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012488518530976052,
+      "loss": 0.5724,
+      "step": 12012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012487452973457229,
+      "loss": 0.7039,
+      "step": 12013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012486387385831935,
+      "loss": 0.5703,
+      "step": 12014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012485321768113065,
+      "loss": 0.6205,
+      "step": 12015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012484256120313513,
+      "loss": 0.5861,
+      "step": 12016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012483190442446186,
+      "loss": 0.6034,
+      "step": 12017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012482124734523978,
+      "loss": 0.7239,
+      "step": 12018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001248105899655978,
+      "loss": 0.6697,
+      "step": 12019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012479993228566502,
+      "loss": 0.6875,
+      "step": 12020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012478927430557038,
+      "loss": 0.6061,
+      "step": 12021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001247786160254429,
+      "loss": 0.6638,
+      "step": 12022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012476795744541158,
+      "loss": 0.6154,
+      "step": 12023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012475729856560537,
+      "loss": 0.631,
+      "step": 12024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012474663938615333,
+      "loss": 0.631,
+      "step": 12025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012473597990718448,
+      "loss": 0.6675,
+      "step": 12026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001247253201288278,
+      "loss": 0.6078,
+      "step": 12027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012471466005121236,
+      "loss": 0.4915,
+      "step": 12028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012470399967446708,
+      "loss": 0.6351,
+      "step": 12029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012469333899872112,
+      "loss": 0.5585,
+      "step": 12030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012468267802410342,
+      "loss": 0.6812,
+      "step": 12031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012467201675074306,
+      "loss": 0.6526,
+      "step": 12032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012466135517876902,
+      "loss": 0.4916,
+      "step": 12033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012465069330831041,
+      "loss": 0.5934,
+      "step": 12034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012464003113949617,
+      "loss": 0.6261,
+      "step": 12035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012462936867245548,
+      "loss": 0.6185,
+      "step": 12036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012461870590731733,
+      "loss": 0.6623,
+      "step": 12037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012460804284421075,
+      "loss": 0.6359,
+      "step": 12038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012459737948326485,
+      "loss": 0.6154,
+      "step": 12039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012458671582460862,
+      "loss": 0.6238,
+      "step": 12040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001245760518683712,
+      "loss": 0.6632,
+      "step": 12041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012456538761468163,
+      "loss": 0.5787,
+      "step": 12042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012455472306366897,
+      "loss": 0.6057,
+      "step": 12043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012454405821546232,
+      "loss": 0.636,
+      "step": 12044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012453339307019073,
+      "loss": 0.6973,
+      "step": 12045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012452272762798335,
+      "loss": 0.6201,
+      "step": 12046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012451206188896915,
+      "loss": 0.6802,
+      "step": 12047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012450139585327733,
+      "loss": 0.629,
+      "step": 12048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012449072952103695,
+      "loss": 0.5766,
+      "step": 12049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001244800628923771,
+      "loss": 0.632,
+      "step": 12050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001244693959674269,
+      "loss": 0.6971,
+      "step": 12051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001244587287463154,
+      "loss": 0.6498,
+      "step": 12052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012444806122917178,
+      "loss": 0.675,
+      "step": 12053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012443739341612512,
+      "loss": 0.6033,
+      "step": 12054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012442672530730452,
+      "loss": 0.7198,
+      "step": 12055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012441605690283915,
+      "loss": 0.5594,
+      "step": 12056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012440538820285808,
+      "loss": 0.6374,
+      "step": 12057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001243947192074905,
+      "loss": 0.5583,
+      "step": 12058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012438404991686546,
+      "loss": 0.4973,
+      "step": 12059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001243733803311121,
+      "loss": 0.6627,
+      "step": 12060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012436271045035966,
+      "loss": 0.595,
+      "step": 12061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012435204027473715,
+      "loss": 0.6842,
+      "step": 12062
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012434136980437382,
+      "loss": 0.7288,
+      "step": 12063
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012433069903939876,
+      "loss": 0.6304,
+      "step": 12064
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012432002797994112,
+      "loss": 0.5744,
+      "step": 12065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001243093566261301,
+      "loss": 0.5865,
+      "step": 12066
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012429868497809482,
+      "loss": 0.57,
+      "step": 12067
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012428801303596444,
+      "loss": 0.6826,
+      "step": 12068
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012427734079986817,
+      "loss": 0.6448,
+      "step": 12069
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012426666826993512,
+      "loss": 0.6945,
+      "step": 12070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012425599544629452,
+      "loss": 0.5544,
+      "step": 12071
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001242453223290755,
+      "loss": 0.8091,
+      "step": 12072
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012423464891840727,
+      "loss": 0.6304,
+      "step": 12073
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012422397521441898,
+      "loss": 0.6994,
+      "step": 12074
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012421330121723984,
+      "loss": 0.5674,
+      "step": 12075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012420262692699907,
+      "loss": 0.6023,
+      "step": 12076
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012419195234382585,
+      "loss": 0.6119,
+      "step": 12077
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001241812774678493,
+      "loss": 0.6522,
+      "step": 12078
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012417060229919874,
+      "loss": 0.7106,
+      "step": 12079
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012415992683800328,
+      "loss": 0.6446,
+      "step": 12080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001241492510843922,
+      "loss": 0.6066,
+      "step": 12081
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001241385750384947,
+      "loss": 0.6082,
+      "step": 12082
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012412789870043994,
+      "loss": 0.702,
+      "step": 12083
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001241172220703572,
+      "loss": 0.6092,
+      "step": 12084
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012410654514837565,
+      "loss": 0.5895,
+      "step": 12085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012409586793462454,
+      "loss": 0.5051,
+      "step": 12086
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012408519042923313,
+      "loss": 0.6358,
+      "step": 12087
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012407451263233066,
+      "loss": 0.5467,
+      "step": 12088
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012406383454404628,
+      "loss": 0.7304,
+      "step": 12089
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012405315616450936,
+      "loss": 0.6036,
+      "step": 12090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012404247749384897,
+      "loss": 0.5697,
+      "step": 12091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012403179853219453,
+      "loss": 0.5347,
+      "step": 12092
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012402111927967518,
+      "loss": 0.6089,
+      "step": 12093
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012401043973642024,
+      "loss": 0.7253,
+      "step": 12094
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012399975990255894,
+      "loss": 0.6619,
+      "step": 12095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001239890797782205,
+      "loss": 0.7444,
+      "step": 12096
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001239783993635343,
+      "loss": 0.6183,
+      "step": 12097
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001239677186586295,
+      "loss": 0.5806,
+      "step": 12098
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001239570376636354,
+      "loss": 0.5948,
+      "step": 12099
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001239463563786813,
+      "loss": 0.645,
+      "step": 12100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012393567480389645,
+      "loss": 0.5681,
+      "step": 12101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012392499293941017,
+      "loss": 0.6032,
+      "step": 12102
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012391431078535172,
+      "loss": 0.6676,
+      "step": 12103
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012390362834185033,
+      "loss": 0.5519,
+      "step": 12104
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012389294560903545,
+      "loss": 0.6443,
+      "step": 12105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012388226258703623,
+      "loss": 0.5271,
+      "step": 12106
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000123871579275982,
+      "loss": 0.5699,
+      "step": 12107
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012386089567600214,
+      "loss": 0.5414,
+      "step": 12108
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012385021178722586,
+      "loss": 0.6126,
+      "step": 12109
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012383952760978257,
+      "loss": 0.5578,
+      "step": 12110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012382884314380147,
+      "loss": 0.519,
+      "step": 12111
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012381815838941198,
+      "loss": 0.5193,
+      "step": 12112
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012380747334674334,
+      "loss": 0.6903,
+      "step": 12113
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012379678801592494,
+      "loss": 0.6473,
+      "step": 12114
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012378610239708606,
+      "loss": 0.5127,
+      "step": 12115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012377541649035605,
+      "loss": 0.7054,
+      "step": 12116
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012376473029586426,
+      "loss": 0.7204,
+      "step": 12117
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012375404381373997,
+      "loss": 0.6132,
+      "step": 12118
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012374335704411262,
+      "loss": 0.6496,
+      "step": 12119
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001237326699871115,
+      "loss": 0.6902,
+      "step": 12120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012372198264286595,
+      "loss": 0.5521,
+      "step": 12121
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012371129501150533,
+      "loss": 0.604,
+      "step": 12122
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000123700607093159,
+      "loss": 0.7035,
+      "step": 12123
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012368991888795633,
+      "loss": 0.7594,
+      "step": 12124
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012367923039602666,
+      "loss": 0.5536,
+      "step": 12125
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012366854161749937,
+      "loss": 0.6106,
+      "step": 12126
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012365785255250383,
+      "loss": 0.6004,
+      "step": 12127
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012364716320116943,
+      "loss": 0.5577,
+      "step": 12128
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012363647356362548,
+      "loss": 0.6307,
+      "step": 12129
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012362578364000145,
+      "loss": 0.528,
+      "step": 12130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012361509343042669,
+      "loss": 0.6302,
+      "step": 12131
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012360440293503056,
+      "loss": 0.7808,
+      "step": 12132
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012359371215394247,
+      "loss": 0.6565,
+      "step": 12133
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001235830210872918,
+      "loss": 0.565,
+      "step": 12134
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012357232973520798,
+      "loss": 0.6028,
+      "step": 12135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001235616380978204,
+      "loss": 0.6617,
+      "step": 12136
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012355094617525844,
+      "loss": 0.5346,
+      "step": 12137
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012354025396765154,
+      "loss": 0.6617,
+      "step": 12138
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012352956147512907,
+      "loss": 0.6424,
+      "step": 12139
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012351886869782052,
+      "loss": 0.553,
+      "step": 12140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012350817563585524,
+      "loss": 0.6407,
+      "step": 12141
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012349748228936264,
+      "loss": 0.6725,
+      "step": 12142
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012348678865847223,
+      "loss": 0.5641,
+      "step": 12143
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012347609474331334,
+      "loss": 0.6204,
+      "step": 12144
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012346540054401547,
+      "loss": 0.5358,
+      "step": 12145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012345470606070803,
+      "loss": 0.554,
+      "step": 12146
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012344401129352046,
+      "loss": 0.7069,
+      "step": 12147
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012343331624258222,
+      "loss": 0.6817,
+      "step": 12148
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012342262090802268,
+      "loss": 0.6662,
+      "step": 12149
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001234119252899714,
+      "loss": 0.6314,
+      "step": 12150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012340122938855777,
+      "loss": 0.6578,
+      "step": 12151
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012339053320391128,
+      "loss": 0.6512,
+      "step": 12152
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012337983673616133,
+      "loss": 0.6752,
+      "step": 12153
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012336913998543743,
+      "loss": 0.5758,
+      "step": 12154
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012335844295186904,
+      "loss": 0.575,
+      "step": 12155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012334774563558563,
+      "loss": 0.5686,
+      "step": 12156
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001233370480367167,
+      "loss": 0.6309,
+      "step": 12157
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012332635015539165,
+      "loss": 0.5468,
+      "step": 12158
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012331565199174,
+      "loss": 0.6765,
+      "step": 12159
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012330495354589128,
+      "loss": 0.5812,
+      "step": 12160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001232942548179749,
+      "loss": 0.588,
+      "step": 12161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001232835558081204,
+      "loss": 0.7214,
+      "step": 12162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012327285651645727,
+      "loss": 0.7007,
+      "step": 12163
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000123262156943115,
+      "loss": 0.6052,
+      "step": 12164
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012325145708822307,
+      "loss": 0.7441,
+      "step": 12165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012324075695191103,
+      "loss": 0.4949,
+      "step": 12166
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001232300565343083,
+      "loss": 0.6303,
+      "step": 12167
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001232193558355445,
+      "loss": 0.6636,
+      "step": 12168
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001232086548557491,
+      "loss": 0.7071,
+      "step": 12169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012319795359505158,
+      "loss": 0.6204,
+      "step": 12170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001231872520535815,
+      "loss": 0.6793,
+      "step": 12171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001231765502314684,
+      "loss": 0.5829,
+      "step": 12172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012316584812884177,
+      "loss": 0.6648,
+      "step": 12173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012315514574583113,
+      "loss": 0.5496,
+      "step": 12174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012314444308256605,
+      "loss": 0.642,
+      "step": 12175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001231337401391761,
+      "loss": 0.5817,
+      "step": 12176
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012312303691579076,
+      "loss": 0.5802,
+      "step": 12177
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012311233341253958,
+      "loss": 0.6262,
+      "step": 12178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012310162962955214,
+      "loss": 0.6406,
+      "step": 12179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012309092556695794,
+      "loss": 0.6423,
+      "step": 12180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012308022122488658,
+      "loss": 0.5957,
+      "step": 12181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012306951660346762,
+      "loss": 0.596,
+      "step": 12182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001230588117028306,
+      "loss": 0.559,
+      "step": 12183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001230481065231051,
+      "loss": 0.6804,
+      "step": 12184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012303740106442066,
+      "loss": 0.6276,
+      "step": 12185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001230266953269069,
+      "loss": 0.6128,
+      "step": 12186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012301598931069337,
+      "loss": 0.6187,
+      "step": 12187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012300528301590962,
+      "loss": 0.7466,
+      "step": 12188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012299457644268527,
+      "loss": 0.6382,
+      "step": 12189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012298386959114986,
+      "loss": 0.5584,
+      "step": 12190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012297316246143308,
+      "loss": 0.5745,
+      "step": 12191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001229624550536644,
+      "loss": 0.5295,
+      "step": 12192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001229517473679735,
+      "loss": 0.623,
+      "step": 12193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012294103940448995,
+      "loss": 0.6031,
+      "step": 12194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012293033116334332,
+      "loss": 0.6572,
+      "step": 12195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012291962264466327,
+      "loss": 0.5845,
+      "step": 12196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001229089138485794,
+      "loss": 0.6177,
+      "step": 12197
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012289820477522128,
+      "loss": 0.5597,
+      "step": 12198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012288749542471855,
+      "loss": 0.5189,
+      "step": 12199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012287678579720083,
+      "loss": 0.6454,
+      "step": 12200
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 0.6077780723571777,
+      "eval_runtime": 843.5156,
+      "eval_samples_per_second": 28.523,
+      "eval_steps_per_second": 14.262,
+      "step": 12200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012286607589279778,
+      "loss": 0.6823,
+      "step": 12201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012285536571163895,
+      "loss": 0.6498,
+      "step": 12202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012284465525385402,
+      "loss": 0.4953,
+      "step": 12203
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012283394451957262,
+      "loss": 0.5662,
+      "step": 12204
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012282323350892437,
+      "loss": 0.5579,
+      "step": 12205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001228125222220389,
+      "loss": 0.5655,
+      "step": 12206
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012280181065904588,
+      "loss": 0.6127,
+      "step": 12207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012279109882007492,
+      "loss": 0.6593,
+      "step": 12208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012278038670525572,
+      "loss": 0.5687,
+      "step": 12209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012276967431471792,
+      "loss": 0.6113,
+      "step": 12210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001227589616485911,
+      "loss": 0.5871,
+      "step": 12211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012274824870700502,
+      "loss": 0.7008,
+      "step": 12212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001227375354900893,
+      "loss": 0.6078,
+      "step": 12213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001227268219979736,
+      "loss": 0.585,
+      "step": 12214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001227161082307876,
+      "loss": 0.5819,
+      "step": 12215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012270539418866093,
+      "loss": 0.6386,
+      "step": 12216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012269467987172336,
+      "loss": 0.5543,
+      "step": 12217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012268396528010447,
+      "loss": 0.6777,
+      "step": 12218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000122673250413934,
+      "loss": 0.6327,
+      "step": 12219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012266253527334163,
+      "loss": 0.675,
+      "step": 12220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012265181985845702,
+      "loss": 0.624,
+      "step": 12221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012264110416940988,
+      "loss": 0.5262,
+      "step": 12222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012263038820632994,
+      "loss": 0.6259,
+      "step": 12223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012261967196934684,
+      "loss": 0.503,
+      "step": 12224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001226089554585903,
+      "loss": 0.6129,
+      "step": 12225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012259823867419002,
+      "loss": 0.7146,
+      "step": 12226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012258752161627576,
+      "loss": 0.752,
+      "step": 12227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001225768042849772,
+      "loss": 0.6592,
+      "step": 12228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000122566086680424,
+      "loss": 0.6221,
+      "step": 12229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012255536880274596,
+      "loss": 0.5678,
+      "step": 12230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012254465065207276,
+      "loss": 0.6866,
+      "step": 12231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012253393222853414,
+      "loss": 0.6009,
+      "step": 12232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012252321353225983,
+      "loss": 0.7095,
+      "step": 12233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012251249456337953,
+      "loss": 0.6511,
+      "step": 12234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012250177532202304,
+      "loss": 0.6517,
+      "step": 12235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012249105580832006,
+      "loss": 0.644,
+      "step": 12236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001224803360224003,
+      "loss": 0.6523,
+      "step": 12237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012246961596439358,
+      "loss": 0.5856,
+      "step": 12238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012245889563442956,
+      "loss": 0.5911,
+      "step": 12239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012244817503263806,
+      "loss": 0.7137,
+      "step": 12240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012243745415914883,
+      "loss": 0.7824,
+      "step": 12241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012242673301409156,
+      "loss": 0.701,
+      "step": 12242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001224160115975961,
+      "loss": 0.6081,
+      "step": 12243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012240528990979215,
+      "loss": 0.6884,
+      "step": 12244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012239456795080955,
+      "loss": 0.7452,
+      "step": 12245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000122383845720778,
+      "loss": 0.5508,
+      "step": 12246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001223731232198273,
+      "loss": 0.6577,
+      "step": 12247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012236240044808725,
+      "loss": 0.6043,
+      "step": 12248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012235167740568757,
+      "loss": 0.6019,
+      "step": 12249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001223409540927581,
+      "loss": 0.588,
+      "step": 12250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012233023050942863,
+      "loss": 0.6913,
+      "step": 12251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001223195066558289,
+      "loss": 0.6302,
+      "step": 12252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001223087825320888,
+      "loss": 0.6401,
+      "step": 12253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000122298058138338,
+      "loss": 0.6015,
+      "step": 12254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001222873334747064,
+      "loss": 0.59,
+      "step": 12255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001222766085413238,
+      "loss": 0.6429,
+      "step": 12256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001222658833383199,
+      "loss": 0.7239,
+      "step": 12257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012225515786582465,
+      "loss": 0.5291,
+      "step": 12258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012224443212396777,
+      "loss": 0.5427,
+      "step": 12259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012223370611287915,
+      "loss": 0.6728,
+      "step": 12260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012222297983268855,
+      "loss": 0.737,
+      "step": 12261
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001222122532835258,
+      "loss": 0.7053,
+      "step": 12262
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012220152646552076,
+      "loss": 0.6737,
+      "step": 12263
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012219079937880322,
+      "loss": 0.5479,
+      "step": 12264
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012218007202350307,
+      "loss": 0.5816,
+      "step": 12265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001221693443997501,
+      "loss": 0.6458,
+      "step": 12266
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012215861650767415,
+      "loss": 0.5882,
+      "step": 12267
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001221478883474051,
+      "loss": 0.5712,
+      "step": 12268
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012213715991907272,
+      "loss": 0.5931,
+      "step": 12269
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000122126431222807,
+      "loss": 0.6725,
+      "step": 12270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012211570225873763,
+      "loss": 0.6046,
+      "step": 12271
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012210497302699456,
+      "loss": 0.5832,
+      "step": 12272
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012209424352770763,
+      "loss": 0.6558,
+      "step": 12273
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001220835137610067,
+      "loss": 0.5593,
+      "step": 12274
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012207278372702165,
+      "loss": 0.7066,
+      "step": 12275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012206205342588235,
+      "loss": 0.5878,
+      "step": 12276
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012205132285771863,
+      "loss": 0.6291,
+      "step": 12277
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012204059202266043,
+      "loss": 0.6952,
+      "step": 12278
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012202986092083757,
+      "loss": 0.5402,
+      "step": 12279
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012201912955237996,
+      "loss": 0.6188,
+      "step": 12280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001220083979174175,
+      "loss": 0.5766,
+      "step": 12281
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012199766601608003,
+      "loss": 0.5312,
+      "step": 12282
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012198693384849748,
+      "loss": 0.5972,
+      "step": 12283
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012197620141479977,
+      "loss": 0.6058,
+      "step": 12284
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012196546871511674,
+      "loss": 0.6643,
+      "step": 12285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012195473574957835,
+      "loss": 0.5903,
+      "step": 12286
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012194400251831444,
+      "loss": 0.7084,
+      "step": 12287
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012193326902145495,
+      "loss": 0.5395,
+      "step": 12288
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001219225352591298,
+      "loss": 0.6432,
+      "step": 12289
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012191180123146889,
+      "loss": 0.6063,
+      "step": 12290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012190106693860215,
+      "loss": 0.6044,
+      "step": 12291
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001218903323806595,
+      "loss": 0.6283,
+      "step": 12292
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012187959755777083,
+      "loss": 0.6214,
+      "step": 12293
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012186886247006613,
+      "loss": 0.6002,
+      "step": 12294
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012185812711767526,
+      "loss": 0.6186,
+      "step": 12295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012184739150072821,
+      "loss": 0.594,
+      "step": 12296
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012183665561935491,
+      "loss": 0.6852,
+      "step": 12297
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012182591947368524,
+      "loss": 0.616,
+      "step": 12298
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012181518306384925,
+      "loss": 0.6083,
+      "step": 12299
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012180444638997677,
+      "loss": 0.5833,
+      "step": 12300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012179370945219784,
+      "loss": 0.5999,
+      "step": 12301
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012178297225064237,
+      "loss": 0.5966,
+      "step": 12302
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012177223478544032,
+      "loss": 0.5482,
+      "step": 12303
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012176149705672164,
+      "loss": 0.6249,
+      "step": 12304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012175075906461633,
+      "loss": 0.6706,
+      "step": 12305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012174002080925431,
+      "loss": 0.707,
+      "step": 12306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012172928229076559,
+      "loss": 0.61,
+      "step": 12307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012171854350928007,
+      "loss": 0.5489,
+      "step": 12308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012170780446492782,
+      "loss": 0.6547,
+      "step": 12309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012169706515783878,
+      "loss": 0.6185,
+      "step": 12310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001216863255881429,
+      "loss": 0.6957,
+      "step": 12311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012167558575597025,
+      "loss": 0.6543,
+      "step": 12312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012166484566145069,
+      "loss": 0.5872,
+      "step": 12313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001216541053047143,
+      "loss": 0.5867,
+      "step": 12314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012164336468589104,
+      "loss": 0.6103,
+      "step": 12315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012163262380511094,
+      "loss": 0.5962,
+      "step": 12316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012162188266250398,
+      "loss": 0.6309,
+      "step": 12317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012161114125820015,
+      "loss": 0.6914,
+      "step": 12318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012160039959232947,
+      "loss": 0.5867,
+      "step": 12319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012158965766502195,
+      "loss": 0.6345,
+      "step": 12320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012157891547640763,
+      "loss": 0.6166,
+      "step": 12321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001215681730266165,
+      "loss": 0.783,
+      "step": 12322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012155743031577854,
+      "loss": 0.7488,
+      "step": 12323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012154668734402384,
+      "loss": 0.6661,
+      "step": 12324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012153594411148239,
+      "loss": 0.5696,
+      "step": 12325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012152520061828423,
+      "loss": 0.6313,
+      "step": 12326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001215144568645594,
+      "loss": 0.5957,
+      "step": 12327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012150371285043789,
+      "loss": 0.7006,
+      "step": 12328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012149296857604982,
+      "loss": 0.5913,
+      "step": 12329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012148222404152516,
+      "loss": 0.5886,
+      "step": 12330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012147147924699398,
+      "loss": 0.6182,
+      "step": 12331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012146073419258632,
+      "loss": 0.6413,
+      "step": 12332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012144998887843225,
+      "loss": 0.5978,
+      "step": 12333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001214392433046618,
+      "loss": 0.5981,
+      "step": 12334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012142849747140508,
+      "loss": 0.6827,
+      "step": 12335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012141775137879205,
+      "loss": 0.6435,
+      "step": 12336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012140700502695288,
+      "loss": 0.5131,
+      "step": 12337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012139625841601759,
+      "loss": 0.5559,
+      "step": 12338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012138551154611622,
+      "loss": 0.7477,
+      "step": 12339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012137476441737888,
+      "loss": 0.5808,
+      "step": 12340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012136401702993563,
+      "loss": 0.5791,
+      "step": 12341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012135326938391658,
+      "loss": 0.5391,
+      "step": 12342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012134252147945177,
+      "loss": 0.5538,
+      "step": 12343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012133177331667131,
+      "loss": 0.674,
+      "step": 12344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012132102489570532,
+      "loss": 0.6209,
+      "step": 12345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001213102762166838,
+      "loss": 0.6153,
+      "step": 12346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012129952727973694,
+      "loss": 0.6303,
+      "step": 12347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012128877808499479,
+      "loss": 0.5667,
+      "step": 12348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012127802863258746,
+      "loss": 0.5594,
+      "step": 12349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012126727892264505,
+      "loss": 0.6419,
+      "step": 12350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012125652895529766,
+      "loss": 0.7427,
+      "step": 12351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012124577873067544,
+      "loss": 0.7231,
+      "step": 12352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012123502824890846,
+      "loss": 0.719,
+      "step": 12353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012122427751012686,
+      "loss": 0.6762,
+      "step": 12354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012121352651446074,
+      "loss": 0.631,
+      "step": 12355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012120277526204024,
+      "loss": 0.5421,
+      "step": 12356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012119202375299548,
+      "loss": 0.6131,
+      "step": 12357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012118127198745661,
+      "loss": 0.6406,
+      "step": 12358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012117051996555371,
+      "loss": 0.5362,
+      "step": 12359
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012115976768741697,
+      "loss": 0.4915,
+      "step": 12360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001211490151531765,
+      "loss": 0.6084,
+      "step": 12361
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012113826236296244,
+      "loss": 0.6114,
+      "step": 12362
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012112750931690496,
+      "loss": 0.5928,
+      "step": 12363
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012111675601513415,
+      "loss": 0.5945,
+      "step": 12364
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012110600245778024,
+      "loss": 0.59,
+      "step": 12365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012109524864497335,
+      "loss": 0.6207,
+      "step": 12366
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012108449457684359,
+      "loss": 0.6617,
+      "step": 12367
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012107374025352118,
+      "loss": 0.5902,
+      "step": 12368
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012106298567513628,
+      "loss": 0.6352,
+      "step": 12369
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012105223084181903,
+      "loss": 0.6409,
+      "step": 12370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012104147575369961,
+      "loss": 0.5653,
+      "step": 12371
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001210307204109082,
+      "loss": 0.5922,
+      "step": 12372
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012101996481357495,
+      "loss": 0.5974,
+      "step": 12373
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012100920896183008,
+      "loss": 0.7087,
+      "step": 12374
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012099845285580373,
+      "loss": 0.5658,
+      "step": 12375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012098769649562612,
+      "loss": 0.5318,
+      "step": 12376
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001209769398814274,
+      "loss": 0.5653,
+      "step": 12377
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001209661830133378,
+      "loss": 0.5659,
+      "step": 12378
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012095542589148749,
+      "loss": 0.5316,
+      "step": 12379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012094466851600667,
+      "loss": 0.6735,
+      "step": 12380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012093391088702557,
+      "loss": 0.601,
+      "step": 12381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012092315300467431,
+      "loss": 0.6523,
+      "step": 12382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012091239486908322,
+      "loss": 0.6264,
+      "step": 12383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012090163648038237,
+      "loss": 0.5816,
+      "step": 12384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001208908778387021,
+      "loss": 0.5875,
+      "step": 12385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012088011894417255,
+      "loss": 0.6469,
+      "step": 12386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012086935979692395,
+      "loss": 0.5704,
+      "step": 12387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012085860039708654,
+      "loss": 0.5703,
+      "step": 12388
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012084784074479054,
+      "loss": 0.612,
+      "step": 12389
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012083708084016614,
+      "loss": 0.6275,
+      "step": 12390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012082632068334363,
+      "loss": 0.5754,
+      "step": 12391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001208155602744532,
+      "loss": 0.5915,
+      "step": 12392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012080479961362511,
+      "loss": 0.5433,
+      "step": 12393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012079403870098962,
+      "loss": 0.5945,
+      "step": 12394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012078327753667691,
+      "loss": 0.5622,
+      "step": 12395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012077251612081728,
+      "loss": 0.5995,
+      "step": 12396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012076175445354095,
+      "loss": 0.7555,
+      "step": 12397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012075099253497819,
+      "loss": 0.6816,
+      "step": 12398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012074023036525926,
+      "loss": 0.5536,
+      "step": 12399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012072946794451436,
+      "loss": 0.6534,
+      "step": 12400
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 0.6013256311416626,
+      "eval_runtime": 843.9337,
+      "eval_samples_per_second": 28.509,
+      "eval_steps_per_second": 14.255,
+      "step": 12400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012071870527287385,
+      "loss": 0.5996,
+      "step": 12401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012070794235046789,
+      "loss": 0.5677,
+      "step": 12402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012069717917742684,
+      "loss": 0.5642,
+      "step": 12403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012068641575388094,
+      "loss": 0.5483,
+      "step": 12404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012067565207996041,
+      "loss": 0.6752,
+      "step": 12405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001206648881557956,
+      "loss": 0.5489,
+      "step": 12406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012065412398151674,
+      "loss": 0.6934,
+      "step": 12407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012064335955725415,
+      "loss": 0.5956,
+      "step": 12408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012063259488313808,
+      "loss": 0.6727,
+      "step": 12409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012062182995929882,
+      "loss": 0.5948,
+      "step": 12410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001206110647858667,
+      "loss": 0.5937,
+      "step": 12411
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000120600299362972,
+      "loss": 0.6972,
+      "step": 12412
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012058953369074497,
+      "loss": 0.5402,
+      "step": 12413
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012057876776931597,
+      "loss": 0.5377,
+      "step": 12414
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012056800159881527,
+      "loss": 0.5663,
+      "step": 12415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012055723517937321,
+      "loss": 0.5538,
+      "step": 12416
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012054646851112008,
+      "loss": 0.6482,
+      "step": 12417
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012053570159418615,
+      "loss": 0.5607,
+      "step": 12418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012052493442870183,
+      "loss": 0.5155,
+      "step": 12419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012051416701479734,
+      "loss": 0.5921,
+      "step": 12420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012050339935260306,
+      "loss": 0.5666,
+      "step": 12421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012049263144224932,
+      "loss": 0.6319,
+      "step": 12422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012048186328386639,
+      "loss": 0.5766,
+      "step": 12423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012047109487758468,
+      "loss": 0.5509,
+      "step": 12424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012046032622353447,
+      "loss": 0.5854,
+      "step": 12425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012044955732184609,
+      "loss": 0.5891,
+      "step": 12426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012043878817264991,
+      "loss": 0.6242,
+      "step": 12427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012042801877607625,
+      "loss": 0.6915,
+      "step": 12428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001204172491322555,
+      "loss": 0.7383,
+      "step": 12429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012040647924131796,
+      "loss": 0.6036,
+      "step": 12430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012039570910339397,
+      "loss": 0.5493,
+      "step": 12431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012038493871861395,
+      "loss": 0.7287,
+      "step": 12432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001203741680871082,
+      "loss": 0.6428,
+      "step": 12433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012036339720900711,
+      "loss": 0.6639,
+      "step": 12434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012035262608444102,
+      "loss": 0.6106,
+      "step": 12435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012034185471354032,
+      "loss": 0.5055,
+      "step": 12436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012033108309643537,
+      "loss": 0.7158,
+      "step": 12437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012032031123325656,
+      "loss": 0.6809,
+      "step": 12438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012030953912413421,
+      "loss": 0.608,
+      "step": 12439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012029876676919876,
+      "loss": 0.6115,
+      "step": 12440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012028799416858054,
+      "loss": 0.6302,
+      "step": 12441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012027722132241,
+      "loss": 0.6497,
+      "step": 12442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012026644823081747,
+      "loss": 0.6135,
+      "step": 12443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012025567489393335,
+      "loss": 0.639,
+      "step": 12444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012024490131188806,
+      "loss": 0.6585,
+      "step": 12445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012023412748481197,
+      "loss": 0.5928,
+      "step": 12446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001202233534128355,
+      "loss": 0.7195,
+      "step": 12447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000120212579096089,
+      "loss": 0.5767,
+      "step": 12448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012020180453470296,
+      "loss": 0.6693,
+      "step": 12449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012019102972880774,
+      "loss": 0.6351,
+      "step": 12450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012018025467853373,
+      "loss": 0.6424,
+      "step": 12451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012016947938401139,
+      "loss": 0.6129,
+      "step": 12452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012015870384537109,
+      "loss": 0.5823,
+      "step": 12453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012014792806274329,
+      "loss": 0.5313,
+      "step": 12454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012013715203625841,
+      "loss": 0.623,
+      "step": 12455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012012637576604684,
+      "loss": 0.5537,
+      "step": 12456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012011559925223906,
+      "loss": 0.5211,
+      "step": 12457
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012010482249496546,
+      "loss": 0.6076,
+      "step": 12458
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012009404549435648,
+      "loss": 0.6894,
+      "step": 12459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001200832682505426,
+      "loss": 0.6435,
+      "step": 12460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012007249076365419,
+      "loss": 0.5652,
+      "step": 12461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012006171303382174,
+      "loss": 0.5512,
+      "step": 12462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012005093506117571,
+      "loss": 0.5764,
+      "step": 12463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012004015684584651,
+      "loss": 0.6032,
+      "step": 12464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001200293783879646,
+      "loss": 0.6192,
+      "step": 12465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012001859968766045,
+      "loss": 0.532,
+      "step": 12466
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012000782074506451,
+      "loss": 0.6879,
+      "step": 12467
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011999704156030726,
+      "loss": 0.6004,
+      "step": 12468
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011998626213351913,
+      "loss": 0.535,
+      "step": 12469
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011997548246483064,
+      "loss": 0.5871,
+      "step": 12470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011996470255437217,
+      "loss": 0.5326,
+      "step": 12471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001199539224022743,
+      "loss": 0.6241,
+      "step": 12472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011994314200866743,
+      "loss": 0.5814,
+      "step": 12473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011993236137368204,
+      "loss": 0.5915,
+      "step": 12474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011992158049744869,
+      "loss": 0.558,
+      "step": 12475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011991079938009777,
+      "loss": 0.5534,
+      "step": 12476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011990001802175981,
+      "loss": 0.6142,
+      "step": 12477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011988923642256529,
+      "loss": 0.6227,
+      "step": 12478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001198784545826447,
+      "loss": 0.6804,
+      "step": 12479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011986767250212858,
+      "loss": 0.5173,
+      "step": 12480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011985689018114738,
+      "loss": 0.6022,
+      "step": 12481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011984610761983162,
+      "loss": 0.6597,
+      "step": 12482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001198353248183118,
+      "loss": 0.6516,
+      "step": 12483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001198245417767184,
+      "loss": 0.6231,
+      "step": 12484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000119813758495182,
+      "loss": 0.6025,
+      "step": 12485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011980297497383306,
+      "loss": 0.5962,
+      "step": 12486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001197921912128021,
+      "loss": 0.6216,
+      "step": 12487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011978140721221967,
+      "loss": 0.5548,
+      "step": 12488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011977062297221627,
+      "loss": 0.6912,
+      "step": 12489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011975983849292241,
+      "loss": 0.598,
+      "step": 12490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011974905377446865,
+      "loss": 0.5492,
+      "step": 12491
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011973826881698549,
+      "loss": 0.5058,
+      "step": 12492
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011972748362060351,
+      "loss": 0.5719,
+      "step": 12493
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011971669818545321,
+      "loss": 0.6649,
+      "step": 12494
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011970591251166511,
+      "loss": 0.7074,
+      "step": 12495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011969512659936981,
+      "loss": 0.5852,
+      "step": 12496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011968434044869781,
+      "loss": 0.5522,
+      "step": 12497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011967355405977967,
+      "loss": 0.6212,
+      "step": 12498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011966276743274595,
+      "loss": 0.576,
+      "step": 12499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011965198056772719,
+      "loss": 0.5477,
+      "step": 12500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011964119346485399,
+      "loss": 0.5997,
+      "step": 12501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011963040612425684,
+      "loss": 0.6097,
+      "step": 12502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011961961854606636,
+      "loss": 0.5843,
+      "step": 12503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011960883073041311,
+      "loss": 0.5512,
+      "step": 12504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011959804267742761,
+      "loss": 0.6156,
+      "step": 12505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011958725438724047,
+      "loss": 0.5289,
+      "step": 12506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011957646585998227,
+      "loss": 0.7058,
+      "step": 12507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011956567709578357,
+      "loss": 0.5973,
+      "step": 12508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011955488809477495,
+      "loss": 0.6687,
+      "step": 12509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011954409885708699,
+      "loss": 0.6829,
+      "step": 12510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011953330938285032,
+      "loss": 0.6158,
+      "step": 12511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011952251967219548,
+      "loss": 0.7242,
+      "step": 12512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011951172972525306,
+      "loss": 0.5128,
+      "step": 12513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011950093954215372,
+      "loss": 0.6389,
+      "step": 12514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011949014912302794,
+      "loss": 0.6785,
+      "step": 12515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011947935846800642,
+      "loss": 0.7175,
+      "step": 12516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001194685675772197,
+      "loss": 0.6585,
+      "step": 12517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011945777645079844,
+      "loss": 0.5332,
+      "step": 12518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011944698508887322,
+      "loss": 0.6699,
+      "step": 12519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011943619349157465,
+      "loss": 0.5187,
+      "step": 12520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011942540165903335,
+      "loss": 0.5384,
+      "step": 12521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011941460959137992,
+      "loss": 0.5275,
+      "step": 12522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011940381728874502,
+      "loss": 0.7143,
+      "step": 12523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011939302475125927,
+      "loss": 0.6175,
+      "step": 12524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011938223197905323,
+      "loss": 0.6382,
+      "step": 12525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011937143897225757,
+      "loss": 0.5625,
+      "step": 12526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011936064573100297,
+      "loss": 0.6484,
+      "step": 12527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011934985225541998,
+      "loss": 0.5775,
+      "step": 12528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011933905854563929,
+      "loss": 0.554,
+      "step": 12529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011932826460179152,
+      "loss": 0.6424,
+      "step": 12530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011931747042400733,
+      "loss": 0.5552,
+      "step": 12531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011930667601241736,
+      "loss": 0.5303,
+      "step": 12532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011929588136715222,
+      "loss": 0.5764,
+      "step": 12533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011928508648834262,
+      "loss": 0.6076,
+      "step": 12534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011927429137611918,
+      "loss": 0.5544,
+      "step": 12535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011926349603061258,
+      "loss": 0.643,
+      "step": 12536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011925270045195344,
+      "loss": 0.633,
+      "step": 12537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011924190464027245,
+      "loss": 0.6255,
+      "step": 12538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001192311085957003,
+      "loss": 0.6614,
+      "step": 12539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001192203123183676,
+      "loss": 0.7233,
+      "step": 12540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011920951580840505,
+      "loss": 0.5795,
+      "step": 12541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011919871906594335,
+      "loss": 0.6473,
+      "step": 12542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011918792209111311,
+      "loss": 0.6024,
+      "step": 12543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011917712488404508,
+      "loss": 0.593,
+      "step": 12544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011916632744486993,
+      "loss": 0.5841,
+      "step": 12545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011915552977371828,
+      "loss": 0.6113,
+      "step": 12546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011914473187072092,
+      "loss": 0.5697,
+      "step": 12547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011913393373600845,
+      "loss": 0.67,
+      "step": 12548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011912313536971162,
+      "loss": 0.544,
+      "step": 12549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001191123367719611,
+      "loss": 0.6128,
+      "step": 12550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011910153794288757,
+      "loss": 0.7618,
+      "step": 12551
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001190907388826218,
+      "loss": 0.6278,
+      "step": 12552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011907993959129439,
+      "loss": 0.5627,
+      "step": 12553
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011906914006903618,
+      "loss": 0.7947,
+      "step": 12554
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011905834031597777,
+      "loss": 0.5256,
+      "step": 12555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011904754033224992,
+      "loss": 0.6428,
+      "step": 12556
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011903674011798334,
+      "loss": 0.5098,
+      "step": 12557
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011902593967330875,
+      "loss": 0.6334,
+      "step": 12558
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011901513899835686,
+      "loss": 0.5945,
+      "step": 12559
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011900433809325839,
+      "loss": 0.6119,
+      "step": 12560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011899353695814408,
+      "loss": 0.6004,
+      "step": 12561
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001189827355931447,
+      "loss": 0.5737,
+      "step": 12562
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001189719339983909,
+      "loss": 0.6616,
+      "step": 12563
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011896113217401348,
+      "loss": 0.6184,
+      "step": 12564
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011895033012014314,
+      "loss": 0.6068,
+      "step": 12565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011893952783691062,
+      "loss": 0.6681,
+      "step": 12566
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011892872532444673,
+      "loss": 0.6068,
+      "step": 12567
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011891792258288212,
+      "loss": 0.6487,
+      "step": 12568
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011890711961234759,
+      "loss": 0.6292,
+      "step": 12569
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011889631641297391,
+      "loss": 0.579,
+      "step": 12570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011888551298489179,
+      "loss": 0.6559,
+      "step": 12571
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011887470932823202,
+      "loss": 0.6673,
+      "step": 12572
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011886390544312536,
+      "loss": 0.4938,
+      "step": 12573
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011885310132970253,
+      "loss": 0.6741,
+      "step": 12574
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011884229698809435,
+      "loss": 0.586,
+      "step": 12575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011883149241843157,
+      "loss": 0.6649,
+      "step": 12576
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011882068762084495,
+      "loss": 0.512,
+      "step": 12577
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011880988259546528,
+      "loss": 0.5935,
+      "step": 12578
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001187990773424233,
+      "loss": 0.5646,
+      "step": 12579
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011878827186184985,
+      "loss": 0.5461,
+      "step": 12580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011877746615387568,
+      "loss": 0.5853,
+      "step": 12581
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011876666021863156,
+      "loss": 0.5503,
+      "step": 12582
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011875585405624831,
+      "loss": 0.5478,
+      "step": 12583
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001187450476668567,
+      "loss": 0.6427,
+      "step": 12584
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011873424105058752,
+      "loss": 0.6604,
+      "step": 12585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011872343420757157,
+      "loss": 0.7038,
+      "step": 12586
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011871262713793967,
+      "loss": 0.6161,
+      "step": 12587
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011870181984182258,
+      "loss": 0.6493,
+      "step": 12588
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011869101231935114,
+      "loss": 0.725,
+      "step": 12589
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011868020457065615,
+      "loss": 0.4973,
+      "step": 12590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011866939659586842,
+      "loss": 0.5511,
+      "step": 12591
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011865858839511873,
+      "loss": 0.5656,
+      "step": 12592
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011864777996853795,
+      "loss": 0.6972,
+      "step": 12593
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011863697131625686,
+      "loss": 0.5641,
+      "step": 12594
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001186261624384063,
+      "loss": 0.6745,
+      "step": 12595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011861535333511708,
+      "loss": 0.6877,
+      "step": 12596
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011860454400652001,
+      "loss": 0.5693,
+      "step": 12597
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011859373445274597,
+      "loss": 0.7513,
+      "step": 12598
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011858292467392575,
+      "loss": 0.6333,
+      "step": 12599
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001185721146701902,
+      "loss": 0.6382,
+      "step": 12600
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 0.596328616142273,
+      "eval_runtime": 847.4761,
+      "eval_samples_per_second": 28.39,
+      "eval_steps_per_second": 14.195,
+      "step": 12600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011856130444167018,
+      "loss": 0.5689,
+      "step": 12601
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011855049398849645,
+      "loss": 0.684,
+      "step": 12602
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011853968331079995,
+      "loss": 0.5794,
+      "step": 12603
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011852887240871145,
+      "loss": 0.5491,
+      "step": 12604
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011851806128236183,
+      "loss": 0.6143,
+      "step": 12605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011850724993188196,
+      "loss": 0.6548,
+      "step": 12606
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011849643835740265,
+      "loss": 0.6031,
+      "step": 12607
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011848562655905479,
+      "loss": 0.5675,
+      "step": 12608
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011847481453696925,
+      "loss": 0.705,
+      "step": 12609
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011846400229127684,
+      "loss": 0.6088,
+      "step": 12610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011845318982210847,
+      "loss": 0.709,
+      "step": 12611
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011844237712959496,
+      "loss": 0.5845,
+      "step": 12612
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011843156421386726,
+      "loss": 0.5424,
+      "step": 12613
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011842075107505617,
+      "loss": 0.7105,
+      "step": 12614
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011840993771329256,
+      "loss": 0.6286,
+      "step": 12615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011839912412870737,
+      "loss": 0.5427,
+      "step": 12616
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011838831032143141,
+      "loss": 0.6359,
+      "step": 12617
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011837749629159562,
+      "loss": 0.5918,
+      "step": 12618
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011836668203933088,
+      "loss": 0.6062,
+      "step": 12619
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011835586756476805,
+      "loss": 0.5309,
+      "step": 12620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011834505286803802,
+      "loss": 0.58,
+      "step": 12621
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011833423794927169,
+      "loss": 0.511,
+      "step": 12622
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001183234228086,
+      "loss": 0.576,
+      "step": 12623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011831260744615382,
+      "loss": 0.5902,
+      "step": 12624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000118301791862064,
+      "loss": 0.5883,
+      "step": 12625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011829097605646153,
+      "loss": 0.5769,
+      "step": 12626
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011828016002947726,
+      "loss": 0.6831,
+      "step": 12627
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011826934378124212,
+      "loss": 0.5753,
+      "step": 12628
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011825852731188702,
+      "loss": 0.5887,
+      "step": 12629
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011824771062154287,
+      "loss": 0.6625,
+      "step": 12630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011823689371034061,
+      "loss": 0.5972,
+      "step": 12631
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011822607657841114,
+      "loss": 0.6146,
+      "step": 12632
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011821525922588536,
+      "loss": 0.515,
+      "step": 12633
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011820444165289426,
+      "loss": 0.6321,
+      "step": 12634
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011819362385956871,
+      "loss": 0.7032,
+      "step": 12635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011818280584603967,
+      "loss": 0.5638,
+      "step": 12636
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011817198761243808,
+      "loss": 0.6407,
+      "step": 12637
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011816116915889484,
+      "loss": 0.5649,
+      "step": 12638
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011815035048554095,
+      "loss": 0.6348,
+      "step": 12639
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001181395315925073,
+      "loss": 0.5777,
+      "step": 12640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011812871247992483,
+      "loss": 0.6211,
+      "step": 12641
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011811789314792452,
+      "loss": 0.5829,
+      "step": 12642
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001181070735966373,
+      "loss": 0.6121,
+      "step": 12643
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011809625382619416,
+      "loss": 0.6749,
+      "step": 12644
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011808543383672602,
+      "loss": 0.6421,
+      "step": 12645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001180746136283638,
+      "loss": 0.6935,
+      "step": 12646
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011806379320123855,
+      "loss": 0.5911,
+      "step": 12647
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011805297255548118,
+      "loss": 0.5871,
+      "step": 12648
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011804215169122266,
+      "loss": 0.5954,
+      "step": 12649
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011803133060859392,
+      "loss": 0.5839,
+      "step": 12650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011802050930772604,
+      "loss": 0.6166,
+      "step": 12651
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001180096877887499,
+      "loss": 0.6757,
+      "step": 12652
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011799886605179648,
+      "loss": 0.5101,
+      "step": 12653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011798804409699682,
+      "loss": 0.688,
+      "step": 12654
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011797722192448184,
+      "loss": 0.5233,
+      "step": 12655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011796639953438255,
+      "loss": 0.6428,
+      "step": 12656
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011795557692682995,
+      "loss": 0.5806,
+      "step": 12657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000117944754101955,
+      "loss": 0.5489,
+      "step": 12658
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011793393105988873,
+      "loss": 0.5863,
+      "step": 12659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001179231078007621,
+      "loss": 0.5799,
+      "step": 12660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011791228432470609,
+      "loss": 0.5308,
+      "step": 12661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011790146063185176,
+      "loss": 0.6569,
+      "step": 12662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011789063672233008,
+      "loss": 0.4754,
+      "step": 12663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011787981259627205,
+      "loss": 0.6913,
+      "step": 12664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011786898825380872,
+      "loss": 0.6485,
+      "step": 12665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011785816369507102,
+      "loss": 0.6396,
+      "step": 12666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011784733892019005,
+      "loss": 0.6556,
+      "step": 12667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011783651392929673,
+      "loss": 0.5527,
+      "step": 12668
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011782568872252217,
+      "loss": 0.5527,
+      "step": 12669
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011781486329999737,
+      "loss": 0.6045,
+      "step": 12670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011780403766185333,
+      "loss": 0.6086,
+      "step": 12671
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011779321180822107,
+      "loss": 0.6282,
+      "step": 12672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011778238573923163,
+      "loss": 0.5609,
+      "step": 12673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011777155945501608,
+      "loss": 0.6116,
+      "step": 12674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001177607329557054,
+      "loss": 0.5391,
+      "step": 12675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011774990624143063,
+      "loss": 0.6786,
+      "step": 12676
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011773907931232287,
+      "loss": 0.6569,
+      "step": 12677
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011772825216851309,
+      "loss": 0.5771,
+      "step": 12678
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011771742481013235,
+      "loss": 0.6114,
+      "step": 12679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011770659723731172,
+      "loss": 0.5749,
+      "step": 12680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011769576945018223,
+      "loss": 0.7011,
+      "step": 12681
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011768494144887496,
+      "loss": 0.6555,
+      "step": 12682
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011767411323352096,
+      "loss": 0.5618,
+      "step": 12683
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011766328480425124,
+      "loss": 0.5854,
+      "step": 12684
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011765245616119691,
+      "loss": 0.5376,
+      "step": 12685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000117641627304489,
+      "loss": 0.5308,
+      "step": 12686
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011763079823425862,
+      "loss": 0.6689,
+      "step": 12687
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011761996895063681,
+      "loss": 0.5508,
+      "step": 12688
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001176091394537546,
+      "loss": 0.7222,
+      "step": 12689
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011759830974374315,
+      "loss": 0.5496,
+      "step": 12690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011758747982073348,
+      "loss": 0.6217,
+      "step": 12691
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011757664968485667,
+      "loss": 0.6872,
+      "step": 12692
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011756581933624379,
+      "loss": 0.6456,
+      "step": 12693
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011755498877502596,
+      "loss": 0.5824,
+      "step": 12694
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011754415800133425,
+      "loss": 0.7001,
+      "step": 12695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011753332701529973,
+      "loss": 0.6812,
+      "step": 12696
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001175224958170535,
+      "loss": 0.6009,
+      "step": 12697
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011751166440672669,
+      "loss": 0.5855,
+      "step": 12698
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011750083278445034,
+      "loss": 0.5952,
+      "step": 12699
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001174900009503556,
+      "loss": 0.6079,
+      "step": 12700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011747916890457356,
+      "loss": 0.5719,
+      "step": 12701
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011746833664723527,
+      "loss": 0.6892,
+      "step": 12702
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011745750417847188,
+      "loss": 0.5976,
+      "step": 12703
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001174466714984145,
+      "loss": 0.4863,
+      "step": 12704
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011743583860719424,
+      "loss": 0.6554,
+      "step": 12705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011742500550494224,
+      "loss": 0.647,
+      "step": 12706
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011741417219178954,
+      "loss": 0.57,
+      "step": 12707
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011740333866786735,
+      "loss": 0.6587,
+      "step": 12708
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011739250493330673,
+      "loss": 0.6681,
+      "step": 12709
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011738167098823883,
+      "loss": 0.6459,
+      "step": 12710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011737083683279475,
+      "loss": 0.5447,
+      "step": 12711
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011736000246710565,
+      "loss": 0.5763,
+      "step": 12712
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011734916789130266,
+      "loss": 0.5502,
+      "step": 12713
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011733833310551692,
+      "loss": 0.6121,
+      "step": 12714
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011732749810987951,
+      "loss": 0.6658,
+      "step": 12715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011731666290452165,
+      "loss": 0.6587,
+      "step": 12716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011730582748957444,
+      "loss": 0.5775,
+      "step": 12717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011729499186516901,
+      "loss": 0.5776,
+      "step": 12718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011728415603143655,
+      "loss": 0.5525,
+      "step": 12719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011727331998850817,
+      "loss": 0.6578,
+      "step": 12720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011726248373651505,
+      "loss": 0.6756,
+      "step": 12721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001172516472755883,
+      "loss": 0.6394,
+      "step": 12722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011724081060585914,
+      "loss": 0.5917,
+      "step": 12723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001172299737274587,
+      "loss": 0.5944,
+      "step": 12724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011721913664051813,
+      "loss": 0.6677,
+      "step": 12725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011720829934516864,
+      "loss": 0.6181,
+      "step": 12726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011719746184154132,
+      "loss": 0.6559,
+      "step": 12727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001171866241297674,
+      "loss": 0.6185,
+      "step": 12728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011717578620997805,
+      "loss": 0.6024,
+      "step": 12729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001171649480823044,
+      "loss": 0.6419,
+      "step": 12730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001171541097468777,
+      "loss": 0.5697,
+      "step": 12731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011714327120382906,
+      "loss": 0.668,
+      "step": 12732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011713243245328973,
+      "loss": 0.5253,
+      "step": 12733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011712159349539082,
+      "loss": 0.5839,
+      "step": 12734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011711075433026357,
+      "loss": 0.528,
+      "step": 12735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011709991495803915,
+      "loss": 0.5826,
+      "step": 12736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011708907537884877,
+      "loss": 0.5231,
+      "step": 12737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001170782355928236,
+      "loss": 0.715,
+      "step": 12738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011706739560009486,
+      "loss": 0.647,
+      "step": 12739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011705655540079373,
+      "loss": 0.571,
+      "step": 12740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011704571499505142,
+      "loss": 0.5844,
+      "step": 12741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011703487438299914,
+      "loss": 0.6747,
+      "step": 12742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001170240335647681,
+      "loss": 0.636,
+      "step": 12743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001170131925404895,
+      "loss": 0.5088,
+      "step": 12744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011700235131029455,
+      "loss": 0.5913,
+      "step": 12745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011699150987431449,
+      "loss": 0.5024,
+      "step": 12746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011698066823268053,
+      "loss": 0.6222,
+      "step": 12747
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011696982638552382,
+      "loss": 0.5937,
+      "step": 12748
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011695898433297573,
+      "loss": 0.5626,
+      "step": 12749
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011694814207516731,
+      "loss": 0.6456,
+      "step": 12750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011693729961222995,
+      "loss": 0.6091,
+      "step": 12751
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011692645694429476,
+      "loss": 0.4453,
+      "step": 12752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011691561407149302,
+      "loss": 0.5272,
+      "step": 12753
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011690477099395598,
+      "loss": 0.6757,
+      "step": 12754
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011689392771181485,
+      "loss": 0.6458,
+      "step": 12755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001168830842252009,
+      "loss": 0.5851,
+      "step": 12756
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011687224053424533,
+      "loss": 0.5986,
+      "step": 12757
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001168613966390794,
+      "loss": 0.5923,
+      "step": 12758
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011685055253983437,
+      "loss": 0.5736,
+      "step": 12759
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001168397082366415,
+      "loss": 0.5815,
+      "step": 12760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011682886372963199,
+      "loss": 0.5654,
+      "step": 12761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011681801901893719,
+      "loss": 0.5667,
+      "step": 12762
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011680717410468825,
+      "loss": 0.6094,
+      "step": 12763
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011679632898701649,
+      "loss": 0.6616,
+      "step": 12764
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011678548366605317,
+      "loss": 0.6154,
+      "step": 12765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011677463814192952,
+      "loss": 0.594,
+      "step": 12766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011676379241477687,
+      "loss": 0.6405,
+      "step": 12767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011675294648472639,
+      "loss": 0.6174,
+      "step": 12768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011674210035190947,
+      "loss": 0.6964,
+      "step": 12769
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001167312540164573,
+      "loss": 0.6427,
+      "step": 12770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001167204074785012,
+      "loss": 0.5629,
+      "step": 12771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001167095607381724,
+      "loss": 0.5903,
+      "step": 12772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011669871379560224,
+      "loss": 0.5588,
+      "step": 12773
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011668786665092197,
+      "loss": 0.6172,
+      "step": 12774
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011667701930426291,
+      "loss": 0.5736,
+      "step": 12775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011666617175575629,
+      "loss": 0.6509,
+      "step": 12776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011665532400553345,
+      "loss": 0.4729,
+      "step": 12777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011664447605372569,
+      "loss": 0.5723,
+      "step": 12778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011663362790046425,
+      "loss": 0.6412,
+      "step": 12779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001166227795458805,
+      "loss": 0.6877,
+      "step": 12780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011661193099010566,
+      "loss": 0.6198,
+      "step": 12781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011660108223327113,
+      "loss": 0.5015,
+      "step": 12782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011659023327550816,
+      "loss": 0.549,
+      "step": 12783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011657938411694804,
+      "loss": 0.5936,
+      "step": 12784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011656853475772211,
+      "loss": 0.6306,
+      "step": 12785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011655768519796169,
+      "loss": 0.6155,
+      "step": 12786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011654683543779808,
+      "loss": 0.7039,
+      "step": 12787
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011653598547736259,
+      "loss": 0.6333,
+      "step": 12788
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011652513531678657,
+      "loss": 0.5451,
+      "step": 12789
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011651428495620134,
+      "loss": 0.6627,
+      "step": 12790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011650343439573819,
+      "loss": 0.5873,
+      "step": 12791
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011649258363552846,
+      "loss": 0.602,
+      "step": 12792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011648173267570353,
+      "loss": 0.7259,
+      "step": 12793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011647088151639464,
+      "loss": 0.5045,
+      "step": 12794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011646003015773324,
+      "loss": 0.5798,
+      "step": 12795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011644917859985054,
+      "loss": 0.5578,
+      "step": 12796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011643832684287801,
+      "loss": 0.5994,
+      "step": 12797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011642747488694689,
+      "loss": 0.5599,
+      "step": 12798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011641662273218856,
+      "loss": 0.5476,
+      "step": 12799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011640577037873442,
+      "loss": 0.6659,
+      "step": 12800
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 0.5907775163650513,
+      "eval_runtime": 843.7438,
+      "eval_samples_per_second": 28.516,
+      "eval_steps_per_second": 14.258,
+      "step": 12800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001163949178267157,
+      "loss": 0.6075,
+      "step": 12801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011638406507626387,
+      "loss": 0.5991,
+      "step": 12802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011637321212751025,
+      "loss": 0.666,
+      "step": 12803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011636235898058614,
+      "loss": 0.5898,
+      "step": 12804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011635150563562296,
+      "loss": 0.5062,
+      "step": 12805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011634065209275204,
+      "loss": 0.6074,
+      "step": 12806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011632979835210478,
+      "loss": 0.5765,
+      "step": 12807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011631894441381254,
+      "loss": 0.625,
+      "step": 12808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011630809027800664,
+      "loss": 0.6008,
+      "step": 12809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011629723594481852,
+      "loss": 0.5918,
+      "step": 12810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011628638141437951,
+      "loss": 0.5836,
+      "step": 12811
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011627552668682095,
+      "loss": 0.5942,
+      "step": 12812
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011626467176227432,
+      "loss": 0.6041,
+      "step": 12813
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011625381664087092,
+      "loss": 0.6616,
+      "step": 12814
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011624296132274214,
+      "loss": 0.5588,
+      "step": 12815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011623210580801944,
+      "loss": 0.6009,
+      "step": 12816
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001162212500968341,
+      "loss": 0.5721,
+      "step": 12817
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011621039418931758,
+      "loss": 0.5194,
+      "step": 12818
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011619953808560123,
+      "loss": 0.5558,
+      "step": 12819
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001161886817858165,
+      "loss": 0.5374,
+      "step": 12820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011617782529009475,
+      "loss": 0.5235,
+      "step": 12821
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011616696859856737,
+      "loss": 0.5855,
+      "step": 12822
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001161561117113658,
+      "loss": 0.6241,
+      "step": 12823
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011614525462862138,
+      "loss": 0.576,
+      "step": 12824
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011613439735046561,
+      "loss": 0.5669,
+      "step": 12825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011612353987702984,
+      "loss": 0.5891,
+      "step": 12826
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011611268220844547,
+      "loss": 0.5686,
+      "step": 12827
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011610182434484395,
+      "loss": 0.6404,
+      "step": 12828
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011609096628635667,
+      "loss": 0.5118,
+      "step": 12829
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011608010803311506,
+      "loss": 0.6439,
+      "step": 12830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011606924958525055,
+      "loss": 0.6109,
+      "step": 12831
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011605839094289455,
+      "loss": 0.6209,
+      "step": 12832
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011604753210617849,
+      "loss": 0.6629,
+      "step": 12833
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001160366730752338,
+      "loss": 0.6734,
+      "step": 12834
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011602581385019188,
+      "loss": 0.5331,
+      "step": 12835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011601495443118424,
+      "loss": 0.5579,
+      "step": 12836
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011600409481834222,
+      "loss": 0.5691,
+      "step": 12837
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011599323501179735,
+      "loss": 0.6583,
+      "step": 12838
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011598237501168101,
+      "loss": 0.5821,
+      "step": 12839
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011597151481812463,
+      "loss": 0.5545,
+      "step": 12840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011596065443125969,
+      "loss": 0.6348,
+      "step": 12841
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011594979385121764,
+      "loss": 0.555,
+      "step": 12842
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011593893307812991,
+      "loss": 0.5697,
+      "step": 12843
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011592807211212797,
+      "loss": 0.6719,
+      "step": 12844
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011591721095334322,
+      "loss": 0.5442,
+      "step": 12845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011590634960190721,
+      "loss": 0.7256,
+      "step": 12846
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011589548805795134,
+      "loss": 0.6399,
+      "step": 12847
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011588462632160704,
+      "loss": 0.5455,
+      "step": 12848
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011587376439300586,
+      "loss": 0.5233,
+      "step": 12849
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011586290227227919,
+      "loss": 0.6024,
+      "step": 12850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011585203995955853,
+      "loss": 0.4831,
+      "step": 12851
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011584117745497533,
+      "loss": 0.6703,
+      "step": 12852
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011583031475866109,
+      "loss": 0.611,
+      "step": 12853
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011581945187074728,
+      "loss": 0.7277,
+      "step": 12854
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011580858879136534,
+      "loss": 0.5128,
+      "step": 12855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001157977255206468,
+      "loss": 0.5916,
+      "step": 12856
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011578686205872312,
+      "loss": 0.4531,
+      "step": 12857
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011577599840572578,
+      "loss": 0.621,
+      "step": 12858
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011576513456178629,
+      "loss": 0.5675,
+      "step": 12859
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011575427052703607,
+      "loss": 0.6535,
+      "step": 12860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011574340630160669,
+      "loss": 0.6638,
+      "step": 12861
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011573254188562963,
+      "loss": 0.5962,
+      "step": 12862
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011572167727923635,
+      "loss": 0.5781,
+      "step": 12863
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011571081248255837,
+      "loss": 0.6697,
+      "step": 12864
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011569994749572719,
+      "loss": 0.5857,
+      "step": 12865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011568908231887432,
+      "loss": 0.6611,
+      "step": 12866
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011567821695213125,
+      "loss": 0.545,
+      "step": 12867
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011566735139562947,
+      "loss": 0.5649,
+      "step": 12868
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011565648564950053,
+      "loss": 0.5688,
+      "step": 12869
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011564561971387591,
+      "loss": 0.6366,
+      "step": 12870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011563475358888719,
+      "loss": 0.6302,
+      "step": 12871
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011562388727466578,
+      "loss": 0.5511,
+      "step": 12872
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011561302077134326,
+      "loss": 0.6125,
+      "step": 12873
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011560215407905115,
+      "loss": 0.6617,
+      "step": 12874
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011559128719792096,
+      "loss": 0.5694,
+      "step": 12875
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011558042012808423,
+      "loss": 0.6432,
+      "step": 12876
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001155695528696725,
+      "loss": 0.7421,
+      "step": 12877
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011555868542281725,
+      "loss": 0.6715,
+      "step": 12878
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011554781778765005,
+      "loss": 0.5206,
+      "step": 12879
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011553694996430244,
+      "loss": 0.5376,
+      "step": 12880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001155260819529059,
+      "loss": 0.5471,
+      "step": 12881
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011551521375359206,
+      "loss": 0.5398,
+      "step": 12882
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001155043453664924,
+      "loss": 0.5699,
+      "step": 12883
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011549347679173847,
+      "loss": 0.6493,
+      "step": 12884
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011548260802946185,
+      "loss": 0.5524,
+      "step": 12885
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011547173907979402,
+      "loss": 0.6034,
+      "step": 12886
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011546086994286663,
+      "loss": 0.6685,
+      "step": 12887
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001154500006188111,
+      "loss": 0.6476,
+      "step": 12888
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011543913110775912,
+      "loss": 0.576,
+      "step": 12889
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001154282614098422,
+      "loss": 0.5539,
+      "step": 12890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011541739152519185,
+      "loss": 0.5712,
+      "step": 12891
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011540652145393967,
+      "loss": 0.5346,
+      "step": 12892
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011539565119621725,
+      "loss": 0.636,
+      "step": 12893
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011538478075215609,
+      "loss": 0.6356,
+      "step": 12894
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011537391012188783,
+      "loss": 0.644,
+      "step": 12895
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000115363039305544,
+      "loss": 0.5916,
+      "step": 12896
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011535216830325617,
+      "loss": 0.526,
+      "step": 12897
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011534129711515596,
+      "loss": 0.6181,
+      "step": 12898
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011533042574137488,
+      "loss": 0.574,
+      "step": 12899
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011531955418204456,
+      "loss": 0.6131,
+      "step": 12900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011530868243729656,
+      "loss": 0.526,
+      "step": 12901
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011529781050726249,
+      "loss": 0.5655,
+      "step": 12902
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011528693839207392,
+      "loss": 0.5944,
+      "step": 12903
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011527606609186241,
+      "loss": 0.5636,
+      "step": 12904
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001152651936067596,
+      "loss": 0.6227,
+      "step": 12905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011525432093689703,
+      "loss": 0.6104,
+      "step": 12906
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011524344808240636,
+      "loss": 0.5997,
+      "step": 12907
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011523257504341918,
+      "loss": 0.5896,
+      "step": 12908
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011522170182006701,
+      "loss": 0.598,
+      "step": 12909
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011521082841248155,
+      "loss": 0.6571,
+      "step": 12910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011519995482079434,
+      "loss": 0.5175,
+      "step": 12911
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011518908104513699,
+      "loss": 0.5794,
+      "step": 12912
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011517820708564116,
+      "loss": 0.5337,
+      "step": 12913
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001151673329424384,
+      "loss": 0.7244,
+      "step": 12914
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011515645861566037,
+      "loss": 0.614,
+      "step": 12915
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011514558410543867,
+      "loss": 0.5743,
+      "step": 12916
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001151347094119049,
+      "loss": 0.5715,
+      "step": 12917
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011512383453519074,
+      "loss": 0.6302,
+      "step": 12918
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011511295947542773,
+      "loss": 0.6228,
+      "step": 12919
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011510208423274754,
+      "loss": 0.6745,
+      "step": 12920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011509120880728177,
+      "loss": 0.5934,
+      "step": 12921
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001150803331991621,
+      "loss": 0.5524,
+      "step": 12922
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011506945740852011,
+      "loss": 0.5686,
+      "step": 12923
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011505858143548743,
+      "loss": 0.5394,
+      "step": 12924
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011504770528019575,
+      "loss": 0.4777,
+      "step": 12925
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011503682894277666,
+      "loss": 0.6435,
+      "step": 12926
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011502595242336182,
+      "loss": 0.5211,
+      "step": 12927
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011501507572208288,
+      "loss": 0.6517,
+      "step": 12928
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011500419883907145,
+      "loss": 0.5472,
+      "step": 12929
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011499332177445921,
+      "loss": 0.7196,
+      "step": 12930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001149824445283778,
+      "loss": 0.6436,
+      "step": 12931
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011497156710095884,
+      "loss": 0.6362,
+      "step": 12932
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011496068949233405,
+      "loss": 0.6726,
+      "step": 12933
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011494981170263501,
+      "loss": 0.566,
+      "step": 12934
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011493893373199345,
+      "loss": 0.5814,
+      "step": 12935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011492805558054098,
+      "loss": 0.5628,
+      "step": 12936
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011491717724840927,
+      "loss": 0.5715,
+      "step": 12937
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011490629873573,
+      "loss": 0.5805,
+      "step": 12938
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001148954200426348,
+      "loss": 0.5912,
+      "step": 12939
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001148845411692554,
+      "loss": 0.5915,
+      "step": 12940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011487366211572344,
+      "loss": 0.4961,
+      "step": 12941
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011486278288217054,
+      "loss": 0.6125,
+      "step": 12942
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011485190346872845,
+      "loss": 0.6056,
+      "step": 12943
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011484102387552884,
+      "loss": 0.5249,
+      "step": 12944
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011483014410270335,
+      "loss": 0.5559,
+      "step": 12945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011481926415038367,
+      "loss": 0.5859,
+      "step": 12946
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011480838401870148,
+      "loss": 0.5651,
+      "step": 12947
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011479750370778852,
+      "loss": 0.5968,
+      "step": 12948
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011478662321777643,
+      "loss": 0.6223,
+      "step": 12949
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011477574254879689,
+      "loss": 0.5353,
+      "step": 12950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011476486170098163,
+      "loss": 0.5899,
+      "step": 12951
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011475398067446231,
+      "loss": 0.6772,
+      "step": 12952
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011474309946937066,
+      "loss": 0.5326,
+      "step": 12953
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011473221808583835,
+      "loss": 0.5633,
+      "step": 12954
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011472133652399708,
+      "loss": 0.6052,
+      "step": 12955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011471045478397859,
+      "loss": 0.5332,
+      "step": 12956
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011469957286591453,
+      "loss": 0.6707,
+      "step": 12957
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011468869076993666,
+      "loss": 0.7251,
+      "step": 12958
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011467780849617669,
+      "loss": 0.6786,
+      "step": 12959
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011466692604476625,
+      "loss": 0.5721,
+      "step": 12960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011465604341583715,
+      "loss": 0.6401,
+      "step": 12961
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011464516060952108,
+      "loss": 0.5912,
+      "step": 12962
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011463427762594974,
+      "loss": 0.5636,
+      "step": 12963
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011462339446525486,
+      "loss": 0.6019,
+      "step": 12964
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011461251112756812,
+      "loss": 0.5331,
+      "step": 12965
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011460162761302134,
+      "loss": 0.5765,
+      "step": 12966
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011459074392174618,
+      "loss": 0.7175,
+      "step": 12967
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011457986005387435,
+      "loss": 0.6447,
+      "step": 12968
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011456897600953764,
+      "loss": 0.5318,
+      "step": 12969
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011455809178886774,
+      "loss": 0.6926,
+      "step": 12970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011454720739199643,
+      "loss": 0.6339,
+      "step": 12971
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001145363228190554,
+      "loss": 0.5624,
+      "step": 12972
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001145254380701764,
+      "loss": 0.7537,
+      "step": 12973
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011451455314549117,
+      "loss": 0.5827,
+      "step": 12974
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011450366804513149,
+      "loss": 0.6008,
+      "step": 12975
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011449278276922906,
+      "loss": 0.5414,
+      "step": 12976
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011448189731791568,
+      "loss": 0.6814,
+      "step": 12977
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011447101169132302,
+      "loss": 0.5111,
+      "step": 12978
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011446012588958291,
+      "loss": 0.5467,
+      "step": 12979
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011444923991282707,
+      "loss": 0.6178,
+      "step": 12980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011443835376118726,
+      "loss": 0.5164,
+      "step": 12981
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011442746743479524,
+      "loss": 0.6049,
+      "step": 12982
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011441658093378275,
+      "loss": 0.5199,
+      "step": 12983
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001144056942582816,
+      "loss": 0.6465,
+      "step": 12984
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011439480740842354,
+      "loss": 0.7175,
+      "step": 12985
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011438392038434032,
+      "loss": 0.6192,
+      "step": 12986
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011437303318616371,
+      "loss": 0.6306,
+      "step": 12987
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011436214581402547,
+      "loss": 0.6317,
+      "step": 12988
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011435125826805739,
+      "loss": 0.5454,
+      "step": 12989
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011434037054839127,
+      "loss": 0.5404,
+      "step": 12990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011432948265515884,
+      "loss": 0.5512,
+      "step": 12991
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011431859458849193,
+      "loss": 0.6626,
+      "step": 12992
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011430770634852226,
+      "loss": 0.5866,
+      "step": 12993
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011429681793538167,
+      "loss": 0.5764,
+      "step": 12994
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011428592934920195,
+      "loss": 0.5745,
+      "step": 12995
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011427504059011482,
+      "loss": 0.6191,
+      "step": 12996
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011426415165825212,
+      "loss": 0.627,
+      "step": 12997
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011425326255374565,
+      "loss": 0.5406,
+      "step": 12998
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011424237327672718,
+      "loss": 0.5204,
+      "step": 12999
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011423148382732853,
+      "loss": 0.6608,
+      "step": 13000
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 0.5851190090179443,
+      "eval_runtime": 845.7966,
+      "eval_samples_per_second": 28.447,
+      "eval_steps_per_second": 14.223,
+      "step": 13000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011422059420568147,
+      "loss": 0.6121,
+      "step": 13001
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011420970441191782,
+      "loss": 0.6125,
+      "step": 13002
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001141988144461694,
+      "loss": 0.5762,
+      "step": 13003
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011418792430856797,
+      "loss": 0.5368,
+      "step": 13004
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001141770339992454,
+      "loss": 0.51,
+      "step": 13005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011416614351833341,
+      "loss": 0.5421,
+      "step": 13006
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011415525286596389,
+      "loss": 0.5453,
+      "step": 13007
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001141443620422686,
+      "loss": 0.701,
+      "step": 13008
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011413347104737939,
+      "loss": 0.5803,
+      "step": 13009
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011412257988142809,
+      "loss": 0.6744,
+      "step": 13010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011411168854454648,
+      "loss": 0.6085,
+      "step": 13011
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001141007970368664,
+      "loss": 0.6354,
+      "step": 13012
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011408990535851967,
+      "loss": 0.6221,
+      "step": 13013
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001140790135096381,
+      "loss": 0.5313,
+      "step": 13014
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011406812149035358,
+      "loss": 0.5352,
+      "step": 13015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011405722930079783,
+      "loss": 0.5452,
+      "step": 13016
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011404633694110281,
+      "loss": 0.5719,
+      "step": 13017
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011403544441140026,
+      "loss": 0.644,
+      "step": 13018
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011402455171182205,
+      "loss": 0.5567,
+      "step": 13019
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011401365884250001,
+      "loss": 0.5975,
+      "step": 13020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011400276580356597,
+      "loss": 0.6583,
+      "step": 13021
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001139918725951518,
+      "loss": 0.5042,
+      "step": 13022
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011398097921738935,
+      "loss": 0.583,
+      "step": 13023
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001139700856704104,
+      "loss": 0.5484,
+      "step": 13024
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001139591919543469,
+      "loss": 0.4963,
+      "step": 13025
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011394829806933061,
+      "loss": 0.5967,
+      "step": 13026
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011393740401549344,
+      "loss": 0.667,
+      "step": 13027
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011392650979296722,
+      "loss": 0.5892,
+      "step": 13028
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011391561540188379,
+      "loss": 0.5235,
+      "step": 13029
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011390472084237502,
+      "loss": 0.6565,
+      "step": 13030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011389382611457282,
+      "loss": 0.5884,
+      "step": 13031
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011388293121860896,
+      "loss": 0.609,
+      "step": 13032
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011387203615461539,
+      "loss": 0.5884,
+      "step": 13033
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001138611409227239,
+      "loss": 0.633,
+      "step": 13034
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011385024552306644,
+      "loss": 0.6291,
+      "step": 13035
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011383934995577483,
+      "loss": 0.7102,
+      "step": 13036
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011382845422098095,
+      "loss": 0.5228,
+      "step": 13037
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011381755831881666,
+      "loss": 0.5666,
+      "step": 13038
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011380666224941388,
+      "loss": 0.6008,
+      "step": 13039
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011379576601290445,
+      "loss": 0.4714,
+      "step": 13040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011378486960942027,
+      "loss": 0.7826,
+      "step": 13041
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011377397303909321,
+      "loss": 0.6261,
+      "step": 13042
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011376307630205516,
+      "loss": 0.5641,
+      "step": 13043
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011375217939843803,
+      "loss": 0.6333,
+      "step": 13044
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011374128232837365,
+      "loss": 0.5934,
+      "step": 13045
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011373038509199399,
+      "loss": 0.6545,
+      "step": 13046
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011371948768943086,
+      "loss": 0.536,
+      "step": 13047
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011370859012081623,
+      "loss": 0.6223,
+      "step": 13048
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011369769238628194,
+      "loss": 0.6112,
+      "step": 13049
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011368679448595991,
+      "loss": 0.6716,
+      "step": 13050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011367589641998207,
+      "loss": 0.5503,
+      "step": 13051
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011366499818848024,
+      "loss": 0.6407,
+      "step": 13052
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011365409979158646,
+      "loss": 0.6084,
+      "step": 13053
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001136432012294325,
+      "loss": 0.5448,
+      "step": 13054
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011363230250215036,
+      "loss": 0.6985,
+      "step": 13055
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001136214036098719,
+      "loss": 0.6148,
+      "step": 13056
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011361050455272902,
+      "loss": 0.689,
+      "step": 13057
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001135996053308537,
+      "loss": 0.5054,
+      "step": 13058
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011358870594437781,
+      "loss": 0.7287,
+      "step": 13059
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001135778063934333,
+      "loss": 0.5765,
+      "step": 13060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011356690667815206,
+      "loss": 0.6141,
+      "step": 13061
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011355600679866601,
+      "loss": 0.6556,
+      "step": 13062
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001135451067551071,
+      "loss": 0.5577,
+      "step": 13063
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011353420654760727,
+      "loss": 0.6892,
+      "step": 13064
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011352330617629839,
+      "loss": 0.5814,
+      "step": 13065
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011351240564131243,
+      "loss": 0.6213,
+      "step": 13066
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011350150494278131,
+      "loss": 0.4958,
+      "step": 13067
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000113490604080837,
+      "loss": 0.5623,
+      "step": 13068
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011347970305561142,
+      "loss": 0.6407,
+      "step": 13069
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011346880186723645,
+      "loss": 0.6012,
+      "step": 13070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011345790051584413,
+      "loss": 0.6908,
+      "step": 13071
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011344699900156632,
+      "loss": 0.5745,
+      "step": 13072
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011343609732453502,
+      "loss": 0.5434,
+      "step": 13073
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011342519548488215,
+      "loss": 0.543,
+      "step": 13074
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011341429348273964,
+      "loss": 0.4854,
+      "step": 13075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011340339131823949,
+      "loss": 0.5781,
+      "step": 13076
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001133924889915136,
+      "loss": 0.6434,
+      "step": 13077
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011338158650269398,
+      "loss": 0.5916,
+      "step": 13078
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011337068385191254,
+      "loss": 0.6622,
+      "step": 13079
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011335978103930125,
+      "loss": 0.6012,
+      "step": 13080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011334887806499208,
+      "loss": 0.5717,
+      "step": 13081
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000113337974929117,
+      "loss": 0.5248,
+      "step": 13082
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011332707163180793,
+      "loss": 0.6872,
+      "step": 13083
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001133161681731969,
+      "loss": 0.7026,
+      "step": 13084
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011330526455341584,
+      "loss": 0.6266,
+      "step": 13085
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011329436077259672,
+      "loss": 0.6193,
+      "step": 13086
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011328345683087152,
+      "loss": 0.6027,
+      "step": 13087
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011327255272837221,
+      "loss": 0.566,
+      "step": 13088
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011326164846523077,
+      "loss": 0.6394,
+      "step": 13089
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011325074404157916,
+      "loss": 0.6042,
+      "step": 13090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011323983945754942,
+      "loss": 0.6069,
+      "step": 13091
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011322893471327348,
+      "loss": 0.5704,
+      "step": 13092
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011321802980888328,
+      "loss": 0.6524,
+      "step": 13093
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011320712474451089,
+      "loss": 0.6373,
+      "step": 13094
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011319621952028829,
+      "loss": 0.615,
+      "step": 13095
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011318531413634741,
+      "loss": 0.5531,
+      "step": 13096
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011317440859282029,
+      "loss": 0.5661,
+      "step": 13097
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001131635028898389,
+      "loss": 0.609,
+      "step": 13098
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011315259702753527,
+      "loss": 0.5372,
+      "step": 13099
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011314169100604135,
+      "loss": 0.5704,
+      "step": 13100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011313078482548915,
+      "loss": 0.6499,
+      "step": 13101
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011311987848601071,
+      "loss": 0.5966,
+      "step": 13102
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011310897198773798,
+      "loss": 0.5399,
+      "step": 13103
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011309806533080303,
+      "loss": 0.5981,
+      "step": 13104
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001130871585153378,
+      "loss": 0.6043,
+      "step": 13105
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011307625154147432,
+      "loss": 0.6332,
+      "step": 13106
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011306534440934465,
+      "loss": 0.4995,
+      "step": 13107
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001130544371190807,
+      "loss": 0.5624,
+      "step": 13108
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011304352967081457,
+      "loss": 0.707,
+      "step": 13109
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011303262206467829,
+      "loss": 0.6326,
+      "step": 13110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011302171430080377,
+      "loss": 0.5949,
+      "step": 13111
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011301080637932316,
+      "loss": 0.5762,
+      "step": 13112
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001129998983003684,
+      "loss": 0.5253,
+      "step": 13113
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011298899006407151,
+      "loss": 0.6926,
+      "step": 13114
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011297808167056456,
+      "loss": 0.626,
+      "step": 13115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011296717311997955,
+      "loss": 0.5907,
+      "step": 13116
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011295626441244855,
+      "loss": 0.7046,
+      "step": 13117
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011294535554810354,
+      "loss": 0.6844,
+      "step": 13118
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011293444652707659,
+      "loss": 0.5554,
+      "step": 13119
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011292353734949972,
+      "loss": 0.567,
+      "step": 13120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011291262801550495,
+      "loss": 0.5556,
+      "step": 13121
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011290171852522435,
+      "loss": 0.5725,
+      "step": 13122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011289080887878993,
+      "loss": 0.7006,
+      "step": 13123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011287989907633378,
+      "loss": 0.6029,
+      "step": 13124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001128689891179879,
+      "loss": 0.5877,
+      "step": 13125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011285807900388436,
+      "loss": 0.6491,
+      "step": 13126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011284716873415521,
+      "loss": 0.5358,
+      "step": 13127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011283625830893247,
+      "loss": 0.517,
+      "step": 13128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011282534772834824,
+      "loss": 0.5469,
+      "step": 13129
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011281443699253453,
+      "loss": 0.6533,
+      "step": 13130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011280352610162342,
+      "loss": 0.5895,
+      "step": 13131
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011279261505574699,
+      "loss": 0.5047,
+      "step": 13132
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011278170385503726,
+      "loss": 0.6368,
+      "step": 13133
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011277079249962627,
+      "loss": 0.5336,
+      "step": 13134
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011275988098964617,
+      "loss": 0.6468,
+      "step": 13135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011274896932522894,
+      "loss": 0.67,
+      "step": 13136
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001127380575065067,
+      "loss": 0.6677,
+      "step": 13137
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001127271455336115,
+      "loss": 0.6063,
+      "step": 13138
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011271623340667541,
+      "loss": 0.6822,
+      "step": 13139
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011270532112583052,
+      "loss": 0.5908,
+      "step": 13140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011269440869120886,
+      "loss": 0.4942,
+      "step": 13141
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011268349610294256,
+      "loss": 0.7542,
+      "step": 13142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011267258336116369,
+      "loss": 0.7171,
+      "step": 13143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011266167046600429,
+      "loss": 0.5798,
+      "step": 13144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001126507574175965,
+      "loss": 0.6347,
+      "step": 13145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011263984421607236,
+      "loss": 0.6205,
+      "step": 13146
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011262893086156396,
+      "loss": 0.6248,
+      "step": 13147
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011261801735420343,
+      "loss": 0.7124,
+      "step": 13148
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001126071036941228,
+      "loss": 0.6283,
+      "step": 13149
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011259618988145419,
+      "loss": 0.5669,
+      "step": 13150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011258527591632972,
+      "loss": 0.5286,
+      "step": 13151
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011257436179888143,
+      "loss": 0.6011,
+      "step": 13152
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011256344752924146,
+      "loss": 0.5553,
+      "step": 13153
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011255253310754188,
+      "loss": 0.637,
+      "step": 13154
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011254161853391483,
+      "loss": 0.6807,
+      "step": 13155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011253070380849238,
+      "loss": 0.5265,
+      "step": 13156
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011251978893140662,
+      "loss": 0.5729,
+      "step": 13157
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011250887390278972,
+      "loss": 0.5808,
+      "step": 13158
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011249795872277371,
+      "loss": 0.5386,
+      "step": 13159
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011248704339149077,
+      "loss": 0.5351,
+      "step": 13160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011247612790907298,
+      "loss": 0.5481,
+      "step": 13161
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011246521227565243,
+      "loss": 0.6206,
+      "step": 13162
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011245429649136128,
+      "loss": 0.5623,
+      "step": 13163
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011244338055633161,
+      "loss": 0.5785,
+      "step": 13164
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011243246447069556,
+      "loss": 0.5587,
+      "step": 13165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011242154823458527,
+      "loss": 0.585,
+      "step": 13166
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011241063184813279,
+      "loss": 0.6131,
+      "step": 13167
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011239971531147033,
+      "loss": 0.5157,
+      "step": 13168
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011238879862473,
+      "loss": 0.6121,
+      "step": 13169
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011237788178804385,
+      "loss": 0.5387,
+      "step": 13170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011236696480154412,
+      "loss": 0.6182,
+      "step": 13171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011235604766536285,
+      "loss": 0.5519,
+      "step": 13172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011234513037963226,
+      "loss": 0.5387,
+      "step": 13173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011233421294448443,
+      "loss": 0.636,
+      "step": 13174
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011232329536005148,
+      "loss": 0.5874,
+      "step": 13175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001123123776264656,
+      "loss": 0.6172,
+      "step": 13176
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001123014597438589,
+      "loss": 0.5103,
+      "step": 13177
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011229054171236352,
+      "loss": 0.6532,
+      "step": 13178
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011227962353211166,
+      "loss": 0.6308,
+      "step": 13179
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011226870520323538,
+      "loss": 0.6103,
+      "step": 13180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011225778672586689,
+      "loss": 0.4752,
+      "step": 13181
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011224686810013832,
+      "loss": 0.626,
+      "step": 13182
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001122359493261818,
+      "loss": 0.6465,
+      "step": 13183
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011222503040412952,
+      "loss": 0.605,
+      "step": 13184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011221411133411362,
+      "loss": 0.5195,
+      "step": 13185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011220319211626624,
+      "loss": 0.6366,
+      "step": 13186
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011219227275071959,
+      "loss": 0.6428,
+      "step": 13187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001121813532376058,
+      "loss": 0.5597,
+      "step": 13188
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011217043357705701,
+      "loss": 0.5653,
+      "step": 13189
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001121595137692054,
+      "loss": 0.5581,
+      "step": 13190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011214859381418314,
+      "loss": 0.4973,
+      "step": 13191
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011213767371212239,
+      "loss": 0.6457,
+      "step": 13192
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011212675346315536,
+      "loss": 0.5403,
+      "step": 13193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011211583306741416,
+      "loss": 0.6242,
+      "step": 13194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000112104912525031,
+      "loss": 0.6017,
+      "step": 13195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011209399183613804,
+      "loss": 0.5455,
+      "step": 13196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001120830710008675,
+      "loss": 0.7306,
+      "step": 13197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011207215001935147,
+      "loss": 0.5629,
+      "step": 13198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011206122889172221,
+      "loss": 0.5773,
+      "step": 13199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011205030761811186,
+      "loss": 0.6864,
+      "step": 13200
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 0.5793399214744568,
+      "eval_runtime": 848.2239,
+      "eval_samples_per_second": 28.365,
+      "eval_steps_per_second": 14.183,
+      "step": 13200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011203938619865266,
+      "loss": 0.5297,
+      "step": 13201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011202846463347673,
+      "loss": 0.5666,
+      "step": 13202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011201754292271627,
+      "loss": 0.6346,
+      "step": 13203
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011200662106650348,
+      "loss": 0.5696,
+      "step": 13204
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011199569906497056,
+      "loss": 0.5319,
+      "step": 13205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001119847769182497,
+      "loss": 0.6567,
+      "step": 13206
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001119738546264731,
+      "loss": 0.5587,
+      "step": 13207
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011196293218977292,
+      "loss": 0.567,
+      "step": 13208
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011195200960828139,
+      "loss": 0.5777,
+      "step": 13209
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001119410868821307,
+      "loss": 0.4592,
+      "step": 13210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011193016401145304,
+      "loss": 0.5355,
+      "step": 13211
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011191924099638067,
+      "loss": 0.7052,
+      "step": 13212
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011190831783704569,
+      "loss": 0.5495,
+      "step": 13213
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011189739453358044,
+      "loss": 0.5566,
+      "step": 13214
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011188647108611703,
+      "loss": 0.5486,
+      "step": 13215
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011187554749478766,
+      "loss": 0.6267,
+      "step": 13216
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011186462375972462,
+      "loss": 0.6141,
+      "step": 13217
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011185369988106005,
+      "loss": 0.5928,
+      "step": 13218
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011184277585892622,
+      "loss": 0.643,
+      "step": 13219
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011183185169345534,
+      "loss": 0.684,
+      "step": 13220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011182092738477959,
+      "loss": 0.5114,
+      "step": 13221
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011181000293303121,
+      "loss": 0.5004,
+      "step": 13222
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011179907833834242,
+      "loss": 0.4771,
+      "step": 13223
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001117881536008455,
+      "loss": 0.7435,
+      "step": 13224
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011177722872067258,
+      "loss": 0.6671,
+      "step": 13225
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011176630369795593,
+      "loss": 0.5285,
+      "step": 13226
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011175537853282781,
+      "loss": 0.5664,
+      "step": 13227
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001117444532254204,
+      "loss": 0.6354,
+      "step": 13228
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011173352777586599,
+      "loss": 0.5932,
+      "step": 13229
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011172260218429676,
+      "loss": 0.6052,
+      "step": 13230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011171167645084495,
+      "loss": 0.6062,
+      "step": 13231
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011170075057564285,
+      "loss": 0.6624,
+      "step": 13232
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011168982455882264,
+      "loss": 0.5505,
+      "step": 13233
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011167889840051657,
+      "loss": 0.6812,
+      "step": 13234
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011166797210085694,
+      "loss": 0.5895,
+      "step": 13235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011165704565997593,
+      "loss": 0.5844,
+      "step": 13236
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011164611907800582,
+      "loss": 0.6609,
+      "step": 13237
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011163519235507884,
+      "loss": 0.5262,
+      "step": 13238
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011162426549132723,
+      "loss": 0.5853,
+      "step": 13239
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011161333848688329,
+      "loss": 0.6805,
+      "step": 13240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011160241134187921,
+      "loss": 0.6384,
+      "step": 13241
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001115914840564473,
+      "loss": 0.5674,
+      "step": 13242
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001115805566307198,
+      "loss": 0.646,
+      "step": 13243
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011156962906482893,
+      "loss": 0.5018,
+      "step": 13244
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000111558701358907,
+      "loss": 0.5018,
+      "step": 13245
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011154777351308627,
+      "loss": 0.5578,
+      "step": 13246
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011153684552749895,
+      "loss": 0.5562,
+      "step": 13247
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011152591740227738,
+      "loss": 0.5377,
+      "step": 13248
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011151498913755375,
+      "loss": 0.5624,
+      "step": 13249
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001115040607334604,
+      "loss": 0.7483,
+      "step": 13250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011149313219012955,
+      "loss": 0.5667,
+      "step": 13251
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001114822035076935,
+      "loss": 0.6565,
+      "step": 13252
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011147127468628449,
+      "loss": 0.6721,
+      "step": 13253
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011146034572603483,
+      "loss": 0.5998,
+      "step": 13254
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011144941662707681,
+      "loss": 0.5414,
+      "step": 13255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011143848738954265,
+      "loss": 0.5795,
+      "step": 13256
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011142755801356468,
+      "loss": 0.5841,
+      "step": 13257
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011141662849927517,
+      "loss": 0.6511,
+      "step": 13258
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011140569884680636,
+      "loss": 0.6957,
+      "step": 13259
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001113947690562906,
+      "loss": 0.6079,
+      "step": 13260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011138383912786015,
+      "loss": 0.5516,
+      "step": 13261
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001113729090616473,
+      "loss": 0.5289,
+      "step": 13262
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011136197885778435,
+      "loss": 0.4997,
+      "step": 13263
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011135104851640354,
+      "loss": 0.5652,
+      "step": 13264
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011134011803763724,
+      "loss": 0.583,
+      "step": 13265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011132918742161773,
+      "loss": 0.5129,
+      "step": 13266
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011131825666847724,
+      "loss": 0.622,
+      "step": 13267
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011130732577834814,
+      "loss": 0.5802,
+      "step": 13268
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011129639475136269,
+      "loss": 0.6555,
+      "step": 13269
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011128546358765324,
+      "loss": 0.5816,
+      "step": 13270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011127453228735203,
+      "loss": 0.6134,
+      "step": 13271
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001112636008505914,
+      "loss": 0.5247,
+      "step": 13272
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011125266927750367,
+      "loss": 0.5297,
+      "step": 13273
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011124173756822111,
+      "loss": 0.6082,
+      "step": 13274
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011123080572287608,
+      "loss": 0.571,
+      "step": 13275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011121987374160083,
+      "loss": 0.6312,
+      "step": 13276
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011120894162452772,
+      "loss": 0.6045,
+      "step": 13277
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011119800937178906,
+      "loss": 0.5576,
+      "step": 13278
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011118707698351714,
+      "loss": 0.4863,
+      "step": 13279
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011117614445984432,
+      "loss": 0.5402,
+      "step": 13280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011116521180090289,
+      "loss": 0.5749,
+      "step": 13281
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011115427900682518,
+      "loss": 0.5878,
+      "step": 13282
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001111433460777435,
+      "loss": 0.7085,
+      "step": 13283
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011113241301379021,
+      "loss": 0.5547,
+      "step": 13284
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011112147981509757,
+      "loss": 0.5683,
+      "step": 13285
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011111054648179799,
+      "loss": 0.5933,
+      "step": 13286
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011109961301402374,
+      "loss": 0.5509,
+      "step": 13287
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011108867941190719,
+      "loss": 0.5723,
+      "step": 13288
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011107774567558067,
+      "loss": 0.5902,
+      "step": 13289
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011106681180517646,
+      "loss": 0.5935,
+      "step": 13290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011105587780082695,
+      "loss": 0.5468,
+      "step": 13291
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011104494366266446,
+      "loss": 0.6168,
+      "step": 13292
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011103400939082135,
+      "loss": 0.4754,
+      "step": 13293
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011102307498542996,
+      "loss": 0.5885,
+      "step": 13294
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011101214044662258,
+      "loss": 0.5923,
+      "step": 13295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011100120577453161,
+      "loss": 0.6325,
+      "step": 13296
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011099027096928938,
+      "loss": 0.6318,
+      "step": 13297
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011097933603102823,
+      "loss": 0.6545,
+      "step": 13298
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011096840095988053,
+      "loss": 0.631,
+      "step": 13299
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011095746575597859,
+      "loss": 0.5084,
+      "step": 13300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001109465304194548,
+      "loss": 0.5439,
+      "step": 13301
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001109355949504415,
+      "loss": 0.5732,
+      "step": 13302
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011092465934907104,
+      "loss": 0.6647,
+      "step": 13303
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011091372361547581,
+      "loss": 0.6115,
+      "step": 13304
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001109027877497881,
+      "loss": 0.4879,
+      "step": 13305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011089185175214037,
+      "loss": 0.6011,
+      "step": 13306
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011088091562266492,
+      "loss": 0.4683,
+      "step": 13307
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011086997936149408,
+      "loss": 0.6576,
+      "step": 13308
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011085904296876028,
+      "loss": 0.5511,
+      "step": 13309
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011084810644459585,
+      "loss": 0.6452,
+      "step": 13310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011083716978913319,
+      "loss": 0.5959,
+      "step": 13311
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011082623300250463,
+      "loss": 0.575,
+      "step": 13312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011081529608484256,
+      "loss": 0.7195,
+      "step": 13313
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011080435903627938,
+      "loss": 0.6202,
+      "step": 13314
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011079342185694743,
+      "loss": 0.5924,
+      "step": 13315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011078248454697908,
+      "loss": 0.5615,
+      "step": 13316
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011077154710650675,
+      "loss": 0.5779,
+      "step": 13317
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011076060953566276,
+      "loss": 0.5668,
+      "step": 13318
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011074967183457957,
+      "loss": 0.5648,
+      "step": 13319
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011073873400338948,
+      "loss": 0.5923,
+      "step": 13320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011072779604222492,
+      "loss": 0.6322,
+      "step": 13321
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011071685795121829,
+      "loss": 0.674,
+      "step": 13322
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011070591973050193,
+      "loss": 0.5796,
+      "step": 13323
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011069498138020827,
+      "loss": 0.5694,
+      "step": 13324
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011068404290046965,
+      "loss": 0.6777,
+      "step": 13325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011067310429141852,
+      "loss": 0.5676,
+      "step": 13326
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011066216555318727,
+      "loss": 0.5731,
+      "step": 13327
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011065122668590824,
+      "loss": 0.5972,
+      "step": 13328
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011064028768971389,
+      "loss": 0.5419,
+      "step": 13329
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011062934856473655,
+      "loss": 0.5369,
+      "step": 13330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011061840931110869,
+      "loss": 0.6581,
+      "step": 13331
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011060746992896269,
+      "loss": 0.4667,
+      "step": 13332
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011059653041843092,
+      "loss": 0.6602,
+      "step": 13333
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011058559077964582,
+      "loss": 0.6255,
+      "step": 13334
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001105746510127398,
+      "loss": 0.5942,
+      "step": 13335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011056371111784522,
+      "loss": 0.5463,
+      "step": 13336
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011055277109509454,
+      "loss": 0.5622,
+      "step": 13337
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011054183094462014,
+      "loss": 0.6051,
+      "step": 13338
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011053089066655447,
+      "loss": 0.4944,
+      "step": 13339
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011051995026102993,
+      "loss": 0.5246,
+      "step": 13340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011050900972817888,
+      "loss": 0.5834,
+      "step": 13341
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001104980690681338,
+      "loss": 0.5652,
+      "step": 13342
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011048712828102708,
+      "loss": 0.4724,
+      "step": 13343
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011047618736699116,
+      "loss": 0.6876,
+      "step": 13344
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011046524632615847,
+      "loss": 0.6461,
+      "step": 13345
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011045430515866137,
+      "loss": 0.6435,
+      "step": 13346
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011044336386463237,
+      "loss": 0.5994,
+      "step": 13347
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011043242244420385,
+      "loss": 0.5809,
+      "step": 13348
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011042148089750821,
+      "loss": 0.6527,
+      "step": 13349
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011041053922467794,
+      "loss": 0.6107,
+      "step": 13350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011039959742584542,
+      "loss": 0.5817,
+      "step": 13351
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011038865550114314,
+      "loss": 0.5502,
+      "step": 13352
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001103777134507035,
+      "loss": 0.6745,
+      "step": 13353
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011036677127465889,
+      "loss": 0.5974,
+      "step": 13354
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011035582897314181,
+      "loss": 0.6242,
+      "step": 13355
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011034488654628468,
+      "loss": 0.5808,
+      "step": 13356
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011033394399421994,
+      "loss": 0.5807,
+      "step": 13357
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011032300131708004,
+      "loss": 0.6248,
+      "step": 13358
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011031205851499737,
+      "loss": 0.5696,
+      "step": 13359
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011030111558810448,
+      "loss": 0.5199,
+      "step": 13360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011029017253653371,
+      "loss": 0.5376,
+      "step": 13361
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011027922936041757,
+      "loss": 0.5926,
+      "step": 13362
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011026828605988849,
+      "loss": 0.5041,
+      "step": 13363
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001102573426350789,
+      "loss": 0.6768,
+      "step": 13364
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011024639908612129,
+      "loss": 0.4949,
+      "step": 13365
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001102354554131481,
+      "loss": 0.6497,
+      "step": 13366
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011022451161629175,
+      "loss": 0.5373,
+      "step": 13367
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011021356769568476,
+      "loss": 0.5648,
+      "step": 13368
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011020262365145952,
+      "loss": 0.6372,
+      "step": 13369
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011019167948374855,
+      "loss": 0.4925,
+      "step": 13370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001101807351926843,
+      "loss": 0.5234,
+      "step": 13371
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011016979077839918,
+      "loss": 0.6976,
+      "step": 13372
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001101588462410257,
+      "loss": 0.5396,
+      "step": 13373
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011014790158069632,
+      "loss": 0.5366,
+      "step": 13374
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011013695679754353,
+      "loss": 0.5622,
+      "step": 13375
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011012601189169975,
+      "loss": 0.5775,
+      "step": 13376
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011011506686329745,
+      "loss": 0.5784,
+      "step": 13377
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011010412171246915,
+      "loss": 0.6091,
+      "step": 13378
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001100931764393473,
+      "loss": 0.5982,
+      "step": 13379
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011008223104406434,
+      "loss": 0.6417,
+      "step": 13380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001100712855267528,
+      "loss": 0.6164,
+      "step": 13381
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011006033988754514,
+      "loss": 0.5966,
+      "step": 13382
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011004939412657383,
+      "loss": 0.6458,
+      "step": 13383
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011003844824397134,
+      "loss": 0.5629,
+      "step": 13384
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011002750223987014,
+      "loss": 0.658,
+      "step": 13385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011001655611440278,
+      "loss": 0.5554,
+      "step": 13386
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011000560986770167,
+      "loss": 0.4935,
+      "step": 13387
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010999466349989935,
+      "loss": 0.5704,
+      "step": 13388
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010998371701112828,
+      "loss": 0.6432,
+      "step": 13389
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010997277040152096,
+      "loss": 0.6071,
+      "step": 13390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010996182367120989,
+      "loss": 0.6232,
+      "step": 13391
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001099508768203275,
+      "loss": 0.5692,
+      "step": 13392
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010993992984900634,
+      "loss": 0.5541,
+      "step": 13393
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001099289827573789,
+      "loss": 0.6477,
+      "step": 13394
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010991803554557768,
+      "loss": 0.5744,
+      "step": 13395
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010990708821373519,
+      "loss": 0.6114,
+      "step": 13396
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010989614076198385,
+      "loss": 0.6534,
+      "step": 13397
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010988519319045628,
+      "loss": 0.5509,
+      "step": 13398
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001098742454992849,
+      "loss": 0.5912,
+      "step": 13399
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010986329768860219,
+      "loss": 0.6046,
+      "step": 13400
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 0.5748456716537476,
+      "eval_runtime": 845.4805,
+      "eval_samples_per_second": 28.457,
+      "eval_steps_per_second": 14.229,
+      "step": 13400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010985234975854075,
+      "loss": 0.503,
+      "step": 13401
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010984140170923301,
+      "loss": 0.5551,
+      "step": 13402
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010983045354081151,
+      "loss": 0.4956,
+      "step": 13403
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010981950525340879,
+      "loss": 0.5699,
+      "step": 13404
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010980855684715726,
+      "loss": 0.5388,
+      "step": 13405
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010979760832218953,
+      "loss": 0.5158,
+      "step": 13406
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010978665967863806,
+      "loss": 0.6551,
+      "step": 13407
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010977571091663541,
+      "loss": 0.4845,
+      "step": 13408
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010976476203631407,
+      "loss": 0.6738,
+      "step": 13409
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010975381303780654,
+      "loss": 0.5851,
+      "step": 13410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010974286392124538,
+      "loss": 0.6353,
+      "step": 13411
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010973191468676306,
+      "loss": 0.6091,
+      "step": 13412
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010972096533449217,
+      "loss": 0.6079,
+      "step": 13413
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010971001586456517,
+      "loss": 0.6257,
+      "step": 13414
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001096990662771146,
+      "loss": 0.4808,
+      "step": 13415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010968811657227303,
+      "loss": 0.6515,
+      "step": 13416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010967716675017294,
+      "loss": 0.6572,
+      "step": 13417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010966621681094684,
+      "loss": 0.6599,
+      "step": 13418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010965526675472735,
+      "loss": 0.5496,
+      "step": 13419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010964431658164689,
+      "loss": 0.6154,
+      "step": 13420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010963336629183809,
+      "loss": 0.5473,
+      "step": 13421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010962241588543345,
+      "loss": 0.6173,
+      "step": 13422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010961146536256548,
+      "loss": 0.5655,
+      "step": 13423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010960051472336676,
+      "loss": 0.4905,
+      "step": 13424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010958956396796977,
+      "loss": 0.5874,
+      "step": 13425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010957861309650714,
+      "loss": 0.6035,
+      "step": 13426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010956766210911133,
+      "loss": 0.6594,
+      "step": 13427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010955671100591493,
+      "loss": 0.5173,
+      "step": 13428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010954575978705046,
+      "loss": 0.5231,
+      "step": 13429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010953480845265047,
+      "loss": 0.651,
+      "step": 13430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010952385700284753,
+      "loss": 0.4995,
+      "step": 13431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001095129054377742,
+      "loss": 0.654,
+      "step": 13432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010950195375756294,
+      "loss": 0.5596,
+      "step": 13433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010949100196234641,
+      "loss": 0.6297,
+      "step": 13434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010948005005225713,
+      "loss": 0.5755,
+      "step": 13435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001094690980274276,
+      "loss": 0.5298,
+      "step": 13436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010945814588799044,
+      "loss": 0.6382,
+      "step": 13437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010944719363407816,
+      "loss": 0.5881,
+      "step": 13438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010943624126582337,
+      "loss": 0.5181,
+      "step": 13439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010942528878335863,
+      "loss": 0.5514,
+      "step": 13440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010941433618681643,
+      "loss": 0.6743,
+      "step": 13441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010940338347632941,
+      "loss": 0.6638,
+      "step": 13442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010939243065203008,
+      "loss": 0.6144,
+      "step": 13443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010938147771405104,
+      "loss": 0.5719,
+      "step": 13444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010937052466252487,
+      "loss": 0.6065,
+      "step": 13445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010935957149758409,
+      "loss": 0.5766,
+      "step": 13446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010934861821936131,
+      "loss": 0.5847,
+      "step": 13447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010933766482798909,
+      "loss": 0.5473,
+      "step": 13448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010932671132359997,
+      "loss": 0.6244,
+      "step": 13449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010931575770632659,
+      "loss": 0.5315,
+      "step": 13450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010930480397630145,
+      "loss": 0.5508,
+      "step": 13451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010929385013365718,
+      "loss": 0.6597,
+      "step": 13452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010928289617852636,
+      "loss": 0.6105,
+      "step": 13453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010927194211104153,
+      "loss": 0.6699,
+      "step": 13454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001092609879313353,
+      "loss": 0.4949,
+      "step": 13455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010925003363954025,
+      "loss": 0.5392,
+      "step": 13456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010923907923578898,
+      "loss": 0.5417,
+      "step": 13457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010922812472021402,
+      "loss": 0.697,
+      "step": 13458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010921717009294801,
+      "loss": 0.6435,
+      "step": 13459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001092062153541235,
+      "loss": 0.5734,
+      "step": 13460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001091952605038731,
+      "loss": 0.5048,
+      "step": 13461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001091843055423294,
+      "loss": 0.6051,
+      "step": 13462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010917335046962495,
+      "loss": 0.6676,
+      "step": 13463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010916239528589245,
+      "loss": 0.6562,
+      "step": 13464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010915143999126438,
+      "loss": 0.5089,
+      "step": 13465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010914048458587338,
+      "loss": 0.5514,
+      "step": 13466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010912952906985207,
+      "loss": 0.6827,
+      "step": 13467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010911857344333303,
+      "loss": 0.7268,
+      "step": 13468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010910761770644881,
+      "loss": 0.5542,
+      "step": 13469
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001090966618593321,
+      "loss": 0.6107,
+      "step": 13470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010908570590211542,
+      "loss": 0.5809,
+      "step": 13471
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010907474983493144,
+      "loss": 0.5381,
+      "step": 13472
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010906379365791276,
+      "loss": 0.5836,
+      "step": 13473
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001090528373711919,
+      "loss": 0.5213,
+      "step": 13474
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010904188097490158,
+      "loss": 0.5573,
+      "step": 13475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010903092446917431,
+      "loss": 0.4617,
+      "step": 13476
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010901996785414281,
+      "loss": 0.5198,
+      "step": 13477
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001090090111299396,
+      "loss": 0.5725,
+      "step": 13478
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010899805429669732,
+      "loss": 0.5197,
+      "step": 13479
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001089870973545486,
+      "loss": 0.6361,
+      "step": 13480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010897614030362604,
+      "loss": 0.561,
+      "step": 13481
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010896518314406225,
+      "loss": 0.5827,
+      "step": 13482
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001089542258759899,
+      "loss": 0.595,
+      "step": 13483
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001089432684995415,
+      "loss": 0.72,
+      "step": 13484
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001089323110148498,
+      "loss": 0.6275,
+      "step": 13485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010892135342204735,
+      "loss": 0.7014,
+      "step": 13486
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010891039572126676,
+      "loss": 0.5753,
+      "step": 13487
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001088994379126407,
+      "loss": 0.6381,
+      "step": 13488
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010888847999630176,
+      "loss": 0.7195,
+      "step": 13489
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010887752197238259,
+      "loss": 0.6634,
+      "step": 13490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010886656384101584,
+      "loss": 0.5663,
+      "step": 13491
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010885560560233406,
+      "loss": 0.6473,
+      "step": 13492
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010884464725646997,
+      "loss": 0.5823,
+      "step": 13493
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010883368880355614,
+      "loss": 0.5455,
+      "step": 13494
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010882273024372524,
+      "loss": 0.6165,
+      "step": 13495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010881177157710991,
+      "loss": 0.5967,
+      "step": 13496
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010880081280384273,
+      "loss": 0.5738,
+      "step": 13497
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010878985392405643,
+      "loss": 0.543,
+      "step": 13498
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010877889493788357,
+      "loss": 0.5955,
+      "step": 13499
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010876793584545679,
+      "loss": 0.5627,
+      "step": 13500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001087569766469088,
+      "loss": 0.6178,
+      "step": 13501
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010874601734237217,
+      "loss": 0.5704,
+      "step": 13502
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010873505793197963,
+      "loss": 0.5436,
+      "step": 13503
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010872409841586373,
+      "loss": 0.5257,
+      "step": 13504
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010871313879415714,
+      "loss": 0.6629,
+      "step": 13505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010870217906699256,
+      "loss": 0.5391,
+      "step": 13506
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010869121923450259,
+      "loss": 0.5654,
+      "step": 13507
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001086802592968199,
+      "loss": 0.5193,
+      "step": 13508
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010866929925407714,
+      "loss": 0.5685,
+      "step": 13509
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010865833910640695,
+      "loss": 0.7103,
+      "step": 13510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010864737885394201,
+      "loss": 0.5939,
+      "step": 13511
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010863641849681494,
+      "loss": 0.598,
+      "step": 13512
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010862545803515842,
+      "loss": 0.6164,
+      "step": 13513
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010861449746910512,
+      "loss": 0.4972,
+      "step": 13514
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010860353679878767,
+      "loss": 0.5626,
+      "step": 13515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010859257602433876,
+      "loss": 0.4851,
+      "step": 13516
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010858161514589105,
+      "loss": 0.5744,
+      "step": 13517
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010857065416357715,
+      "loss": 0.6076,
+      "step": 13518
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001085596930775298,
+      "loss": 0.5629,
+      "step": 13519
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010854873188788159,
+      "loss": 0.5328,
+      "step": 13520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010853777059476527,
+      "loss": 0.5986,
+      "step": 13521
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010852680919831346,
+      "loss": 0.5942,
+      "step": 13522
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010851584769865881,
+      "loss": 0.6095,
+      "step": 13523
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010850488609593406,
+      "loss": 0.631,
+      "step": 13524
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001084939243902718,
+      "loss": 0.5747,
+      "step": 13525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010848296258180475,
+      "loss": 0.5206,
+      "step": 13526
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010847200067066556,
+      "loss": 0.4797,
+      "step": 13527
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010846103865698696,
+      "loss": 0.6702,
+      "step": 13528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010845007654090155,
+      "loss": 0.6404,
+      "step": 13529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010843911432254206,
+      "loss": 0.4972,
+      "step": 13530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010842815200204115,
+      "loss": 0.5433,
+      "step": 13531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001084171895795315,
+      "loss": 0.5594,
+      "step": 13532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010840622705514582,
+      "loss": 0.5742,
+      "step": 13533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010839526442901675,
+      "loss": 0.6407,
+      "step": 13534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000108384301701277,
+      "loss": 0.5744,
+      "step": 13535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010837333887205927,
+      "loss": 0.6358,
+      "step": 13536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010836237594149621,
+      "loss": 0.5338,
+      "step": 13537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010835141290972053,
+      "loss": 0.5831,
+      "step": 13538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001083404497768649,
+      "loss": 0.6018,
+      "step": 13539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010832948654306204,
+      "loss": 0.4805,
+      "step": 13540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010831852320844462,
+      "loss": 0.4976,
+      "step": 13541
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010830755977314535,
+      "loss": 0.6393,
+      "step": 13542
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010829659623729692,
+      "loss": 0.5812,
+      "step": 13543
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000108285632601032,
+      "loss": 0.6132,
+      "step": 13544
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010827466886448329,
+      "loss": 0.5095,
+      "step": 13545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010826370502778354,
+      "loss": 0.538,
+      "step": 13546
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001082527410910654,
+      "loss": 0.5635,
+      "step": 13547
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010824177705446157,
+      "loss": 0.5315,
+      "step": 13548
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010823081291810479,
+      "loss": 0.6254,
+      "step": 13549
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010821984868212774,
+      "loss": 0.5176,
+      "step": 13550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010820888434666307,
+      "loss": 0.5256,
+      "step": 13551
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010819791991184358,
+      "loss": 0.5924,
+      "step": 13552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010818695537780192,
+      "loss": 0.5932,
+      "step": 13553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010817599074467082,
+      "loss": 0.602,
+      "step": 13554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010816502601258296,
+      "loss": 0.593,
+      "step": 13555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010815406118167107,
+      "loss": 0.5147,
+      "step": 13556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010814309625206786,
+      "loss": 0.5467,
+      "step": 13557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010813213122390604,
+      "loss": 0.4887,
+      "step": 13558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010812116609731834,
+      "loss": 0.6753,
+      "step": 13559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010811020087243746,
+      "loss": 0.589,
+      "step": 13560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010809923554939607,
+      "loss": 0.5885,
+      "step": 13561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010808827012832699,
+      "loss": 0.5144,
+      "step": 13562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010807730460936283,
+      "loss": 0.5043,
+      "step": 13563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010806633899263639,
+      "loss": 0.5891,
+      "step": 13564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010805537327828036,
+      "loss": 0.6741,
+      "step": 13565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010804440746642743,
+      "loss": 0.519,
+      "step": 13566
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001080334415572104,
+      "loss": 0.5821,
+      "step": 13567
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010802247555076193,
+      "loss": 0.5622,
+      "step": 13568
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010801150944721474,
+      "loss": 0.5292,
+      "step": 13569
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001080005432467016,
+      "loss": 0.6245,
+      "step": 13570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001079895769493552,
+      "loss": 0.6041,
+      "step": 13571
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010797861055530831,
+      "loss": 0.5077,
+      "step": 13572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00010796764406469365,
+      "loss": 0.5216,
+      "step": 13573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010795667747764388,
+      "loss": 0.5689,
+      "step": 13574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010794571079429183,
+      "loss": 0.5487,
+      "step": 13575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010793474401477018,
+      "loss": 0.4944,
+      "step": 13576
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001079237771392117,
+      "loss": 0.5862,
+      "step": 13577
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010791281016774908,
+      "loss": 0.5716,
+      "step": 13578
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010790184310051507,
+      "loss": 0.5708,
+      "step": 13579
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010789087593764246,
+      "loss": 0.5582,
+      "step": 13580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001078799086792639,
+      "loss": 0.5385,
+      "step": 13581
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010786894132551221,
+      "loss": 0.594,
+      "step": 13582
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001078579738765201,
+      "loss": 0.6865,
+      "step": 13583
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010784700633242029,
+      "loss": 0.6619,
+      "step": 13584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010783603869334555,
+      "loss": 0.6485,
+      "step": 13585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010782507095942865,
+      "loss": 0.5444,
+      "step": 13586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010781410313080227,
+      "loss": 0.5261,
+      "step": 13587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010780313520759921,
+      "loss": 0.6697,
+      "step": 13588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010779216718995218,
+      "loss": 0.4987,
+      "step": 13589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010778119907799398,
+      "loss": 0.4849,
+      "step": 13590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010777023087185734,
+      "loss": 0.5705,
+      "step": 13591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000107759262571675,
+      "loss": 0.6073,
+      "step": 13592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010774829417757973,
+      "loss": 0.6648,
+      "step": 13593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010773732568970423,
+      "loss": 0.664,
+      "step": 13594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010772635710818133,
+      "loss": 0.6358,
+      "step": 13595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010771538843314373,
+      "loss": 0.5497,
+      "step": 13596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010770441966472422,
+      "loss": 0.5998,
+      "step": 13597
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010769345080305556,
+      "loss": 0.6103,
+      "step": 13598
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010768248184827048,
+      "loss": 0.5571,
+      "step": 13599
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010767151280050179,
+      "loss": 0.5331,
+      "step": 13600
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 0.5691545009613037,
+      "eval_runtime": 853.2104,
+      "eval_samples_per_second": 28.199,
+      "eval_steps_per_second": 14.1,
+      "step": 13600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001076605436598822,
+      "loss": 0.5365,
+      "step": 13601
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010764957442654448,
+      "loss": 0.5327,
+      "step": 13602
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010763860510062145,
+      "loss": 0.5871,
+      "step": 13603
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010762763568224579,
+      "loss": 0.5505,
+      "step": 13604
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010761666617155034,
+      "loss": 0.6023,
+      "step": 13605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010760569656866786,
+      "loss": 0.6605,
+      "step": 13606
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010759472687373105,
+      "loss": 0.6413,
+      "step": 13607
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010758375708687276,
+      "loss": 0.5934,
+      "step": 13608
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010757278720822572,
+      "loss": 0.5848,
+      "step": 13609
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010756181723792272,
+      "loss": 0.5283,
+      "step": 13610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010755084717609651,
+      "loss": 0.5453,
+      "step": 13611
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010753987702287989,
+      "loss": 0.4971,
+      "step": 13612
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010752890677840564,
+      "loss": 0.6002,
+      "step": 13613
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010751793644280647,
+      "loss": 0.6057,
+      "step": 13614
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010750696601621526,
+      "loss": 0.66,
+      "step": 13615
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010749599549876472,
+      "loss": 0.6077,
+      "step": 13616
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010748502489058763,
+      "loss": 0.7188,
+      "step": 13617
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010747405419181681,
+      "loss": 0.5677,
+      "step": 13618
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010746308340258502,
+      "loss": 0.6069,
+      "step": 13619
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010745211252302502,
+      "loss": 0.5331,
+      "step": 13620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010744114155326965,
+      "loss": 0.5815,
+      "step": 13621
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010743017049345163,
+      "loss": 0.5324,
+      "step": 13622
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010741919934370378,
+      "loss": 0.5303,
+      "step": 13623
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010740822810415893,
+      "loss": 0.547,
+      "step": 13624
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010739725677494979,
+      "loss": 0.5178,
+      "step": 13625
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010738628535620918,
+      "loss": 0.6632,
+      "step": 13626
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001073753138480699,
+      "loss": 0.643,
+      "step": 13627
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010736434225066474,
+      "loss": 0.5441,
+      "step": 13628
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001073533705641265,
+      "loss": 0.5804,
+      "step": 13629
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010734239878858796,
+      "loss": 0.5827,
+      "step": 13630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010733142692418193,
+      "loss": 0.6546,
+      "step": 13631
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010732045497104116,
+      "loss": 0.5394,
+      "step": 13632
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001073094829292985,
+      "loss": 0.6336,
+      "step": 13633
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010729851079908675,
+      "loss": 0.6014,
+      "step": 13634
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010728753858053867,
+      "loss": 0.6816,
+      "step": 13635
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010727656627378708,
+      "loss": 0.5735,
+      "step": 13636
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001072655938789648,
+      "loss": 0.5837,
+      "step": 13637
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010725462139620459,
+      "loss": 0.5614,
+      "step": 13638
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010724364882563928,
+      "loss": 0.5461,
+      "step": 13639
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010723267616740168,
+      "loss": 0.5284,
+      "step": 13640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010722170342162461,
+      "loss": 0.5754,
+      "step": 13641
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010721073058844084,
+      "loss": 0.5098,
+      "step": 13642
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010719975766798318,
+      "loss": 0.6345,
+      "step": 13643
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010718878466038447,
+      "loss": 0.5076,
+      "step": 13644
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010717781156577747,
+      "loss": 0.6265,
+      "step": 13645
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010716683838429506,
+      "loss": 0.5091,
+      "step": 13646
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010715586511607001,
+      "loss": 0.5022,
+      "step": 13647
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010714489176123512,
+      "loss": 0.5053,
+      "step": 13648
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010713391831992323,
+      "loss": 0.5472,
+      "step": 13649
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010712294479226716,
+      "loss": 0.5837,
+      "step": 13650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001071119711783997,
+      "loss": 0.6143,
+      "step": 13651
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001071009974784537,
+      "loss": 0.4968,
+      "step": 13652
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010709002369256193,
+      "loss": 0.5716,
+      "step": 13653
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010707904982085726,
+      "loss": 0.5245,
+      "step": 13654
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001070680758634725,
+      "loss": 0.5452,
+      "step": 13655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010705710182054043,
+      "loss": 0.5602,
+      "step": 13656
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010704612769219393,
+      "loss": 0.6161,
+      "step": 13657
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010703515347856576,
+      "loss": 0.629,
+      "step": 13658
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010702417917978884,
+      "loss": 0.6183,
+      "step": 13659
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001070132047959959,
+      "loss": 0.4983,
+      "step": 13660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010700223032731981,
+      "loss": 0.5521,
+      "step": 13661
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010699125577389342,
+      "loss": 0.5199,
+      "step": 13662
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010698028113584949,
+      "loss": 0.589,
+      "step": 13663
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001069693064133209,
+      "loss": 0.6468,
+      "step": 13664
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010695833160644044,
+      "loss": 0.572,
+      "step": 13665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010694735671534101,
+      "loss": 0.5854,
+      "step": 13666
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010693638174015542,
+      "loss": 0.5797,
+      "step": 13667
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010692540668101646,
+      "loss": 0.5603,
+      "step": 13668
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000106914431538057,
+      "loss": 0.5859,
+      "step": 13669
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010690345631140987,
+      "loss": 0.6823,
+      "step": 13670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001068924810012079,
+      "loss": 0.5625,
+      "step": 13671
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010688150560758394,
+      "loss": 0.5382,
+      "step": 13672
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010687053013067083,
+      "loss": 0.5822,
+      "step": 13673
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001068595545706014,
+      "loss": 0.648,
+      "step": 13674
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001068485789275085,
+      "loss": 0.7084,
+      "step": 13675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010683760320152494,
+      "loss": 0.6376,
+      "step": 13676
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010682662739278362,
+      "loss": 0.5778,
+      "step": 13677
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010681565150141732,
+      "loss": 0.6424,
+      "step": 13678
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010680467552755893,
+      "loss": 0.5207,
+      "step": 13679
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001067936994713413,
+      "loss": 0.6012,
+      "step": 13680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010678272333289722,
+      "loss": 0.6454,
+      "step": 13681
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010677174711235962,
+      "loss": 0.717,
+      "step": 13682
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001067607708098613,
+      "loss": 0.6494,
+      "step": 13683
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001067497944255351,
+      "loss": 0.5442,
+      "step": 13684
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001067388179595139,
+      "loss": 0.6547,
+      "step": 13685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010672784141193052,
+      "loss": 0.5697,
+      "step": 13686
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010671686478291787,
+      "loss": 0.5461,
+      "step": 13687
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010670588807260875,
+      "loss": 0.592,
+      "step": 13688
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000106694911281136,
+      "loss": 0.6149,
+      "step": 13689
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010668393440863254,
+      "loss": 0.5074,
+      "step": 13690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010667295745523116,
+      "loss": 0.5444,
+      "step": 13691
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010666198042106478,
+      "loss": 0.4813,
+      "step": 13692
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010665100330626625,
+      "loss": 0.5828,
+      "step": 13693
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010664002611096835,
+      "loss": 0.6227,
+      "step": 13694
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010662904883530404,
+      "loss": 0.5199,
+      "step": 13695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010661807147940612,
+      "loss": 0.572,
+      "step": 13696
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010660709404340749,
+      "loss": 0.535,
+      "step": 13697
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010659611652744101,
+      "loss": 0.5322,
+      "step": 13698
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001065851389316395,
+      "loss": 0.5017,
+      "step": 13699
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010657416125613589,
+      "loss": 0.5685,
+      "step": 13700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010656318350106301,
+      "loss": 0.5146,
+      "step": 13701
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010655220566655371,
+      "loss": 0.775,
+      "step": 13702
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001065412277527409,
+      "loss": 0.7363,
+      "step": 13703
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001065302497597574,
+      "loss": 0.5129,
+      "step": 13704
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010651927168773614,
+      "loss": 0.5285,
+      "step": 13705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010650829353680996,
+      "loss": 0.6394,
+      "step": 13706
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010649731530711171,
+      "loss": 0.5766,
+      "step": 13707
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001064863369987743,
+      "loss": 0.5252,
+      "step": 13708
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010647535861193058,
+      "loss": 0.6062,
+      "step": 13709
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010646438014671345,
+      "loss": 0.5453,
+      "step": 13710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010645340160325578,
+      "loss": 0.5294,
+      "step": 13711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010644242298169043,
+      "loss": 0.5572,
+      "step": 13712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010643144428215028,
+      "loss": 0.6106,
+      "step": 13713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010642046550476821,
+      "loss": 0.529,
+      "step": 13714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010640948664967712,
+      "loss": 0.7143,
+      "step": 13715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010639850771700988,
+      "loss": 0.6449,
+      "step": 13716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010638752870689934,
+      "loss": 0.6372,
+      "step": 13717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010637654961947843,
+      "loss": 0.6347,
+      "step": 13718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010636557045488,
+      "loss": 0.5562,
+      "step": 13719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010635459121323697,
+      "loss": 0.5687,
+      "step": 13720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010634361189468219,
+      "loss": 0.5376,
+      "step": 13721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010633263249934855,
+      "loss": 0.4986,
+      "step": 13722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010632165302736896,
+      "loss": 0.5985,
+      "step": 13723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010631067347887631,
+      "loss": 0.5227,
+      "step": 13724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010629969385400345,
+      "loss": 0.5705,
+      "step": 13725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010628871415288332,
+      "loss": 0.561,
+      "step": 13726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010627773437564877,
+      "loss": 0.6701,
+      "step": 13727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010626675452243271,
+      "loss": 0.5694,
+      "step": 13728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010625577459336802,
+      "loss": 0.5946,
+      "step": 13729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001062447945885876,
+      "loss": 0.5957,
+      "step": 13730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010623381450822439,
+      "loss": 0.5473,
+      "step": 13731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010622283435241118,
+      "loss": 0.5994,
+      "step": 13732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010621185412128099,
+      "loss": 0.658,
+      "step": 13733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010620087381496661,
+      "loss": 0.5246,
+      "step": 13734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010618989343360102,
+      "loss": 0.6235,
+      "step": 13735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010617891297731708,
+      "loss": 0.5276,
+      "step": 13736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010616793244624767,
+      "loss": 0.5253,
+      "step": 13737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010615695184052577,
+      "loss": 0.6131,
+      "step": 13738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001061459711602842,
+      "loss": 0.6743,
+      "step": 13739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010613499040565587,
+      "loss": 0.5751,
+      "step": 13740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010612400957677373,
+      "loss": 0.586,
+      "step": 13741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010611302867377064,
+      "loss": 0.5844,
+      "step": 13742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010610204769677956,
+      "loss": 0.6754,
+      "step": 13743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010609106664593333,
+      "loss": 0.578,
+      "step": 13744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001060800855213649,
+      "loss": 0.5535,
+      "step": 13745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010606910432320718,
+      "loss": 0.6691,
+      "step": 13746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010605812305159304,
+      "loss": 0.5795,
+      "step": 13747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010604714170665544,
+      "loss": 0.5754,
+      "step": 13748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010603616028852727,
+      "loss": 0.5027,
+      "step": 13749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010602517879734142,
+      "loss": 0.6389,
+      "step": 13750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010601419723323082,
+      "loss": 0.5707,
+      "step": 13751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010600321559632842,
+      "loss": 0.5384,
+      "step": 13752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010599223388676706,
+      "loss": 0.6167,
+      "step": 13753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010598125210467973,
+      "loss": 0.6168,
+      "step": 13754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010597027025019927,
+      "loss": 0.547,
+      "step": 13755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010595928832345867,
+      "loss": 0.5028,
+      "step": 13756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010594830632459081,
+      "loss": 0.5314,
+      "step": 13757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001059373242537286,
+      "loss": 0.6673,
+      "step": 13758
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000105926342111005,
+      "loss": 0.5527,
+      "step": 13759
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010591535989655289,
+      "loss": 0.5285,
+      "step": 13760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001059043776105052,
+      "loss": 0.6569,
+      "step": 13761
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010589339525299488,
+      "loss": 0.575,
+      "step": 13762
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010588241282415481,
+      "loss": 0.5569,
+      "step": 13763
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010587143032411795,
+      "loss": 0.5629,
+      "step": 13764
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001058604477530172,
+      "loss": 0.5744,
+      "step": 13765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001058494651109855,
+      "loss": 0.5641,
+      "step": 13766
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010583848239815579,
+      "loss": 0.571,
+      "step": 13767
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010582749961466096,
+      "loss": 0.5057,
+      "step": 13768
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010581651676063398,
+      "loss": 0.5267,
+      "step": 13769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010580553383620775,
+      "loss": 0.5837,
+      "step": 13770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001057945508415152,
+      "loss": 0.594,
+      "step": 13771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010578356777668927,
+      "loss": 0.6896,
+      "step": 13772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010577258464186288,
+      "loss": 0.5436,
+      "step": 13773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000105761601437169,
+      "loss": 0.6405,
+      "step": 13774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010575061816274053,
+      "loss": 0.5134,
+      "step": 13775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010573963481871039,
+      "loss": 0.6063,
+      "step": 13776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010572865140521155,
+      "loss": 0.564,
+      "step": 13777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010571766792237692,
+      "loss": 0.6293,
+      "step": 13778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010570668437033946,
+      "loss": 0.5492,
+      "step": 13779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010569570074923212,
+      "loss": 0.5395,
+      "step": 13780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010568471705918777,
+      "loss": 0.56,
+      "step": 13781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010567373330033943,
+      "loss": 0.5385,
+      "step": 13782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010566274947281995,
+      "loss": 0.6081,
+      "step": 13783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010565176557676238,
+      "loss": 0.4943,
+      "step": 13784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001056407816122996,
+      "loss": 0.5906,
+      "step": 13785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010562979757956456,
+      "loss": 0.573,
+      "step": 13786
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010561881347869018,
+      "loss": 0.5831,
+      "step": 13787
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010560782930980943,
+      "loss": 0.5473,
+      "step": 13788
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010559684507305527,
+      "loss": 0.5755,
+      "step": 13789
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010558586076856064,
+      "loss": 0.5866,
+      "step": 13790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010557487639645842,
+      "loss": 0.6596,
+      "step": 13791
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010556389195688165,
+      "loss": 0.4786,
+      "step": 13792
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010555290744996324,
+      "loss": 0.6296,
+      "step": 13793
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010554192287583615,
+      "loss": 0.5485,
+      "step": 13794
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010553093823463331,
+      "loss": 0.5339,
+      "step": 13795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010551995352648767,
+      "loss": 0.6385,
+      "step": 13796
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001055089687515322,
+      "loss": 0.6005,
+      "step": 13797
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010549798390989983,
+      "loss": 0.5219,
+      "step": 13798
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010548699900172353,
+      "loss": 0.6142,
+      "step": 13799
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010547601402713627,
+      "loss": 0.5314,
+      "step": 13800
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 0.5640407204627991,
+      "eval_runtime": 853.0057,
+      "eval_samples_per_second": 28.206,
+      "eval_steps_per_second": 14.103,
+      "step": 13800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010546502898627098,
+      "loss": 0.5736,
+      "step": 13801
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010545404387926061,
+      "loss": 0.6372,
+      "step": 13802
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010544305870623814,
+      "loss": 0.6267,
+      "step": 13803
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001054320734673365,
+      "loss": 0.6665,
+      "step": 13804
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001054210881626887,
+      "loss": 0.5974,
+      "step": 13805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010541010279242763,
+      "loss": 0.5272,
+      "step": 13806
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010539911735668628,
+      "loss": 0.548,
+      "step": 13807
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010538813185559764,
+      "loss": 0.5722,
+      "step": 13808
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001053771462892946,
+      "loss": 0.589,
+      "step": 13809
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001053661606579102,
+      "loss": 0.6494,
+      "step": 13810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010535517496157734,
+      "loss": 0.5752,
+      "step": 13811
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010534418920042906,
+      "loss": 0.4711,
+      "step": 13812
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010533320337459824,
+      "loss": 0.5273,
+      "step": 13813
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010532221748421787,
+      "loss": 0.5263,
+      "step": 13814
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010531123152942096,
+      "loss": 0.508,
+      "step": 13815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010530024551034041,
+      "loss": 0.5822,
+      "step": 13816
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010528925942710926,
+      "loss": 0.5783,
+      "step": 13817
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010527827327986042,
+      "loss": 0.5429,
+      "step": 13818
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010526728706872686,
+      "loss": 0.6712,
+      "step": 13819
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010525630079384159,
+      "loss": 0.5193,
+      "step": 13820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010524531445533757,
+      "loss": 0.4862,
+      "step": 13821
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010523432805334771,
+      "loss": 0.527,
+      "step": 13822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010522334158800509,
+      "loss": 0.5252,
+      "step": 13823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001052123550594426,
+      "loss": 0.591,
+      "step": 13824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010520136846779325,
+      "loss": 0.5122,
+      "step": 13825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010519038181318999,
+      "loss": 0.4997,
+      "step": 13826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001051793950957658,
+      "loss": 0.5103,
+      "step": 13827
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010516840831565369,
+      "loss": 0.5605,
+      "step": 13828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010515742147298657,
+      "loss": 0.5046,
+      "step": 13829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001051464345678975,
+      "loss": 0.6018,
+      "step": 13830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010513544760051941,
+      "loss": 0.6755,
+      "step": 13831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010512446057098526,
+      "loss": 0.7032,
+      "step": 13832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010511347347942808,
+      "loss": 0.6849,
+      "step": 13833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010510248632598081,
+      "loss": 0.5757,
+      "step": 13834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010509149911077644,
+      "loss": 0.5505,
+      "step": 13835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000105080511833948,
+      "loss": 0.6118,
+      "step": 13836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010506952449562839,
+      "loss": 0.6465,
+      "step": 13837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010505853709595066,
+      "loss": 0.7372,
+      "step": 13838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010504754963504775,
+      "loss": 0.5852,
+      "step": 13839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010503656211305267,
+      "loss": 0.6469,
+      "step": 13840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001050255745300984,
+      "loss": 0.5708,
+      "step": 13841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010501458688631791,
+      "loss": 0.4868,
+      "step": 13842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010500359918184423,
+      "loss": 0.5653,
+      "step": 13843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010499261141681032,
+      "loss": 0.5652,
+      "step": 13844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010498162359134913,
+      "loss": 0.598,
+      "step": 13845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010497063570559374,
+      "loss": 0.5842,
+      "step": 13846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010495964775967704,
+      "loss": 0.5844,
+      "step": 13847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010494865975373211,
+      "loss": 0.6663,
+      "step": 13848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001049376716878919,
+      "loss": 0.6216,
+      "step": 13849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010492668356228936,
+      "loss": 0.5825,
+      "step": 13850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010491569537705757,
+      "loss": 0.5599,
+      "step": 13851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010490470713232947,
+      "loss": 0.5274,
+      "step": 13852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010489371882823804,
+      "loss": 0.586,
+      "step": 13853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010488273046491633,
+      "loss": 0.5674,
+      "step": 13854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010487174204249728,
+      "loss": 0.5769,
+      "step": 13855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010486075356111394,
+      "loss": 0.5814,
+      "step": 13856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010484976502089928,
+      "loss": 0.539,
+      "step": 13857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010483877642198625,
+      "loss": 0.6271,
+      "step": 13858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010482778776450793,
+      "loss": 0.569,
+      "step": 13859
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010481679904859727,
+      "loss": 0.5556,
+      "step": 13860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010480581027438732,
+      "loss": 0.5336,
+      "step": 13861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010479482144201101,
+      "loss": 0.5237,
+      "step": 13862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001047838325516014,
+      "loss": 0.5484,
+      "step": 13863
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010477284360329148,
+      "loss": 0.5718,
+      "step": 13864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001047618545972142,
+      "loss": 0.6003,
+      "step": 13865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010475086553350262,
+      "loss": 0.6776,
+      "step": 13866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010473987641228973,
+      "loss": 0.5367,
+      "step": 13867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010472888723370854,
+      "loss": 0.5948,
+      "step": 13868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010471789799789206,
+      "loss": 0.6204,
+      "step": 13869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010470690870497326,
+      "loss": 0.5541,
+      "step": 13870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001046959193550852,
+      "loss": 0.6323,
+      "step": 13871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010468492994836087,
+      "loss": 0.6237,
+      "step": 13872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010467394048493324,
+      "loss": 0.5331,
+      "step": 13873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010466295096493539,
+      "loss": 0.5598,
+      "step": 13874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010465196138850025,
+      "loss": 0.5938,
+      "step": 13875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010464097175576089,
+      "loss": 0.605,
+      "step": 13876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001046299820668503,
+      "loss": 0.6245,
+      "step": 13877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010461899232190148,
+      "loss": 0.4947,
+      "step": 13878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010460800252104748,
+      "loss": 0.5499,
+      "step": 13879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010459701266442126,
+      "loss": 0.5856,
+      "step": 13880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001045860227521559,
+      "loss": 0.5878,
+      "step": 13881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010457503278438436,
+      "loss": 0.6257,
+      "step": 13882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010456404276123966,
+      "loss": 0.6013,
+      "step": 13883
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010455305268285482,
+      "loss": 0.5648,
+      "step": 13884
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010454206254936287,
+      "loss": 0.7318,
+      "step": 13885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010453107236089685,
+      "loss": 0.5638,
+      "step": 13886
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010452008211758974,
+      "loss": 0.5691,
+      "step": 13887
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010450909181957455,
+      "loss": 0.6647,
+      "step": 13888
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010449810146698435,
+      "loss": 0.5632,
+      "step": 13889
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010448711105995212,
+      "loss": 0.5486,
+      "step": 13890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010447612059861084,
+      "loss": 0.556,
+      "step": 13891
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010446513008309364,
+      "loss": 0.5623,
+      "step": 13892
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010445413951353344,
+      "loss": 0.6939,
+      "step": 13893
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010444314889006335,
+      "loss": 0.5329,
+      "step": 13894
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010443215821281633,
+      "loss": 0.5157,
+      "step": 13895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010442116748192539,
+      "loss": 0.5804,
+      "step": 13896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010441017669752362,
+      "loss": 0.5889,
+      "step": 13897
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010439918585974399,
+      "loss": 0.5274,
+      "step": 13898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010438819496871957,
+      "loss": 0.5113,
+      "step": 13899
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010437720402458335,
+      "loss": 0.5204,
+      "step": 13900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010436621302746836,
+      "loss": 0.566,
+      "step": 13901
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010435522197750765,
+      "loss": 0.5922,
+      "step": 13902
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010434423087483425,
+      "loss": 0.5648,
+      "step": 13903
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010433323971958113,
+      "loss": 0.5406,
+      "step": 13904
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010432224851188141,
+      "loss": 0.654,
+      "step": 13905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010431125725186805,
+      "loss": 0.5965,
+      "step": 13906
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010430026593967412,
+      "loss": 0.6049,
+      "step": 13907
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010428927457543263,
+      "loss": 0.6397,
+      "step": 13908
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010427828315927662,
+      "loss": 0.6514,
+      "step": 13909
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010426729169133913,
+      "loss": 0.5775,
+      "step": 13910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010425630017175316,
+      "loss": 0.5523,
+      "step": 13911
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001042453086006518,
+      "loss": 0.5936,
+      "step": 13912
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010423431697816806,
+      "loss": 0.6906,
+      "step": 13913
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010422332530443493,
+      "loss": 0.498,
+      "step": 13914
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010421233357958551,
+      "loss": 0.5364,
+      "step": 13915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010420134180375281,
+      "loss": 0.5877,
+      "step": 13916
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010419034997706987,
+      "loss": 0.548,
+      "step": 13917
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010417935809966974,
+      "loss": 0.5976,
+      "step": 13918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010416836617168542,
+      "loss": 0.6839,
+      "step": 13919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010415737419325,
+      "loss": 0.6523,
+      "step": 13920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010414638216449648,
+      "loss": 0.5949,
+      "step": 13921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010413539008555791,
+      "loss": 0.5876,
+      "step": 13922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010412439795656734,
+      "loss": 0.5364,
+      "step": 13923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010411340577765778,
+      "loss": 0.5884,
+      "step": 13924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010410241354896233,
+      "loss": 0.6359,
+      "step": 13925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010409142127061402,
+      "loss": 0.546,
+      "step": 13926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010408042894274583,
+      "loss": 0.5854,
+      "step": 13927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010406943656549088,
+      "loss": 0.5471,
+      "step": 13928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010405844413898218,
+      "loss": 0.5343,
+      "step": 13929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010404745166335277,
+      "loss": 0.6112,
+      "step": 13930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010403645913873568,
+      "loss": 0.5576,
+      "step": 13931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010402546656526402,
+      "loss": 0.6426,
+      "step": 13932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010401447394307078,
+      "loss": 0.7129,
+      "step": 13933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010400348127228899,
+      "loss": 0.5453,
+      "step": 13934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010399248855305176,
+      "loss": 0.6861,
+      "step": 13935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001039814957854921,
+      "loss": 0.5463,
+      "step": 13936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001039705029697431,
+      "loss": 0.6862,
+      "step": 13937
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010395951010593777,
+      "loss": 0.5036,
+      "step": 13938
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010394851719420913,
+      "loss": 0.5677,
+      "step": 13939
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001039375242346903,
+      "loss": 0.6159,
+      "step": 13940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001039265312275143,
+      "loss": 0.6205,
+      "step": 13941
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010391553817281416,
+      "loss": 0.5546,
+      "step": 13942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010390454507072299,
+      "loss": 0.5338,
+      "step": 13943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010389355192137377,
+      "loss": 0.4852,
+      "step": 13944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010388255872489963,
+      "loss": 0.5482,
+      "step": 13945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010387156548143359,
+      "loss": 0.5527,
+      "step": 13946
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010386057219110867,
+      "loss": 0.4544,
+      "step": 13947
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010384957885405798,
+      "loss": 0.5843,
+      "step": 13948
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010383858547041454,
+      "loss": 0.5461,
+      "step": 13949
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010382759204031142,
+      "loss": 0.5617,
+      "step": 13950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010381659856388171,
+      "loss": 0.6208,
+      "step": 13951
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010380560504125839,
+      "loss": 0.56,
+      "step": 13952
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010379461147257459,
+      "loss": 0.6022,
+      "step": 13953
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010378361785796334,
+      "loss": 0.5303,
+      "step": 13954
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010377262419755769,
+      "loss": 0.58,
+      "step": 13955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010376163049149072,
+      "loss": 0.4697,
+      "step": 13956
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010375063673989547,
+      "loss": 0.5437,
+      "step": 13957
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010373964294290502,
+      "loss": 0.5357,
+      "step": 13958
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010372864910065243,
+      "loss": 0.5391,
+      "step": 13959
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010371765521327075,
+      "loss": 0.5141,
+      "step": 13960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010370666128089305,
+      "loss": 0.6656,
+      "step": 13961
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010369566730365239,
+      "loss": 0.5081,
+      "step": 13962
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010368467328168184,
+      "loss": 0.5085,
+      "step": 13963
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010367367921511446,
+      "loss": 0.517,
+      "step": 13964
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010366268510408333,
+      "loss": 0.5248,
+      "step": 13965
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010365169094872147,
+      "loss": 0.6062,
+      "step": 13966
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010364069674916199,
+      "loss": 0.5584,
+      "step": 13967
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010362970250553796,
+      "loss": 0.5845,
+      "step": 13968
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010361870821798242,
+      "loss": 0.5422,
+      "step": 13969
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010360771388662844,
+      "loss": 0.5967,
+      "step": 13970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001035967195116091,
+      "loss": 0.6388,
+      "step": 13971
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010358572509305749,
+      "loss": 0.609,
+      "step": 13972
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010357473063110661,
+      "loss": 0.5605,
+      "step": 13973
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010356373612588958,
+      "loss": 0.5248,
+      "step": 13974
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010355274157753948,
+      "loss": 0.5415,
+      "step": 13975
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010354174698618937,
+      "loss": 0.5662,
+      "step": 13976
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010353075235197231,
+      "loss": 0.5325,
+      "step": 13977
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010351975767502135,
+      "loss": 0.5923,
+      "step": 13978
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010350876295546963,
+      "loss": 0.567,
+      "step": 13979
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010349776819345014,
+      "loss": 0.5072,
+      "step": 13980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010348677338909604,
+      "loss": 0.5585,
+      "step": 13981
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010347577854254034,
+      "loss": 0.4972,
+      "step": 13982
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010346478365391612,
+      "loss": 0.5517,
+      "step": 13983
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001034537887233565,
+      "loss": 0.5293,
+      "step": 13984
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010344279375099449,
+      "loss": 0.6024,
+      "step": 13985
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001034317987369632,
+      "loss": 0.5628,
+      "step": 13986
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010342080368139573,
+      "loss": 0.5883,
+      "step": 13987
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010340980858442511,
+      "loss": 0.58,
+      "step": 13988
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010339881344618445,
+      "loss": 0.6555,
+      "step": 13989
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010338781826680683,
+      "loss": 0.5633,
+      "step": 13990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010337682304642529,
+      "loss": 0.5023,
+      "step": 13991
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010336582778517297,
+      "loss": 0.5863,
+      "step": 13992
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010335483248318287,
+      "loss": 0.6279,
+      "step": 13993
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010334383714058812,
+      "loss": 0.5929,
+      "step": 13994
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010333284175752182,
+      "loss": 0.5603,
+      "step": 13995
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010332184633411703,
+      "loss": 0.5776,
+      "step": 13996
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010331085087050682,
+      "loss": 0.6009,
+      "step": 13997
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010329985536682425,
+      "loss": 0.6008,
+      "step": 13998
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010328885982320247,
+      "loss": 0.6488,
+      "step": 13999
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010327786423977448,
+      "loss": 0.5528,
+      "step": 14000
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 0.5594451427459717,
+      "eval_runtime": 851.4366,
+      "eval_samples_per_second": 28.258,
+      "eval_steps_per_second": 14.129,
+      "step": 14000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010326686861667344,
+      "loss": 0.5301,
+      "step": 14001
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010325587295403241,
+      "loss": 0.5787,
+      "step": 14002
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010324487725198442,
+      "loss": 0.5683,
+      "step": 14003
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010323388151066262,
+      "loss": 0.6516,
+      "step": 14004
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010322288573020009,
+      "loss": 0.5318,
+      "step": 14005
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010321188991072985,
+      "loss": 0.5444,
+      "step": 14006
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010320089405238509,
+      "loss": 0.5669,
+      "step": 14007
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010318989815529881,
+      "loss": 0.6639,
+      "step": 14008
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010317890221960413,
+      "loss": 0.6811,
+      "step": 14009
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010316790624543417,
+      "loss": 0.6259,
+      "step": 14010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010315691023292193,
+      "loss": 0.6069,
+      "step": 14011
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001031459141822006,
+      "loss": 0.5201,
+      "step": 14012
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010313491809340319,
+      "loss": 0.6188,
+      "step": 14013
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010312392196666285,
+      "loss": 0.6078,
+      "step": 14014
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010311292580211262,
+      "loss": 0.5469,
+      "step": 14015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010310192959988562,
+      "loss": 0.6219,
+      "step": 14016
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010309093336011492,
+      "loss": 0.5787,
+      "step": 14017
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010307993708293364,
+      "loss": 0.4531,
+      "step": 14018
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010306894076847484,
+      "loss": 0.5655,
+      "step": 14019
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010305794441687164,
+      "loss": 0.5538,
+      "step": 14020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001030469480282571,
+      "loss": 0.5992,
+      "step": 14021
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010303595160276435,
+      "loss": 0.5219,
+      "step": 14022
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010302495514052648,
+      "loss": 0.521,
+      "step": 14023
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010301395864167654,
+      "loss": 0.6326,
+      "step": 14024
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010300296210634766,
+      "loss": 0.5413,
+      "step": 14025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010299196553467292,
+      "loss": 0.5671,
+      "step": 14026
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010298096892678544,
+      "loss": 0.6397,
+      "step": 14027
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001029699722828183,
+      "loss": 0.6105,
+      "step": 14028
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010295897560290457,
+      "loss": 0.6282,
+      "step": 14029
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001029479788871774,
+      "loss": 0.6294,
+      "step": 14030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010293698213576984,
+      "loss": 0.6204,
+      "step": 14031
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010292598534881501,
+      "loss": 0.497,
+      "step": 14032
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010291498852644603,
+      "loss": 0.5933,
+      "step": 14033
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010290399166879594,
+      "loss": 0.7162,
+      "step": 14034
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010289299477599788,
+      "loss": 0.5841,
+      "step": 14035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010288199784818493,
+      "loss": 0.5272,
+      "step": 14036
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010287100088549022,
+      "loss": 0.6053,
+      "step": 14037
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010286000388804681,
+      "loss": 0.5138,
+      "step": 14038
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010284900685598781,
+      "loss": 0.5524,
+      "step": 14039
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010283800978944636,
+      "loss": 0.5446,
+      "step": 14040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010282701268855553,
+      "loss": 0.5434,
+      "step": 14041
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001028160155534484,
+      "loss": 0.5513,
+      "step": 14042
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010280501838425811,
+      "loss": 0.546,
+      "step": 14043
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010279402118111774,
+      "loss": 0.5036,
+      "step": 14044
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001027830239441604,
+      "loss": 0.5192,
+      "step": 14045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010277202667351921,
+      "loss": 0.4841,
+      "step": 14046
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010276102936932723,
+      "loss": 0.5238,
+      "step": 14047
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001027500320317176,
+      "loss": 0.55,
+      "step": 14048
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010273903466082341,
+      "loss": 0.7758,
+      "step": 14049
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010272803725677779,
+      "loss": 0.5546,
+      "step": 14050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010271703981971384,
+      "loss": 0.6683,
+      "step": 14051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001027060423497646,
+      "loss": 0.4839,
+      "step": 14052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010269504484706324,
+      "loss": 0.5639,
+      "step": 14053
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010268404731174288,
+      "loss": 0.5838,
+      "step": 14054
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010267304974393659,
+      "loss": 0.6491,
+      "step": 14055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010266205214377748,
+      "loss": 0.5151,
+      "step": 14056
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010265105451139867,
+      "loss": 0.5069,
+      "step": 14057
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010264005684693325,
+      "loss": 0.539,
+      "step": 14058
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010262905915051437,
+      "loss": 0.5467,
+      "step": 14059
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010261806142227507,
+      "loss": 0.5205,
+      "step": 14060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010260706366234856,
+      "loss": 0.4994,
+      "step": 14061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010259606587086783,
+      "loss": 0.5442,
+      "step": 14062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001025850680479661,
+      "loss": 0.5393,
+      "step": 14063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010257407019377638,
+      "loss": 0.5275,
+      "step": 14064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010256307230843188,
+      "loss": 0.605,
+      "step": 14065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010255207439206564,
+      "loss": 0.6328,
+      "step": 14066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010254107644481077,
+      "loss": 0.6317,
+      "step": 14067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010253007846680042,
+      "loss": 0.6206,
+      "step": 14068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010251908045816767,
+      "loss": 0.5645,
+      "step": 14069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010250808241904568,
+      "loss": 0.5918,
+      "step": 14070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010249708434956752,
+      "loss": 0.5243,
+      "step": 14071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001024860862498663,
+      "loss": 0.6042,
+      "step": 14072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010247508812007519,
+      "loss": 0.5647,
+      "step": 14073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010246408996032723,
+      "loss": 0.4788,
+      "step": 14074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010245309177075558,
+      "loss": 0.6077,
+      "step": 14075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010244209355149333,
+      "loss": 0.5989,
+      "step": 14076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010243109530267361,
+      "loss": 0.5262,
+      "step": 14077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010242009702442955,
+      "loss": 0.5495,
+      "step": 14078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010240909871689424,
+      "loss": 0.5792,
+      "step": 14079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010239810038020077,
+      "loss": 0.5851,
+      "step": 14080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010238710201448235,
+      "loss": 0.6329,
+      "step": 14081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010237610361987197,
+      "loss": 0.5334,
+      "step": 14082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010236510519650289,
+      "loss": 0.601,
+      "step": 14083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010235410674450812,
+      "loss": 0.7276,
+      "step": 14084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010234310826402079,
+      "loss": 0.5966,
+      "step": 14085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010233210975517406,
+      "loss": 0.5255,
+      "step": 14086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010232111121810101,
+      "loss": 0.5266,
+      "step": 14087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010231011265293482,
+      "loss": 0.6179,
+      "step": 14088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010229911405980855,
+      "loss": 0.5992,
+      "step": 14089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001022881154388553,
+      "loss": 0.6308,
+      "step": 14090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010227711679020827,
+      "loss": 0.625,
+      "step": 14091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010226611811400053,
+      "loss": 0.5702,
+      "step": 14092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010225511941036518,
+      "loss": 0.5743,
+      "step": 14093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010224412067943539,
+      "loss": 0.581,
+      "step": 14094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010223312192134424,
+      "loss": 0.5361,
+      "step": 14095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010222212313622487,
+      "loss": 0.5426,
+      "step": 14096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010221112432421045,
+      "loss": 0.508,
+      "step": 14097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000102200125485434,
+      "loss": 0.5559,
+      "step": 14098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010218912662002873,
+      "loss": 0.5075,
+      "step": 14099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001021781277281277,
+      "loss": 0.5157,
+      "step": 14100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001021671288098641,
+      "loss": 0.5351,
+      "step": 14101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000102156129865371,
+      "loss": 0.5917,
+      "step": 14102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010214513089478152,
+      "loss": 0.4888,
+      "step": 14103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010213413189822885,
+      "loss": 0.602,
+      "step": 14104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010212313287584605,
+      "loss": 0.5426,
+      "step": 14105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010211213382776623,
+      "loss": 0.622,
+      "step": 14106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010210113475412259,
+      "loss": 0.572,
+      "step": 14107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010209013565504821,
+      "loss": 0.5858,
+      "step": 14108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010207913653067622,
+      "loss": 0.5307,
+      "step": 14109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010206813738113976,
+      "loss": 0.5222,
+      "step": 14110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010205713820657192,
+      "loss": 0.6031,
+      "step": 14111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010204613900710587,
+      "loss": 0.5738,
+      "step": 14112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001020351397828747,
+      "loss": 0.533,
+      "step": 14113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010202414053401155,
+      "loss": 0.5136,
+      "step": 14114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010201314126064958,
+      "loss": 0.513,
+      "step": 14115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010200214196292186,
+      "loss": 0.5219,
+      "step": 14116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010199114264096156,
+      "loss": 0.5372,
+      "step": 14117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010198014329490179,
+      "loss": 0.5222,
+      "step": 14118
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001019691439248757,
+      "loss": 0.5585,
+      "step": 14119
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010195814453101642,
+      "loss": 0.5596,
+      "step": 14120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010194714511345702,
+      "loss": 0.487,
+      "step": 14121
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010193614567233069,
+      "loss": 0.5434,
+      "step": 14122
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010192514620777057,
+      "loss": 0.6122,
+      "step": 14123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010191414671990973,
+      "loss": 0.6672,
+      "step": 14124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010190314720888135,
+      "loss": 0.6092,
+      "step": 14125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010189214767481853,
+      "loss": 0.5611,
+      "step": 14126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010188114811785443,
+      "loss": 0.5128,
+      "step": 14127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010187014853812218,
+      "loss": 0.6652,
+      "step": 14128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010185914893575486,
+      "loss": 0.5119,
+      "step": 14129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010184814931088571,
+      "loss": 0.5163,
+      "step": 14130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010183714966364772,
+      "loss": 0.5452,
+      "step": 14131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010182614999417413,
+      "loss": 0.5949,
+      "step": 14132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000101815150302598,
+      "loss": 0.6483,
+      "step": 14133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010180415058905254,
+      "loss": 0.6161,
+      "step": 14134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010179315085367084,
+      "loss": 0.5141,
+      "step": 14135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010178215109658602,
+      "loss": 0.5716,
+      "step": 14136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010177115131793127,
+      "loss": 0.5916,
+      "step": 14137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010176015151783964,
+      "loss": 0.5361,
+      "step": 14138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010174915169644432,
+      "loss": 0.5045,
+      "step": 14139
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010173815185387845,
+      "loss": 0.5522,
+      "step": 14140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010172715199027514,
+      "loss": 0.4967,
+      "step": 14141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010171615210576753,
+      "loss": 0.4996,
+      "step": 14142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010170515220048877,
+      "loss": 0.6288,
+      "step": 14143
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010169415227457196,
+      "loss": 0.5646,
+      "step": 14144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010168315232815029,
+      "loss": 0.5754,
+      "step": 14145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010167215236135683,
+      "loss": 0.6168,
+      "step": 14146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010166115237432479,
+      "loss": 0.6127,
+      "step": 14147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010165015236718724,
+      "loss": 0.6069,
+      "step": 14148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010163915234007735,
+      "loss": 0.5801,
+      "step": 14149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010162815229312827,
+      "loss": 0.6214,
+      "step": 14150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001016171522264731,
+      "loss": 0.5454,
+      "step": 14151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000101606152140245,
+      "loss": 0.5692,
+      "step": 14152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010159515203457713,
+      "loss": 0.6002,
+      "step": 14153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010158415190960258,
+      "loss": 0.5612,
+      "step": 14154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010157315176545452,
+      "loss": 0.5928,
+      "step": 14155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010156215160226607,
+      "loss": 0.6773,
+      "step": 14156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010155115142017038,
+      "loss": 0.5735,
+      "step": 14157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010154015121930061,
+      "loss": 0.6613,
+      "step": 14158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010152915099978983,
+      "loss": 0.5714,
+      "step": 14159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010151815076177127,
+      "loss": 0.6154,
+      "step": 14160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010150715050537802,
+      "loss": 0.5539,
+      "step": 14161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010149615023074319,
+      "loss": 0.5355,
+      "step": 14162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000101485149938,
+      "loss": 0.5469,
+      "step": 14163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001014741496272815,
+      "loss": 0.5829,
+      "step": 14164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010146314929872092,
+      "loss": 0.625,
+      "step": 14165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010145214895245134,
+      "loss": 0.5815,
+      "step": 14166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010144114858860589,
+      "loss": 0.5817,
+      "step": 14167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010143014820731778,
+      "loss": 0.5421,
+      "step": 14168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010141914780872006,
+      "loss": 0.575,
+      "step": 14169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010140814739294597,
+      "loss": 0.5213,
+      "step": 14170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010139714696012859,
+      "loss": 0.7336,
+      "step": 14171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010138614651040106,
+      "loss": 0.5802,
+      "step": 14172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010137514604389655,
+      "loss": 0.5822,
+      "step": 14173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010136414556074817,
+      "loss": 0.6084,
+      "step": 14174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010135314506108911,
+      "loss": 0.5592,
+      "step": 14175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010134214454505246,
+      "loss": 0.665,
+      "step": 14176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010133114401277139,
+      "loss": 0.5569,
+      "step": 14177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010132014346437906,
+      "loss": 0.5374,
+      "step": 14178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010130914290000859,
+      "loss": 0.6278,
+      "step": 14179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001012981423197931,
+      "loss": 0.6439,
+      "step": 14180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010128714172386578,
+      "loss": 0.6383,
+      "step": 14181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010127614111235975,
+      "loss": 0.5016,
+      "step": 14182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010126514048540816,
+      "loss": 0.4564,
+      "step": 14183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010125413984314416,
+      "loss": 0.5567,
+      "step": 14184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010124313918570085,
+      "loss": 0.4611,
+      "step": 14185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010123213851321145,
+      "loss": 0.6169,
+      "step": 14186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010122113782580906,
+      "loss": 0.6379,
+      "step": 14187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010121013712362684,
+      "loss": 0.5623,
+      "step": 14188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010119913640679792,
+      "loss": 0.5095,
+      "step": 14189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010118813567545544,
+      "loss": 0.6157,
+      "step": 14190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010117713492973257,
+      "loss": 0.5173,
+      "step": 14191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010116613416976246,
+      "loss": 0.6966,
+      "step": 14192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010115513339567819,
+      "loss": 0.5865,
+      "step": 14193
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010114413260761299,
+      "loss": 0.549,
+      "step": 14194
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010113313180569996,
+      "loss": 0.5438,
+      "step": 14195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010112213099007229,
+      "loss": 0.5956,
+      "step": 14196
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010111113016086306,
+      "loss": 0.4695,
+      "step": 14197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010110012931820547,
+      "loss": 0.6512,
+      "step": 14198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010108912846223264,
+      "loss": 0.496,
+      "step": 14199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010107812759307772,
+      "loss": 0.5617,
+      "step": 14200
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 0.5544564127922058,
+      "eval_runtime": 854.2321,
+      "eval_samples_per_second": 28.166,
+      "eval_steps_per_second": 14.083,
+      "step": 14200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010106712671087386,
+      "loss": 0.6357,
+      "step": 14201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010105612581575421,
+      "loss": 0.4744,
+      "step": 14202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010104512490785194,
+      "loss": 0.601,
+      "step": 14203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010103412398730017,
+      "loss": 0.5159,
+      "step": 14204
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000101023123054232,
+      "loss": 0.5774,
+      "step": 14205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010101212210878069,
+      "loss": 0.6217,
+      "step": 14206
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010100112115107931,
+      "loss": 0.6228,
+      "step": 14207
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010099012018126101,
+      "loss": 0.5141,
+      "step": 14208
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010097911919945899,
+      "loss": 0.6021,
+      "step": 14209
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010096811820580634,
+      "loss": 0.6225,
+      "step": 14210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010095711720043623,
+      "loss": 0.6484,
+      "step": 14211
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010094611618348184,
+      "loss": 0.5711,
+      "step": 14212
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010093511515507625,
+      "loss": 0.709,
+      "step": 14213
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010092411411535268,
+      "loss": 0.6243,
+      "step": 14214
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010091311306444421,
+      "loss": 0.542,
+      "step": 14215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010090211200248407,
+      "loss": 0.5785,
+      "step": 14216
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010089111092960537,
+      "loss": 0.705,
+      "step": 14217
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010088010984594123,
+      "loss": 0.4549,
+      "step": 14218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010086910875162482,
+      "loss": 0.539,
+      "step": 14219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010085810764678931,
+      "loss": 0.6278,
+      "step": 14220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010084710653156786,
+      "loss": 0.6367,
+      "step": 14221
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010083610540609358,
+      "loss": 0.4743,
+      "step": 14222
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010082510427049962,
+      "loss": 0.4498,
+      "step": 14223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010081410312491917,
+      "loss": 0.5918,
+      "step": 14224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010080310196948536,
+      "loss": 0.5011,
+      "step": 14225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010079210080433132,
+      "loss": 0.6594,
+      "step": 14226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010078109962959023,
+      "loss": 0.5844,
+      "step": 14227
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010077009844539523,
+      "loss": 0.5781,
+      "step": 14228
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010075909725187948,
+      "loss": 0.5699,
+      "step": 14229
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010074809604917612,
+      "loss": 0.6055,
+      "step": 14230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001007370948374183,
+      "loss": 0.5799,
+      "step": 14231
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010072609361673917,
+      "loss": 0.5946,
+      "step": 14232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010071509238727188,
+      "loss": 0.6965,
+      "step": 14233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010070409114914963,
+      "loss": 0.485,
+      "step": 14234
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001006930899025055,
+      "loss": 0.6306,
+      "step": 14235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010068208864747266,
+      "loss": 0.5112,
+      "step": 14236
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010067108738418429,
+      "loss": 0.5119,
+      "step": 14237
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010066008611277352,
+      "loss": 0.6396,
+      "step": 14238
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010064908483337353,
+      "loss": 0.5048,
+      "step": 14239
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001006380835461174,
+      "loss": 0.5914,
+      "step": 14240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010062708225113837,
+      "loss": 0.5294,
+      "step": 14241
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010061608094856956,
+      "loss": 0.6144,
+      "step": 14242
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010060507963854411,
+      "loss": 0.5644,
+      "step": 14243
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010059407832119514,
+      "loss": 0.4977,
+      "step": 14244
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001005830769966559,
+      "loss": 0.5853,
+      "step": 14245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010057207566505944,
+      "loss": 0.5255,
+      "step": 14246
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010056107432653898,
+      "loss": 0.5022,
+      "step": 14247
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010055007298122765,
+      "loss": 0.6448,
+      "step": 14248
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001005390716292586,
+      "loss": 0.6288,
+      "step": 14249
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010052807027076498,
+      "loss": 0.5314,
+      "step": 14250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010051706890587996,
+      "loss": 0.5593,
+      "step": 14251
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010050606753473668,
+      "loss": 0.5159,
+      "step": 14252
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001004950661574683,
+      "loss": 0.5704,
+      "step": 14253
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010048406477420795,
+      "loss": 0.5844,
+      "step": 14254
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010047306338508882,
+      "loss": 0.5347,
+      "step": 14255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010046206199024406,
+      "loss": 0.5258,
+      "step": 14256
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010045106058980678,
+      "loss": 0.672,
+      "step": 14257
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010044005918391017,
+      "loss": 0.5491,
+      "step": 14258
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010042905777268739,
+      "loss": 0.5802,
+      "step": 14259
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010041805635627158,
+      "loss": 0.4981,
+      "step": 14260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010040705493479588,
+      "loss": 0.61,
+      "step": 14261
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010039605350839345,
+      "loss": 0.6081,
+      "step": 14262
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010038505207719749,
+      "loss": 0.5482,
+      "step": 14263
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010037405064134107,
+      "loss": 0.634,
+      "step": 14264
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010036304920095745,
+      "loss": 0.5691,
+      "step": 14265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010035204775617968,
+      "loss": 0.631,
+      "step": 14266
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010034104630714096,
+      "loss": 0.5351,
+      "step": 14267
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010033004485397445,
+      "loss": 0.5591,
+      "step": 14268
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010031904339681331,
+      "loss": 0.565,
+      "step": 14269
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010030804193579066,
+      "loss": 0.5272,
+      "step": 14270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010029704047103967,
+      "loss": 0.5123,
+      "step": 14271
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010028603900269353,
+      "loss": 0.5991,
+      "step": 14272
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010027503753088537,
+      "loss": 0.4906,
+      "step": 14273
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001002640360557483,
+      "loss": 0.4707,
+      "step": 14274
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010025303457741553,
+      "loss": 0.6119,
+      "step": 14275
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010024203309602021,
+      "loss": 0.5843,
+      "step": 14276
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010023103161169548,
+      "loss": 0.4743,
+      "step": 14277
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010022003012457448,
+      "loss": 0.5645,
+      "step": 14278
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010020902863479038,
+      "loss": 0.6084,
+      "step": 14279
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010019802714247637,
+      "loss": 0.5003,
+      "step": 14280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010018702564776554,
+      "loss": 0.3628,
+      "step": 14281
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010017602415079108,
+      "loss": 0.584,
+      "step": 14282
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010016502265168616,
+      "loss": 0.601,
+      "step": 14283
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010015402115058389,
+      "loss": 0.5531,
+      "step": 14284
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010014301964761747,
+      "loss": 0.5808,
+      "step": 14285
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010013201814292004,
+      "loss": 0.6227,
+      "step": 14286
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010012101663662472,
+      "loss": 0.5222,
+      "step": 14287
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010011001512886473,
+      "loss": 0.5901,
+      "step": 14288
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010009901361977316,
+      "loss": 0.5717,
+      "step": 14289
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010008801210948322,
+      "loss": 0.6155,
+      "step": 14290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010007701059812804,
+      "loss": 0.5641,
+      "step": 14291
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010006600908584076,
+      "loss": 0.5959,
+      "step": 14292
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010005500757275456,
+      "loss": 0.5371,
+      "step": 14293
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010004400605900259,
+      "loss": 0.5902,
+      "step": 14294
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010003300454471798,
+      "loss": 0.5553,
+      "step": 14295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010002200303003393,
+      "loss": 0.6082,
+      "step": 14296
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010001100151508354,
+      "loss": 0.5748,
+      "step": 14297
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001,
+      "loss": 0.4938,
+      "step": 14298
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.998899848491648e-05,
+      "loss": 0.5222,
+      "step": 14299
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.99779969699661e-05,
+      "loss": 0.5688,
+      "step": 14300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.996699545528205e-05,
+      "loss": 0.5815,
+      "step": 14301
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.995599394099743e-05,
+      "loss": 0.531,
+      "step": 14302
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.994499242724546e-05,
+      "loss": 0.5232,
+      "step": 14303
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.993399091415927e-05,
+      "loss": 0.5625,
+      "step": 14304
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.992298940187198e-05,
+      "loss": 0.5583,
+      "step": 14305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.99119878905168e-05,
+      "loss": 0.4679,
+      "step": 14306
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.990098638022686e-05,
+      "loss": 0.5065,
+      "step": 14307
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.98899848711353e-05,
+      "loss": 0.6253,
+      "step": 14308
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.98789833633753e-05,
+      "loss": 0.5626,
+      "step": 14309
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.986798185707999e-05,
+      "loss": 0.6539,
+      "step": 14310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.985698035238254e-05,
+      "loss": 0.5478,
+      "step": 14311
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.984597884941613e-05,
+      "loss": 0.608,
+      "step": 14312
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.983497734831386e-05,
+      "loss": 0.5794,
+      "step": 14313
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.982397584920894e-05,
+      "loss": 0.5171,
+      "step": 14314
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.981297435223447e-05,
+      "loss": 0.5392,
+      "step": 14315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.980197285752367e-05,
+      "loss": 0.5875,
+      "step": 14316
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.979097136520964e-05,
+      "loss": 0.5993,
+      "step": 14317
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.977996987542554e-05,
+      "loss": 0.6113,
+      "step": 14318
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.976896838830456e-05,
+      "loss": 0.5656,
+      "step": 14319
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.97579669039798e-05,
+      "loss": 0.6858,
+      "step": 14320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.974696542258448e-05,
+      "loss": 0.5812,
+      "step": 14321
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.973596394425172e-05,
+      "loss": 0.5483,
+      "step": 14322
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.972496246911466e-05,
+      "loss": 0.623,
+      "step": 14323
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.971396099730648e-05,
+      "loss": 0.6466,
+      "step": 14324
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.970295952896034e-05,
+      "loss": 0.5975,
+      "step": 14325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.969195806420935e-05,
+      "loss": 0.5,
+      "step": 14326
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.968095660318673e-05,
+      "loss": 0.5711,
+      "step": 14327
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.966995514602556e-05,
+      "loss": 0.5549,
+      "step": 14328
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.965895369285906e-05,
+      "loss": 0.5666,
+      "step": 14329
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.964795224382035e-05,
+      "loss": 0.5748,
+      "step": 14330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.963695079904257e-05,
+      "loss": 0.5243,
+      "step": 14331
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.962594935865893e-05,
+      "loss": 0.4759,
+      "step": 14332
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.961494792280252e-05,
+      "loss": 0.6543,
+      "step": 14333
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.960394649160654e-05,
+      "loss": 0.644,
+      "step": 14334
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.959294506520413e-05,
+      "loss": 0.5205,
+      "step": 14335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.958194364372843e-05,
+      "loss": 0.5503,
+      "step": 14336
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.957094222731262e-05,
+      "loss": 0.5097,
+      "step": 14337
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.955994081608983e-05,
+      "loss": 0.5498,
+      "step": 14338
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.954893941019323e-05,
+      "loss": 0.5805,
+      "step": 14339
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.953793800975597e-05,
+      "loss": 0.5912,
+      "step": 14340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.952693661491117e-05,
+      "loss": 0.5807,
+      "step": 14341
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.951593522579204e-05,
+      "loss": 0.5811,
+      "step": 14342
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.950493384253172e-05,
+      "loss": 0.5945,
+      "step": 14343
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.949393246526332e-05,
+      "loss": 0.5277,
+      "step": 14344
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.948293109412005e-05,
+      "loss": 0.6124,
+      "step": 14345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.947192972923502e-05,
+      "loss": 0.5046,
+      "step": 14346
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.94609283707414e-05,
+      "loss": 0.5318,
+      "step": 14347
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.944992701877236e-05,
+      "loss": 0.5231,
+      "step": 14348
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.943892567346101e-05,
+      "loss": 0.5867,
+      "step": 14349
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.942792433494056e-05,
+      "loss": 0.4277,
+      "step": 14350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.941692300334411e-05,
+      "loss": 0.6649,
+      "step": 14351
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.940592167880485e-05,
+      "loss": 0.5055,
+      "step": 14352
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.939492036145593e-05,
+      "loss": 0.5273,
+      "step": 14353
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.938391905143045e-05,
+      "loss": 0.566,
+      "step": 14354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.937291774886164e-05,
+      "loss": 0.511,
+      "step": 14355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.93619164538826e-05,
+      "loss": 0.5749,
+      "step": 14356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.935091516662648e-05,
+      "loss": 0.6261,
+      "step": 14357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.933991388722649e-05,
+      "loss": 0.557,
+      "step": 14358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.93289126158157e-05,
+      "loss": 0.5258,
+      "step": 14359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.931791135252734e-05,
+      "loss": 0.5029,
+      "step": 14360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.930691009749452e-05,
+      "loss": 0.5335,
+      "step": 14361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.929590885085038e-05,
+      "loss": 0.6382,
+      "step": 14362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.92849076127281e-05,
+      "loss": 0.5401,
+      "step": 14363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.927390638326081e-05,
+      "loss": 0.5509,
+      "step": 14364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.92629051625817e-05,
+      "loss": 0.5166,
+      "step": 14365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.92519039508239e-05,
+      "loss": 0.6006,
+      "step": 14366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.924090274812055e-05,
+      "loss": 0.5864,
+      "step": 14367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.922990155460478e-05,
+      "loss": 0.6476,
+      "step": 14368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.921890037040979e-05,
+      "loss": 0.5769,
+      "step": 14369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.920789919566872e-05,
+      "loss": 0.5795,
+      "step": 14370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.919689803051467e-05,
+      "loss": 0.6302,
+      "step": 14371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.918589687508087e-05,
+      "loss": 0.5449,
+      "step": 14372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.917489572950039e-05,
+      "loss": 0.4621,
+      "step": 14373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.916389459390645e-05,
+      "loss": 0.5325,
+      "step": 14374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.915289346843219e-05,
+      "loss": 0.5214,
+      "step": 14375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.914189235321071e-05,
+      "loss": 0.5154,
+      "step": 14376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.91308912483752e-05,
+      "loss": 0.5018,
+      "step": 14377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.911989015405882e-05,
+      "loss": 0.4656,
+      "step": 14378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.910888907039468e-05,
+      "loss": 0.6141,
+      "step": 14379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.909788799751598e-05,
+      "loss": 0.5424,
+      "step": 14380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.90868869355558e-05,
+      "loss": 0.5587,
+      "step": 14381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.907588588464737e-05,
+      "loss": 0.4981,
+      "step": 14382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.906488484492378e-05,
+      "loss": 0.5407,
+      "step": 14383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.905388381651821e-05,
+      "loss": 0.6075,
+      "step": 14384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.90428827995638e-05,
+      "loss": 0.605,
+      "step": 14385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.903188179419369e-05,
+      "loss": 0.4483,
+      "step": 14386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.902088080054105e-05,
+      "loss": 0.5744,
+      "step": 14387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.900987981873902e-05,
+      "loss": 0.6089,
+      "step": 14388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.899887884892072e-05,
+      "loss": 0.532,
+      "step": 14389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.898787789121935e-05,
+      "loss": 0.5709,
+      "step": 14390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.8976876945768e-05,
+      "loss": 0.6205,
+      "step": 14391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.896587601269989e-05,
+      "loss": 0.7432,
+      "step": 14392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.895487509214812e-05,
+      "loss": 0.5582,
+      "step": 14393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.894387418424582e-05,
+      "loss": 0.5614,
+      "step": 14394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.893287328912617e-05,
+      "loss": 0.6256,
+      "step": 14395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.892187240692232e-05,
+      "loss": 0.5941,
+      "step": 14396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.89108715377674e-05,
+      "loss": 0.5118,
+      "step": 14397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.889987068179457e-05,
+      "loss": 0.6354,
+      "step": 14398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.888886983913695e-05,
+      "loss": 0.5775,
+      "step": 14399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.887786900992775e-05,
+      "loss": 0.6572,
+      "step": 14400
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 0.5487639307975769,
+      "eval_runtime": 843.9853,
+      "eval_samples_per_second": 28.508,
+      "eval_steps_per_second": 14.254,
+      "step": 14400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.886686819430006e-05,
+      "loss": 0.5631,
+      "step": 14401
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.885586739238703e-05,
+      "loss": 0.6072,
+      "step": 14402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.884486660432182e-05,
+      "loss": 0.5003,
+      "step": 14403
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.883386583023756e-05,
+      "loss": 0.5545,
+      "step": 14404
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.882286507026745e-05,
+      "loss": 0.5915,
+      "step": 14405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.88118643245446e-05,
+      "loss": 0.5749,
+      "step": 14406
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.880086359320209e-05,
+      "loss": 0.6309,
+      "step": 14407
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.878986287637318e-05,
+      "loss": 0.8132,
+      "step": 14408
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.877886217419096e-05,
+      "loss": 0.6478,
+      "step": 14409
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.876786148678856e-05,
+      "loss": 0.5312,
+      "step": 14410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.875686081429916e-05,
+      "loss": 0.5428,
+      "step": 14411
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.874586015685586e-05,
+      "loss": 0.5827,
+      "step": 14412
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.873485951459187e-05,
+      "loss": 0.6514,
+      "step": 14413
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.872385888764029e-05,
+      "loss": 0.5465,
+      "step": 14414
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.871285827613425e-05,
+      "loss": 0.4965,
+      "step": 14415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.870185768020693e-05,
+      "loss": 0.5281,
+      "step": 14416
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.869085709999144e-05,
+      "loss": 0.5835,
+      "step": 14417
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.867985653562098e-05,
+      "loss": 0.5404,
+      "step": 14418
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.866885598722863e-05,
+      "loss": 0.6164,
+      "step": 14419
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.865785545494756e-05,
+      "loss": 0.5682,
+      "step": 14420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.864685493891093e-05,
+      "loss": 0.5776,
+      "step": 14421
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.863585443925183e-05,
+      "loss": 0.5949,
+      "step": 14422
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.862485395610347e-05,
+      "loss": 0.5085,
+      "step": 14423
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.861385348959897e-05,
+      "loss": 0.6075,
+      "step": 14424
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.860285303987144e-05,
+      "loss": 0.4908,
+      "step": 14425
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.859185260705404e-05,
+      "loss": 0.6506,
+      "step": 14426
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.858085219127995e-05,
+      "loss": 0.4995,
+      "step": 14427
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.856985179268224e-05,
+      "loss": 0.6625,
+      "step": 14428
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.855885141139413e-05,
+      "loss": 0.7336,
+      "step": 14429
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.854785104754868e-05,
+      "loss": 0.5746,
+      "step": 14430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.853685070127911e-05,
+      "loss": 0.5245,
+      "step": 14431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.852585037271853e-05,
+      "loss": 0.5589,
+      "step": 14432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.851485006200002e-05,
+      "loss": 0.5388,
+      "step": 14433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.850384976925682e-05,
+      "loss": 0.4889,
+      "step": 14434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.8492849494622e-05,
+      "loss": 0.6384,
+      "step": 14435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.848184923822875e-05,
+      "loss": 0.594,
+      "step": 14436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.847084900021018e-05,
+      "loss": 0.5422,
+      "step": 14437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.845984878069941e-05,
+      "loss": 0.5974,
+      "step": 14438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.844884857982963e-05,
+      "loss": 0.511,
+      "step": 14439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.843784839773394e-05,
+      "loss": 0.6518,
+      "step": 14440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.84268482345455e-05,
+      "loss": 0.6108,
+      "step": 14441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.841584809039744e-05,
+      "loss": 0.4898,
+      "step": 14442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.840484796542288e-05,
+      "loss": 0.5997,
+      "step": 14443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.839384785975501e-05,
+      "loss": 0.4524,
+      "step": 14444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.838284777352692e-05,
+      "loss": 0.6314,
+      "step": 14445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.837184770687175e-05,
+      "loss": 0.6412,
+      "step": 14446
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.836084765992267e-05,
+      "loss": 0.4925,
+      "step": 14447
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.834984763281278e-05,
+      "loss": 0.5203,
+      "step": 14448
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.833884762567525e-05,
+      "loss": 0.5965,
+      "step": 14449
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.83278476386432e-05,
+      "loss": 0.5463,
+      "step": 14450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.831684767184974e-05,
+      "loss": 0.6653,
+      "step": 14451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.830584772542806e-05,
+      "loss": 0.5431,
+      "step": 14452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.829484779951126e-05,
+      "loss": 0.6168,
+      "step": 14453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.82838478942325e-05,
+      "loss": 0.5565,
+      "step": 14454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.82728480097249e-05,
+      "loss": 0.5211,
+      "step": 14455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.826184814612157e-05,
+      "loss": 0.6141,
+      "step": 14456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.82508483035557e-05,
+      "loss": 0.5678,
+      "step": 14457
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.823984848216039e-05,
+      "loss": 0.5636,
+      "step": 14458
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.822884868206876e-05,
+      "loss": 0.5304,
+      "step": 14459
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.821784890341399e-05,
+      "loss": 0.5914,
+      "step": 14460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.820684914632917e-05,
+      "loss": 0.5429,
+      "step": 14461
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.819584941094747e-05,
+      "loss": 0.5797,
+      "step": 14462
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.8184849697402e-05,
+      "loss": 0.498,
+      "step": 14463
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.81738500058259e-05,
+      "loss": 0.5531,
+      "step": 14464
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.81628503363523e-05,
+      "loss": 0.6235,
+      "step": 14465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.815185068911433e-05,
+      "loss": 0.5896,
+      "step": 14466
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.814085106424514e-05,
+      "loss": 0.6426,
+      "step": 14467
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.812985146187785e-05,
+      "loss": 0.5522,
+      "step": 14468
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.811885188214557e-05,
+      "loss": 0.5509,
+      "step": 14469
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.810785232518148e-05,
+      "loss": 0.5669,
+      "step": 14470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.809685279111864e-05,
+      "loss": 0.5237,
+      "step": 14471
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.808585328009027e-05,
+      "loss": 0.5829,
+      "step": 14472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.807485379222944e-05,
+      "loss": 0.5166,
+      "step": 14473
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.80638543276693e-05,
+      "loss": 0.4819,
+      "step": 14474
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.805285488654297e-05,
+      "loss": 0.5977,
+      "step": 14475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.804185546898361e-05,
+      "loss": 0.5022,
+      "step": 14476
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.80308560751243e-05,
+      "loss": 0.5635,
+      "step": 14477
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.801985670509822e-05,
+      "loss": 0.495,
+      "step": 14478
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.800885735903843e-05,
+      "loss": 0.4514,
+      "step": 14479
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.799785803707815e-05,
+      "loss": 0.5305,
+      "step": 14480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.798685873935046e-05,
+      "loss": 0.5784,
+      "step": 14481
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.797585946598845e-05,
+      "loss": 0.5702,
+      "step": 14482
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.796486021712531e-05,
+      "loss": 0.5593,
+      "step": 14483
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.795386099289414e-05,
+      "loss": 0.507,
+      "step": 14484
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.794286179342809e-05,
+      "loss": 0.4777,
+      "step": 14485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.793186261886028e-05,
+      "loss": 0.5935,
+      "step": 14486
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.792086346932378e-05,
+      "loss": 0.6288,
+      "step": 14487
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.79098643449518e-05,
+      "loss": 0.6835,
+      "step": 14488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.789886524587739e-05,
+      "loss": 0.5749,
+      "step": 14489
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.788786617223376e-05,
+      "loss": 0.5948,
+      "step": 14490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.787686712415398e-05,
+      "loss": 0.5648,
+      "step": 14491
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.786586810177116e-05,
+      "loss": 0.6627,
+      "step": 14492
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.785486910521847e-05,
+      "loss": 0.4932,
+      "step": 14493
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.784387013462903e-05,
+      "loss": 0.5362,
+      "step": 14494
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.783287119013591e-05,
+      "loss": 0.5175,
+      "step": 14495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.782187227187231e-05,
+      "loss": 0.5027,
+      "step": 14496
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.781087337997128e-05,
+      "loss": 0.5327,
+      "step": 14497
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.7799874514566e-05,
+      "loss": 0.5973,
+      "step": 14498
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.778887567578958e-05,
+      "loss": 0.4774,
+      "step": 14499
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.777787686377511e-05,
+      "loss": 0.4575,
+      "step": 14500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.776687807865579e-05,
+      "loss": 0.5011,
+      "step": 14501
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.775587932056466e-05,
+      "loss": 0.5857,
+      "step": 14502
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.774488058963487e-05,
+      "loss": 0.4452,
+      "step": 14503
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.773388188599952e-05,
+      "loss": 0.6235,
+      "step": 14504
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.772288320979178e-05,
+      "loss": 0.717,
+      "step": 14505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.771188456114472e-05,
+      "loss": 0.5492,
+      "step": 14506
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.77008859401915e-05,
+      "loss": 0.5158,
+      "step": 14507
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.768988734706523e-05,
+      "loss": 0.4955,
+      "step": 14508
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.7678888781899e-05,
+      "loss": 0.5385,
+      "step": 14509
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.766789024482598e-05,
+      "loss": 0.4823,
+      "step": 14510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.765689173597925e-05,
+      "loss": 0.5101,
+      "step": 14511
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.764589325549193e-05,
+      "loss": 0.5896,
+      "step": 14512
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.763489480349717e-05,
+      "loss": 0.5136,
+      "step": 14513
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.762389638012804e-05,
+      "loss": 0.4455,
+      "step": 14514
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.76128979855177e-05,
+      "loss": 0.5368,
+      "step": 14515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.760189961979926e-05,
+      "loss": 0.561,
+      "step": 14516
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.759090128310581e-05,
+      "loss": 0.5716,
+      "step": 14517
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.75799029755705e-05,
+      "loss": 0.6004,
+      "step": 14518
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.756890469732642e-05,
+      "loss": 0.5791,
+      "step": 14519
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.755790644850671e-05,
+      "loss": 0.6333,
+      "step": 14520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.754690822924447e-05,
+      "loss": 0.4402,
+      "step": 14521
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.75359100396728e-05,
+      "loss": 0.5362,
+      "step": 14522
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.752491187992487e-05,
+      "loss": 0.4906,
+      "step": 14523
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.751391375013372e-05,
+      "loss": 0.6409,
+      "step": 14524
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.750291565043251e-05,
+      "loss": 0.5256,
+      "step": 14525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.749191758095437e-05,
+      "loss": 0.4863,
+      "step": 14526
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.748091954183234e-05,
+      "loss": 0.5328,
+      "step": 14527
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.746992153319961e-05,
+      "loss": 0.6298,
+      "step": 14528
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.745892355518928e-05,
+      "loss": 0.682,
+      "step": 14529
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.74479256079344e-05,
+      "loss": 0.5295,
+      "step": 14530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.743692769156817e-05,
+      "loss": 0.543,
+      "step": 14531
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.742592980622363e-05,
+      "loss": 0.511,
+      "step": 14532
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.741493195203394e-05,
+      "loss": 0.6376,
+      "step": 14533
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.740393412913219e-05,
+      "loss": 0.6029,
+      "step": 14534
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.739293633765148e-05,
+      "loss": 0.5155,
+      "step": 14535
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.738193857772493e-05,
+      "loss": 0.4827,
+      "step": 14536
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.737094084948565e-05,
+      "loss": 0.4659,
+      "step": 14537
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.735994315306676e-05,
+      "loss": 0.6028,
+      "step": 14538
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.734894548860136e-05,
+      "loss": 0.5714,
+      "step": 14539
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.733794785622253e-05,
+      "loss": 0.6404,
+      "step": 14540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.732695025606345e-05,
+      "loss": 0.5595,
+      "step": 14541
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.731595268825713e-05,
+      "loss": 0.5281,
+      "step": 14542
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.730495515293677e-05,
+      "loss": 0.5029,
+      "step": 14543
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.729395765023543e-05,
+      "loss": 0.5453,
+      "step": 14544
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.72829601802862e-05,
+      "loss": 0.5129,
+      "step": 14545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.727196274322223e-05,
+      "loss": 0.5246,
+      "step": 14546
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.726096533917661e-05,
+      "loss": 0.5937,
+      "step": 14547
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.724996796828242e-05,
+      "loss": 0.5304,
+      "step": 14548
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.72389706306728e-05,
+      "loss": 0.5037,
+      "step": 14549
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.722797332648081e-05,
+      "loss": 0.4956,
+      "step": 14550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.721697605583961e-05,
+      "loss": 0.4888,
+      "step": 14551
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.72059788188823e-05,
+      "loss": 0.5739,
+      "step": 14552
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.719498161574192e-05,
+      "loss": 0.4963,
+      "step": 14553
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.718398444655163e-05,
+      "loss": 0.6034,
+      "step": 14554
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.717298731144448e-05,
+      "loss": 0.5399,
+      "step": 14555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.716199021055365e-05,
+      "loss": 0.5111,
+      "step": 14556
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.715099314401221e-05,
+      "loss": 0.5049,
+      "step": 14557
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.71399961119532e-05,
+      "loss": 0.6021,
+      "step": 14558
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.712899911450982e-05,
+      "loss": 0.5565,
+      "step": 14559
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.71180021518151e-05,
+      "loss": 0.5245,
+      "step": 14560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.710700522400214e-05,
+      "loss": 0.4773,
+      "step": 14561
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.709600833120409e-05,
+      "loss": 0.552,
+      "step": 14562
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.7085011473554e-05,
+      "loss": 0.5143,
+      "step": 14563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.707401465118501e-05,
+      "loss": 0.5672,
+      "step": 14564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.706301786423018e-05,
+      "loss": 0.4927,
+      "step": 14565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.705202111282262e-05,
+      "loss": 0.5668,
+      "step": 14566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.704102439709545e-05,
+      "loss": 0.5029,
+      "step": 14567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.703002771718172e-05,
+      "loss": 0.6386,
+      "step": 14568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.701903107321459e-05,
+      "loss": 0.4825,
+      "step": 14569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.700803446532711e-05,
+      "loss": 0.6366,
+      "step": 14570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.699703789365236e-05,
+      "loss": 0.6228,
+      "step": 14571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.698604135832349e-05,
+      "loss": 0.576,
+      "step": 14572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.697504485947356e-05,
+      "loss": 0.5672,
+      "step": 14573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.696404839723566e-05,
+      "loss": 0.5048,
+      "step": 14574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.695305197174291e-05,
+      "loss": 0.5574,
+      "step": 14575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.694205558312837e-05,
+      "loss": 0.5489,
+      "step": 14576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.693105923152517e-05,
+      "loss": 0.5755,
+      "step": 14577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.692006291706639e-05,
+      "loss": 0.6132,
+      "step": 14578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.690906663988509e-05,
+      "loss": 0.4875,
+      "step": 14579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.68980704001144e-05,
+      "loss": 0.5605,
+      "step": 14580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.688707419788739e-05,
+      "loss": 0.6591,
+      "step": 14581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.687607803333717e-05,
+      "loss": 0.5573,
+      "step": 14582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.686508190659682e-05,
+      "loss": 0.6153,
+      "step": 14583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.685408581779942e-05,
+      "loss": 0.4973,
+      "step": 14584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.684308976707808e-05,
+      "loss": 0.4969,
+      "step": 14585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.683209375456586e-05,
+      "loss": 0.5595,
+      "step": 14586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.682109778039588e-05,
+      "loss": 0.4745,
+      "step": 14587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.681010184470121e-05,
+      "loss": 0.521,
+      "step": 14588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.679910594761493e-05,
+      "loss": 0.6186,
+      "step": 14589
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.678811008927016e-05,
+      "loss": 0.5432,
+      "step": 14590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.677711426979993e-05,
+      "loss": 0.6339,
+      "step": 14591
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.67661184893374e-05,
+      "loss": 0.6219,
+      "step": 14592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.67551227480156e-05,
+      "loss": 0.5279,
+      "step": 14593
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.674412704596763e-05,
+      "loss": 0.5921,
+      "step": 14594
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.673313138332658e-05,
+      "loss": 0.5614,
+      "step": 14595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.672213576022554e-05,
+      "loss": 0.5742,
+      "step": 14596
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.671114017679755e-05,
+      "loss": 0.5075,
+      "step": 14597
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.670014463317576e-05,
+      "loss": 0.5636,
+      "step": 14598
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.66891491294932e-05,
+      "loss": 0.5635,
+      "step": 14599
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.667815366588298e-05,
+      "loss": 0.6108,
+      "step": 14600
+    },
+    {
+      "epoch": 0.51,
+      "eval_loss": 0.5440958142280579,
+      "eval_runtime": 853.0607,
+      "eval_samples_per_second": 28.204,
+      "eval_steps_per_second": 14.102,
+      "step": 14600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.666715824247819e-05,
+      "loss": 0.5183,
+      "step": 14601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.665616285941186e-05,
+      "loss": 0.48,
+      "step": 14602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.664516751681714e-05,
+      "loss": 0.5643,
+      "step": 14603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.663417221482704e-05,
+      "loss": 0.6001,
+      "step": 14604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.662317695357472e-05,
+      "loss": 0.6235,
+      "step": 14605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.661218173319318e-05,
+      "loss": 0.5215,
+      "step": 14606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.660118655381555e-05,
+      "loss": 0.5374,
+      "step": 14607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.659019141557489e-05,
+      "loss": 0.5324,
+      "step": 14608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.657919631860429e-05,
+      "loss": 0.4571,
+      "step": 14609
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.65682012630368e-05,
+      "loss": 0.517,
+      "step": 14610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.655720624900552e-05,
+      "loss": 0.56,
+      "step": 14611
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.654621127664351e-05,
+      "loss": 0.536,
+      "step": 14612
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.653521634608388e-05,
+      "loss": 0.4693,
+      "step": 14613
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.652422145745967e-05,
+      "loss": 0.5176,
+      "step": 14614
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.651322661090397e-05,
+      "loss": 0.5736,
+      "step": 14615
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.650223180654984e-05,
+      "loss": 0.525,
+      "step": 14616
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.649123704453038e-05,
+      "loss": 0.7036,
+      "step": 14617
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.648024232497864e-05,
+      "loss": 0.5051,
+      "step": 14618
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.646924764802771e-05,
+      "loss": 0.5772,
+      "step": 14619
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.645825301381063e-05,
+      "loss": 0.6369,
+      "step": 14620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.644725842246052e-05,
+      "loss": 0.5874,
+      "step": 14621
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.64362638741104e-05,
+      "loss": 0.5828,
+      "step": 14622
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.642526936889339e-05,
+      "loss": 0.5772,
+      "step": 14623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.641427490694254e-05,
+      "loss": 0.5373,
+      "step": 14624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.640328048839089e-05,
+      "loss": 0.4969,
+      "step": 14625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.639228611337157e-05,
+      "loss": 0.5002,
+      "step": 14626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.63812917820176e-05,
+      "loss": 0.62,
+      "step": 14627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.637029749446205e-05,
+      "loss": 0.5043,
+      "step": 14628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.6359303250838e-05,
+      "loss": 0.5214,
+      "step": 14629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.634830905127851e-05,
+      "loss": 0.5164,
+      "step": 14630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.633731489591668e-05,
+      "loss": 0.4254,
+      "step": 14631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.632632078488555e-05,
+      "loss": 0.6342,
+      "step": 14632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.631532671831816e-05,
+      "loss": 0.5029,
+      "step": 14633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.630433269634762e-05,
+      "loss": 0.4758,
+      "step": 14634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.6293338719107e-05,
+      "loss": 0.6189,
+      "step": 14635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.62823447867293e-05,
+      "loss": 0.5512,
+      "step": 14636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.627135089934761e-05,
+      "loss": 0.5532,
+      "step": 14637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.626035705709503e-05,
+      "loss": 0.5545,
+      "step": 14638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.624936326010457e-05,
+      "loss": 0.6135,
+      "step": 14639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.623836950850933e-05,
+      "loss": 0.5406,
+      "step": 14640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.622737580244236e-05,
+      "loss": 0.6232,
+      "step": 14641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.621638214203671e-05,
+      "loss": 0.6375,
+      "step": 14642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.620538852742546e-05,
+      "loss": 0.5797,
+      "step": 14643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.619439495874164e-05,
+      "loss": 0.601,
+      "step": 14644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.618340143611835e-05,
+      "loss": 0.6596,
+      "step": 14645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.617240795968861e-05,
+      "loss": 0.4878,
+      "step": 14646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.616141452958549e-05,
+      "loss": 0.5336,
+      "step": 14647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.615042114594206e-05,
+      "loss": 0.505,
+      "step": 14648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.613942780889138e-05,
+      "loss": 0.4071,
+      "step": 14649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.612843451856646e-05,
+      "loss": 0.6403,
+      "step": 14650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.61174412751004e-05,
+      "loss": 0.5489,
+      "step": 14651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.610644807862625e-05,
+      "loss": 0.535,
+      "step": 14652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.609545492927705e-05,
+      "loss": 0.549,
+      "step": 14653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.608446182718588e-05,
+      "loss": 0.5068,
+      "step": 14654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.607346877248574e-05,
+      "loss": 0.5505,
+      "step": 14655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.606247576530975e-05,
+      "loss": 0.4755,
+      "step": 14656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.605148280579089e-05,
+      "loss": 0.6366,
+      "step": 14657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.60404898940623e-05,
+      "loss": 0.5122,
+      "step": 14658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.602949703025696e-05,
+      "loss": 0.535,
+      "step": 14659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.601850421450792e-05,
+      "loss": 0.4861,
+      "step": 14660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.600751144694827e-05,
+      "loss": 0.6706,
+      "step": 14661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.599651872771105e-05,
+      "loss": 0.5315,
+      "step": 14662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.598552605692927e-05,
+      "loss": 0.5509,
+      "step": 14663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.597453343473604e-05,
+      "loss": 0.565,
+      "step": 14664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.596354086126435e-05,
+      "loss": 0.4684,
+      "step": 14665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.595254833664728e-05,
+      "loss": 0.618,
+      "step": 14666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.594155586101787e-05,
+      "loss": 0.579,
+      "step": 14667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.593056343450914e-05,
+      "loss": 0.5162,
+      "step": 14668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.591957105725419e-05,
+      "loss": 0.585,
+      "step": 14669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.5908578729386e-05,
+      "loss": 0.4538,
+      "step": 14670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.589758645103768e-05,
+      "loss": 0.6356,
+      "step": 14671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.588659422234223e-05,
+      "loss": 0.5069,
+      "step": 14672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.587560204343268e-05,
+      "loss": 0.5497,
+      "step": 14673
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.586460991444213e-05,
+      "loss": 0.4827,
+      "step": 14674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.585361783550353e-05,
+      "loss": 0.469,
+      "step": 14675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.584262580675003e-05,
+      "loss": 0.5262,
+      "step": 14676
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.58316338283146e-05,
+      "loss": 0.5295,
+      "step": 14677
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.58206419003303e-05,
+      "loss": 0.6153,
+      "step": 14678
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.580965002293015e-05,
+      "loss": 0.711,
+      "step": 14679
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.579865819624721e-05,
+      "loss": 0.5406,
+      "step": 14680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.57876664204145e-05,
+      "loss": 0.5417,
+      "step": 14681
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.57766746955651e-05,
+      "loss": 0.5591,
+      "step": 14682
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.576568302183198e-05,
+      "loss": 0.5488,
+      "step": 14683
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.575469139934823e-05,
+      "loss": 0.5756,
+      "step": 14684
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.574369982824686e-05,
+      "loss": 0.5968,
+      "step": 14685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.573270830866091e-05,
+      "loss": 0.5846,
+      "step": 14686
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.57217168407234e-05,
+      "loss": 0.4764,
+      "step": 14687
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.571072542456738e-05,
+      "loss": 0.6323,
+      "step": 14688
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.56997340603259e-05,
+      "loss": 0.5892,
+      "step": 14689
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.568874274813197e-05,
+      "loss": 0.6026,
+      "step": 14690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.567775148811861e-05,
+      "loss": 0.4605,
+      "step": 14691
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.566676028041889e-05,
+      "loss": 0.532,
+      "step": 14692
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.565576912516577e-05,
+      "loss": 0.5991,
+      "step": 14693
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.564477802249236e-05,
+      "loss": 0.5316,
+      "step": 14694
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.563378697253167e-05,
+      "loss": 0.5854,
+      "step": 14695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.562279597541667e-05,
+      "loss": 0.4872,
+      "step": 14696
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.561180503128045e-05,
+      "loss": 0.6212,
+      "step": 14697
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.560081414025603e-05,
+      "loss": 0.6486,
+      "step": 14698
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.55898233024764e-05,
+      "loss": 0.6302,
+      "step": 14699
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.557883251807462e-05,
+      "loss": 0.6806,
+      "step": 14700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.55678417871837e-05,
+      "loss": 0.5283,
+      "step": 14701
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.555685110993667e-05,
+      "loss": 0.4896,
+      "step": 14702
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.554586048646657e-05,
+      "loss": 0.5844,
+      "step": 14703
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.553486991690638e-05,
+      "loss": 0.5818,
+      "step": 14704
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.552387940138917e-05,
+      "loss": 0.5616,
+      "step": 14705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.55128889400479e-05,
+      "loss": 0.5194,
+      "step": 14706
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.550189853301567e-05,
+      "loss": 0.4133,
+      "step": 14707
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.549090818042547e-05,
+      "loss": 0.6001,
+      "step": 14708
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.547991788241028e-05,
+      "loss": 0.6067,
+      "step": 14709
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.546892763910316e-05,
+      "loss": 0.5789,
+      "step": 14710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.545793745063714e-05,
+      "loss": 0.4615,
+      "step": 14711
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.544694731714519e-05,
+      "loss": 0.5553,
+      "step": 14712
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.543595723876038e-05,
+      "loss": 0.5682,
+      "step": 14713
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.542496721561566e-05,
+      "loss": 0.6098,
+      "step": 14714
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.541397724784413e-05,
+      "loss": 0.5703,
+      "step": 14715
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.540298733557877e-05,
+      "loss": 0.5746,
+      "step": 14716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.539199747895254e-05,
+      "loss": 0.5702,
+      "step": 14717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.538100767809853e-05,
+      "loss": 0.513,
+      "step": 14718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.537001793314972e-05,
+      "loss": 0.5302,
+      "step": 14719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.535902824423912e-05,
+      "loss": 0.6188,
+      "step": 14720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.534803861149976e-05,
+      "loss": 0.5472,
+      "step": 14721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.533704903506464e-05,
+      "loss": 0.5914,
+      "step": 14722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.532605951506677e-05,
+      "loss": 0.4951,
+      "step": 14723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.531507005163915e-05,
+      "loss": 0.6035,
+      "step": 14724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.530408064491481e-05,
+      "loss": 0.5685,
+      "step": 14725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.529309129502675e-05,
+      "loss": 0.5158,
+      "step": 14726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.528210200210796e-05,
+      "loss": 0.5466,
+      "step": 14727
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.527111276629148e-05,
+      "loss": 0.4919,
+      "step": 14728
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.526012358771029e-05,
+      "loss": 0.5552,
+      "step": 14729
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.524913446649741e-05,
+      "loss": 0.5403,
+      "step": 14730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.523814540278583e-05,
+      "loss": 0.6635,
+      "step": 14731
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.522715639670856e-05,
+      "loss": 0.5422,
+      "step": 14732
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.521616744839862e-05,
+      "loss": 0.5675,
+      "step": 14733
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.5205178557989e-05,
+      "loss": 0.6113,
+      "step": 14734
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.519418972561269e-05,
+      "loss": 0.6254,
+      "step": 14735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.518320095140273e-05,
+      "loss": 0.5855,
+      "step": 14736
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.517221223549207e-05,
+      "loss": 0.5588,
+      "step": 14737
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.516122357801375e-05,
+      "loss": 0.5245,
+      "step": 14738
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.515023497910076e-05,
+      "loss": 0.5816,
+      "step": 14739
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.513924643888606e-05,
+      "loss": 0.5804,
+      "step": 14740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.512825795750272e-05,
+      "loss": 0.455,
+      "step": 14741
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.511726953508367e-05,
+      "loss": 0.4679,
+      "step": 14742
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.510628117176196e-05,
+      "loss": 0.5585,
+      "step": 14743
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.509529286767055e-05,
+      "loss": 0.5201,
+      "step": 14744
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.508430462294243e-05,
+      "loss": 0.5529,
+      "step": 14745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.507331643771063e-05,
+      "loss": 0.6291,
+      "step": 14746
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.506232831210812e-05,
+      "loss": 0.5333,
+      "step": 14747
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.50513402462679e-05,
+      "loss": 0.5249,
+      "step": 14748
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.504035224032295e-05,
+      "loss": 0.5658,
+      "step": 14749
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.502936429440626e-05,
+      "loss": 0.5603,
+      "step": 14750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.501837640865086e-05,
+      "loss": 0.568,
+      "step": 14751
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.500738858318971e-05,
+      "loss": 0.5268,
+      "step": 14752
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.499640081815578e-05,
+      "loss": 0.4955,
+      "step": 14753
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.498541311368208e-05,
+      "loss": 0.6618,
+      "step": 14754
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.49744254699016e-05,
+      "loss": 0.5048,
+      "step": 14755
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.496343788694734e-05,
+      "loss": 0.5019,
+      "step": 14756
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.495245036495226e-05,
+      "loss": 0.6081,
+      "step": 14757
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.494146290404935e-05,
+      "loss": 0.5331,
+      "step": 14758
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.493047550437161e-05,
+      "loss": 0.4715,
+      "step": 14759
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.491948816605203e-05,
+      "loss": 0.6192,
+      "step": 14760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.490850088922354e-05,
+      "loss": 0.5261,
+      "step": 14761
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.489751367401918e-05,
+      "loss": 0.6606,
+      "step": 14762
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.488652652057192e-05,
+      "loss": 0.5947,
+      "step": 14763
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.487553942901474e-05,
+      "loss": 0.5914,
+      "step": 14764
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.486455239948061e-05,
+      "loss": 0.6184,
+      "step": 14765
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.48535654321025e-05,
+      "loss": 0.5854,
+      "step": 14766
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.484257852701343e-05,
+      "loss": 0.5076,
+      "step": 14767
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.483159168434631e-05,
+      "loss": 0.6114,
+      "step": 14768
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.48206049042342e-05,
+      "loss": 0.5916,
+      "step": 14769
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.480961818681004e-05,
+      "loss": 0.5643,
+      "step": 14770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.47986315322068e-05,
+      "loss": 0.617,
+      "step": 14771
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.478764494055743e-05,
+      "loss": 0.4946,
+      "step": 14772
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.477665841199494e-05,
+      "loss": 0.4918,
+      "step": 14773
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.476567194665231e-05,
+      "loss": 0.5352,
+      "step": 14774
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.475468554466248e-05,
+      "loss": 0.5421,
+      "step": 14775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.474369920615844e-05,
+      "loss": 0.5408,
+      "step": 14776
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.473271293127315e-05,
+      "loss": 0.4664,
+      "step": 14777
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.472172672013962e-05,
+      "loss": 0.4369,
+      "step": 14778
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.47107405728908e-05,
+      "loss": 0.5478,
+      "step": 14779
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.469975448965961e-05,
+      "loss": 0.566,
+      "step": 14780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.468876847057908e-05,
+      "loss": 0.5055,
+      "step": 14781
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.467778251578217e-05,
+      "loss": 0.5207,
+      "step": 14782
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.46667966254018e-05,
+      "loss": 0.5688,
+      "step": 14783
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.4655810799571e-05,
+      "loss": 0.5705,
+      "step": 14784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.464482503842268e-05,
+      "loss": 0.6182,
+      "step": 14785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.463383934208985e-05,
+      "loss": 0.449,
+      "step": 14786
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.462285371070544e-05,
+      "loss": 0.5661,
+      "step": 14787
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.461186814440241e-05,
+      "loss": 0.4688,
+      "step": 14788
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.460088264331377e-05,
+      "loss": 0.5282,
+      "step": 14789
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.458989720757241e-05,
+      "loss": 0.5802,
+      "step": 14790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.457891183731136e-05,
+      "loss": 0.532,
+      "step": 14791
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.456792653266353e-05,
+      "loss": 0.5224,
+      "step": 14792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.45569412937619e-05,
+      "loss": 0.5238,
+      "step": 14793
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.454595612073944e-05,
+      "loss": 0.4837,
+      "step": 14794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.453497101372906e-05,
+      "loss": 0.633,
+      "step": 14795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.452398597286375e-05,
+      "loss": 0.6117,
+      "step": 14796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.45130009982765e-05,
+      "loss": 0.5728,
+      "step": 14797
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.45020160901002e-05,
+      "loss": 0.5293,
+      "step": 14798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.449103124846784e-05,
+      "loss": 0.5863,
+      "step": 14799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.448004647351238e-05,
+      "loss": 0.4776,
+      "step": 14800
+    },
+    {
+      "epoch": 0.52,
+      "eval_loss": 0.5387113690376282,
+      "eval_runtime": 845.9644,
+      "eval_samples_per_second": 28.441,
+      "eval_steps_per_second": 14.22,
+      "step": 14800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.446906176536673e-05,
+      "loss": 0.5673,
+      "step": 14801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.445807712416389e-05,
+      "loss": 0.5682,
+      "step": 14802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.444709255003678e-05,
+      "loss": 0.5942,
+      "step": 14803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.443610804311837e-05,
+      "loss": 0.4998,
+      "step": 14804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.44251236035416e-05,
+      "loss": 0.5507,
+      "step": 14805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.44141392314394e-05,
+      "loss": 0.5807,
+      "step": 14806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.440315492694476e-05,
+      "loss": 0.4851,
+      "step": 14807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.439217069019058e-05,
+      "loss": 0.5742,
+      "step": 14808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.438118652130983e-05,
+      "loss": 0.6057,
+      "step": 14809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.437020242043548e-05,
+      "loss": 0.4802,
+      "step": 14810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.435921838770042e-05,
+      "loss": 0.5743,
+      "step": 14811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.434823442323763e-05,
+      "loss": 0.5924,
+      "step": 14812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.433725052718006e-05,
+      "loss": 0.5583,
+      "step": 14813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.43262666996606e-05,
+      "loss": 0.4949,
+      "step": 14814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.431528294081225e-05,
+      "loss": 0.5622,
+      "step": 14815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.430429925076791e-05,
+      "loss": 0.6268,
+      "step": 14816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.429331562966056e-05,
+      "loss": 0.5794,
+      "step": 14817
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.42823320776231e-05,
+      "loss": 0.5484,
+      "step": 14818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.427134859478847e-05,
+      "loss": 0.6035,
+      "step": 14819
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.426036518128965e-05,
+      "loss": 0.5213,
+      "step": 14820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.42493818372595e-05,
+      "loss": 0.5212,
+      "step": 14821
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.423839856283103e-05,
+      "loss": 0.6209,
+      "step": 14822
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.422741535813714e-05,
+      "loss": 0.553,
+      "step": 14823
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.421643222331075e-05,
+      "loss": 0.5141,
+      "step": 14824
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.420544915848484e-05,
+      "loss": 0.5916,
+      "step": 14825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.419446616379227e-05,
+      "loss": 0.5199,
+      "step": 14826
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.418348323936605e-05,
+      "loss": 0.4656,
+      "step": 14827
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.417250038533907e-05,
+      "loss": 0.5477,
+      "step": 14828
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.416151760184423e-05,
+      "loss": 0.5645,
+      "step": 14829
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.415053488901451e-05,
+      "loss": 0.557,
+      "step": 14830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.413955224698282e-05,
+      "loss": 0.4968,
+      "step": 14831
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.412856967588207e-05,
+      "loss": 0.6871,
+      "step": 14832
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.411758717584521e-05,
+      "loss": 0.5964,
+      "step": 14833
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.410660474700514e-05,
+      "loss": 0.5194,
+      "step": 14834
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.409562238949481e-05,
+      "loss": 0.511,
+      "step": 14835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.408464010344714e-05,
+      "loss": 0.5469,
+      "step": 14836
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.407365788899502e-05,
+      "loss": 0.5228,
+      "step": 14837
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.406267574627142e-05,
+      "loss": 0.5188,
+      "step": 14838
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.40516936754092e-05,
+      "loss": 0.5865,
+      "step": 14839
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.404071167654135e-05,
+      "loss": 0.5655,
+      "step": 14840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.402972974980075e-05,
+      "loss": 0.534,
+      "step": 14841
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.401874789532031e-05,
+      "loss": 0.5876,
+      "step": 14842
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.400776611323296e-05,
+      "loss": 0.5355,
+      "step": 14843
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.39967844036716e-05,
+      "loss": 0.4795,
+      "step": 14844
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.398580276676919e-05,
+      "loss": 0.5948,
+      "step": 14845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.39748212026586e-05,
+      "loss": 0.5119,
+      "step": 14846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.396383971147275e-05,
+      "loss": 0.5376,
+      "step": 14847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.395285829334458e-05,
+      "loss": 0.5596,
+      "step": 14848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.394187694840699e-05,
+      "loss": 0.4959,
+      "step": 14849
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.393089567679284e-05,
+      "loss": 0.565,
+      "step": 14850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.391991447863512e-05,
+      "loss": 0.5564,
+      "step": 14851
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.390893335406668e-05,
+      "loss": 0.4558,
+      "step": 14852
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.389795230322046e-05,
+      "loss": 0.4924,
+      "step": 14853
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.388697132622938e-05,
+      "loss": 0.5953,
+      "step": 14854
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.38759904232263e-05,
+      "loss": 0.5815,
+      "step": 14855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.386500959434414e-05,
+      "loss": 0.597,
+      "step": 14856
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.385402883971583e-05,
+      "loss": 0.5817,
+      "step": 14857
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.384304815947425e-05,
+      "loss": 0.6491,
+      "step": 14858
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.383206755375234e-05,
+      "loss": 0.6314,
+      "step": 14859
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.382108702268295e-05,
+      "loss": 0.5168,
+      "step": 14860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.3810106566399e-05,
+      "loss": 0.5136,
+      "step": 14861
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.37991261850334e-05,
+      "loss": 0.5989,
+      "step": 14862
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.378814587871903e-05,
+      "loss": 0.568,
+      "step": 14863
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.377716564758883e-05,
+      "loss": 0.6256,
+      "step": 14864
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.376618549177564e-05,
+      "loss": 0.6037,
+      "step": 14865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.375520541141241e-05,
+      "loss": 0.5785,
+      "step": 14866
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.374422540663201e-05,
+      "loss": 0.5879,
+      "step": 14867
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.373324547756732e-05,
+      "loss": 0.6056,
+      "step": 14868
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.372226562435126e-05,
+      "loss": 0.5862,
+      "step": 14869
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.371128584711669e-05,
+      "loss": 0.5142,
+      "step": 14870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.370030614599655e-05,
+      "loss": 0.5313,
+      "step": 14871
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.368932652112371e-05,
+      "loss": 0.5867,
+      "step": 14872
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.367834697263104e-05,
+      "loss": 0.5501,
+      "step": 14873
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.366736750065145e-05,
+      "loss": 0.6393,
+      "step": 14874
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.365638810531781e-05,
+      "loss": 0.5236,
+      "step": 14875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.364540878676304e-05,
+      "loss": 0.5314,
+      "step": 14876
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.363442954512e-05,
+      "loss": 0.599,
+      "step": 14877
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.362345038052157e-05,
+      "loss": 0.5105,
+      "step": 14878
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.361247129310067e-05,
+      "loss": 0.503,
+      "step": 14879
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.360149228299016e-05,
+      "loss": 0.5039,
+      "step": 14880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.359051335032288e-05,
+      "loss": 0.5593,
+      "step": 14881
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.35795344952318e-05,
+      "loss": 0.4728,
+      "step": 14882
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.356855571784973e-05,
+      "loss": 0.5713,
+      "step": 14883
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.355757701830958e-05,
+      "loss": 0.5607,
+      "step": 14884
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.354659839674424e-05,
+      "loss": 0.5601,
+      "step": 14885
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.353561985328654e-05,
+      "loss": 0.587,
+      "step": 14886
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.352464138806941e-05,
+      "loss": 0.4835,
+      "step": 14887
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.35136630012257e-05,
+      "loss": 0.6166,
+      "step": 14888
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.350268469288828e-05,
+      "loss": 0.4504,
+      "step": 14889
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.349170646319006e-05,
+      "loss": 0.561,
+      "step": 14890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.348072831226386e-05,
+      "loss": 0.5969,
+      "step": 14891
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.34697502402426e-05,
+      "loss": 0.4814,
+      "step": 14892
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.34587722472591e-05,
+      "loss": 0.5609,
+      "step": 14893
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.34477943334463e-05,
+      "loss": 0.5429,
+      "step": 14894
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.3436816498937e-05,
+      "loss": 0.464,
+      "step": 14895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.34258387438641e-05,
+      "loss": 0.6691,
+      "step": 14896
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.341486106836049e-05,
+      "loss": 0.5513,
+      "step": 14897
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.340388347255901e-05,
+      "loss": 0.5354,
+      "step": 14898
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.339290595659249e-05,
+      "loss": 0.6106,
+      "step": 14899
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.338192852059387e-05,
+      "loss": 0.5743,
+      "step": 14900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.337095116469595e-05,
+      "loss": 0.5058,
+      "step": 14901
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.335997388903163e-05,
+      "loss": 0.4986,
+      "step": 14902
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.334899669373379e-05,
+      "loss": 0.4968,
+      "step": 14903
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.333801957893521e-05,
+      "loss": 0.552,
+      "step": 14904
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.332704254476885e-05,
+      "loss": 0.5736,
+      "step": 14905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.33160655913675e-05,
+      "loss": 0.5518,
+      "step": 14906
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.330508871886404e-05,
+      "loss": 0.5249,
+      "step": 14907
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.32941119273913e-05,
+      "loss": 0.4895,
+      "step": 14908
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.328313521708218e-05,
+      "loss": 0.5644,
+      "step": 14909
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.327215858806949e-05,
+      "loss": 0.5073,
+      "step": 14910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.326118204048613e-05,
+      "loss": 0.5915,
+      "step": 14911
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.325020557446494e-05,
+      "loss": 0.5172,
+      "step": 14912
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.323922919013873e-05,
+      "loss": 0.5689,
+      "step": 14913
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.322825288764042e-05,
+      "loss": 0.6348,
+      "step": 14914
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.32172766671028e-05,
+      "loss": 0.5336,
+      "step": 14915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.320630052865874e-05,
+      "loss": 0.5673,
+      "step": 14916
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.319532447244111e-05,
+      "loss": 0.6234,
+      "step": 14917
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.31843484985827e-05,
+      "loss": 0.5464,
+      "step": 14918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.317337260721644e-05,
+      "loss": 0.5863,
+      "step": 14919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.31623967984751e-05,
+      "loss": 0.4416,
+      "step": 14920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.315142107249155e-05,
+      "loss": 0.5948,
+      "step": 14921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.314044542939865e-05,
+      "loss": 0.5268,
+      "step": 14922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.31294698693292e-05,
+      "loss": 0.4965,
+      "step": 14923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.311849439241608e-05,
+      "loss": 0.5045,
+      "step": 14924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.310751899879215e-05,
+      "loss": 0.5047,
+      "step": 14925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.309654368859017e-05,
+      "loss": 0.4851,
+      "step": 14926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.308556846194304e-05,
+      "loss": 0.4871,
+      "step": 14927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.307459331898357e-05,
+      "loss": 0.5825,
+      "step": 14928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.306361825984462e-05,
+      "loss": 0.5959,
+      "step": 14929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.305264328465902e-05,
+      "loss": 0.4755,
+      "step": 14930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.304166839355957e-05,
+      "loss": 0.4232,
+      "step": 14931
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.303069358667915e-05,
+      "loss": 0.508,
+      "step": 14932
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.301971886415058e-05,
+      "loss": 0.5102,
+      "step": 14933
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.300874422610663e-05,
+      "loss": 0.5086,
+      "step": 14934
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.299776967268022e-05,
+      "loss": 0.4772,
+      "step": 14935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.298679520400412e-05,
+      "loss": 0.5699,
+      "step": 14936
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.29758208202112e-05,
+      "loss": 0.5477,
+      "step": 14937
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.296484652143425e-05,
+      "loss": 0.5198,
+      "step": 14938
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.29538723078061e-05,
+      "loss": 0.5316,
+      "step": 14939
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.294289817945959e-05,
+      "loss": 0.5078,
+      "step": 14940
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.293192413652752e-05,
+      "loss": 0.5013,
+      "step": 14941
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.292095017914275e-05,
+      "loss": 0.5244,
+      "step": 14942
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.29099763074381e-05,
+      "loss": 0.541,
+      "step": 14943
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.289900252154633e-05,
+      "loss": 0.5629,
+      "step": 14944
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.288802882160032e-05,
+      "loss": 0.5699,
+      "step": 14945
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.287705520773286e-05,
+      "loss": 0.5098,
+      "step": 14946
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.286608168007678e-05,
+      "loss": 0.5771,
+      "step": 14947
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.285510823876489e-05,
+      "loss": 0.5486,
+      "step": 14948
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.284413488393001e-05,
+      "loss": 0.5412,
+      "step": 14949
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.283316161570496e-05,
+      "loss": 0.5716,
+      "step": 14950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.282218843422254e-05,
+      "loss": 0.5815,
+      "step": 14951
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.281121533961555e-05,
+      "loss": 0.6216,
+      "step": 14952
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.280024233201685e-05,
+      "loss": 0.5433,
+      "step": 14953
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.278926941155919e-05,
+      "loss": 0.4924,
+      "step": 14954
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.277829657837542e-05,
+      "loss": 0.4515,
+      "step": 14955
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.276732383259834e-05,
+      "loss": 0.6019,
+      "step": 14956
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.275635117436073e-05,
+      "loss": 0.5929,
+      "step": 14957
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.274537860379544e-05,
+      "loss": 0.5001,
+      "step": 14958
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.273440612103522e-05,
+      "loss": 0.5733,
+      "step": 14959
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.272343372621294e-05,
+      "loss": 0.6214,
+      "step": 14960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.271246141946136e-05,
+      "loss": 0.5587,
+      "step": 14961
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.270148920091328e-05,
+      "loss": 0.4817,
+      "step": 14962
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.26905170707015e-05,
+      "loss": 0.6379,
+      "step": 14963
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.267954502895887e-05,
+      "loss": 0.5049,
+      "step": 14964
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.266857307581811e-05,
+      "loss": 0.5497,
+      "step": 14965
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.265760121141207e-05,
+      "loss": 0.4948,
+      "step": 14966
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.26466294358735e-05,
+      "loss": 0.5516,
+      "step": 14967
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.263565774933527e-05,
+      "loss": 0.5744,
+      "step": 14968
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.262468615193011e-05,
+      "loss": 0.5826,
+      "step": 14969
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.261371464379084e-05,
+      "loss": 0.5682,
+      "step": 14970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.260274322505025e-05,
+      "loss": 0.5453,
+      "step": 14971
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.259177189584109e-05,
+      "loss": 0.5035,
+      "step": 14972
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.258080065629622e-05,
+      "loss": 0.6077,
+      "step": 14973
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.25698295065484e-05,
+      "loss": 0.5253,
+      "step": 14974
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.255885844673038e-05,
+      "loss": 0.4762,
+      "step": 14975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.2547887476975e-05,
+      "loss": 0.5459,
+      "step": 14976
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.2536916597415e-05,
+      "loss": 0.4398,
+      "step": 14977
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.252594580818321e-05,
+      "loss": 0.5489,
+      "step": 14978
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.25149751094124e-05,
+      "loss": 0.6036,
+      "step": 14979
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.25040045012353e-05,
+      "loss": 0.4838,
+      "step": 14980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.249303398378478e-05,
+      "loss": 0.6257,
+      "step": 14981
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.248206355719354e-05,
+      "loss": 0.5808,
+      "step": 14982
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.24710932215944e-05,
+      "loss": 0.6933,
+      "step": 14983
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.246012297712014e-05,
+      "loss": 0.6314,
+      "step": 14984
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.24491528239035e-05,
+      "loss": 0.5587,
+      "step": 14985
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.243818276207731e-05,
+      "loss": 0.4876,
+      "step": 14986
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.242721279177431e-05,
+      "loss": 0.5906,
+      "step": 14987
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.241624291312725e-05,
+      "loss": 0.4846,
+      "step": 14988
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.240527312626896e-05,
+      "loss": 0.535,
+      "step": 14989
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.239430343133217e-05,
+      "loss": 0.6475,
+      "step": 14990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.238333382844966e-05,
+      "loss": 0.5341,
+      "step": 14991
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.237236431775422e-05,
+      "loss": 0.5441,
+      "step": 14992
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.236139489937858e-05,
+      "loss": 0.6182,
+      "step": 14993
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.235042557345554e-05,
+      "loss": 0.5686,
+      "step": 14994
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.233945634011782e-05,
+      "loss": 0.5642,
+      "step": 14995
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.232848719949825e-05,
+      "loss": 0.6239,
+      "step": 14996
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.231751815172955e-05,
+      "loss": 0.4819,
+      "step": 14997
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.230654919694447e-05,
+      "loss": 0.4872,
+      "step": 14998
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.229558033527582e-05,
+      "loss": 0.5897,
+      "step": 14999
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.22846115668563e-05,
+      "loss": 0.493,
+      "step": 15000
+    },
+    {
+      "epoch": 0.52,
+      "eval_loss": 0.5356741547584534,
+      "eval_runtime": 850.642,
+      "eval_samples_per_second": 28.285,
+      "eval_steps_per_second": 14.142,
+      "step": 15000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.22736428918187e-05,
+      "loss": 0.5269,
+      "step": 15001
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.22626743102958e-05,
+      "loss": 0.5947,
+      "step": 15002
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.225170582242031e-05,
+      "loss": 0.5791,
+      "step": 15003
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.224073742832501e-05,
+      "loss": 0.5249,
+      "step": 15004
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.222976912814269e-05,
+      "loss": 0.5648,
+      "step": 15005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.221880092200601e-05,
+      "loss": 0.4947,
+      "step": 15006
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.22078328100478e-05,
+      "loss": 0.5495,
+      "step": 15007
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.21968647924008e-05,
+      "loss": 0.5756,
+      "step": 15008
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.218589686919774e-05,
+      "loss": 0.7151,
+      "step": 15009
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.217492904057138e-05,
+      "loss": 0.5932,
+      "step": 15010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.216396130665444e-05,
+      "loss": 0.5353,
+      "step": 15011
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.215299366757972e-05,
+      "loss": 0.4727,
+      "step": 15012
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.214202612347993e-05,
+      "loss": 0.49,
+      "step": 15013
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.21310586744878e-05,
+      "loss": 0.4916,
+      "step": 15014
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.212009132073611e-05,
+      "loss": 0.5161,
+      "step": 15015
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.210912406235755e-05,
+      "loss": 0.5392,
+      "step": 15016
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.209815689948493e-05,
+      "loss": 0.6153,
+      "step": 15017
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.208718983225093e-05,
+      "loss": 0.5255,
+      "step": 15018
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.207622286078831e-05,
+      "loss": 0.4367,
+      "step": 15019
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.206525598522982e-05,
+      "loss": 0.5612,
+      "step": 15020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.205428920570815e-05,
+      "loss": 0.4845,
+      "step": 15021
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.204332252235611e-05,
+      "loss": 0.5244,
+      "step": 15022
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.203235593530638e-05,
+      "loss": 0.4922,
+      "step": 15023
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.202138944469168e-05,
+      "loss": 0.6106,
+      "step": 15024
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.201042305064479e-05,
+      "loss": 0.4988,
+      "step": 15025
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.19994567532984e-05,
+      "loss": 0.5555,
+      "step": 15026
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.198849055278526e-05,
+      "loss": 0.4863,
+      "step": 15027
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.19775244492381e-05,
+      "loss": 0.5107,
+      "step": 15028
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.19665584427896e-05,
+      "loss": 0.6217,
+      "step": 15029
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.195559253357255e-05,
+      "loss": 0.5167,
+      "step": 15030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.194462672171966e-05,
+      "loss": 0.4893,
+      "step": 15031
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.19336610073636e-05,
+      "loss": 0.4784,
+      "step": 15032
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.192269539063717e-05,
+      "loss": 0.5897,
+      "step": 15033
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.191172987167301e-05,
+      "loss": 0.6248,
+      "step": 15034
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.190076445060392e-05,
+      "loss": 0.5319,
+      "step": 15035
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.188979912756257e-05,
+      "loss": 0.5297,
+      "step": 15036
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.187883390268167e-05,
+      "loss": 0.484,
+      "step": 15037
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.186786877609396e-05,
+      "loss": 0.5305,
+      "step": 15038
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.185690374793217e-05,
+      "loss": 0.5345,
+      "step": 15039
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.184593881832897e-05,
+      "loss": 0.5164,
+      "step": 15040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.183497398741707e-05,
+      "loss": 0.5719,
+      "step": 15041
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.182400925532923e-05,
+      "loss": 0.6437,
+      "step": 15042
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.18130446221981e-05,
+      "loss": 0.6053,
+      "step": 15043
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.180208008815646e-05,
+      "loss": 0.5225,
+      "step": 15044
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.179111565333695e-05,
+      "loss": 0.4486,
+      "step": 15045
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.178015131787231e-05,
+      "loss": 0.552,
+      "step": 15046
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.176918708189524e-05,
+      "loss": 0.5265,
+      "step": 15047
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.175822294553844e-05,
+      "loss": 0.5446,
+      "step": 15048
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.174725890893463e-05,
+      "loss": 0.5513,
+      "step": 15049
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.17362949722165e-05,
+      "loss": 0.5292,
+      "step": 15050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.172533113551672e-05,
+      "loss": 0.506,
+      "step": 15051
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.171436739896804e-05,
+      "loss": 0.4647,
+      "step": 15052
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.170340376270315e-05,
+      "loss": 0.5835,
+      "step": 15053
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.169244022685469e-05,
+      "loss": 0.5028,
+      "step": 15054
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.168147679155542e-05,
+      "loss": 0.5486,
+      "step": 15055
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.167051345693799e-05,
+      "loss": 0.487,
+      "step": 15056
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.165955022313513e-05,
+      "loss": 0.5115,
+      "step": 15057
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.164858709027952e-05,
+      "loss": 0.4729,
+      "step": 15058
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.163762405850382e-05,
+      "loss": 0.6735,
+      "step": 15059
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.162666112794078e-05,
+      "loss": 0.5234,
+      "step": 15060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.161569829872301e-05,
+      "loss": 0.5628,
+      "step": 15061
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.160473557098329e-05,
+      "loss": 0.5097,
+      "step": 15062
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.159377294485423e-05,
+      "loss": 0.5219,
+      "step": 15063
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.158281042046852e-05,
+      "loss": 0.5654,
+      "step": 15064
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.157184799795889e-05,
+      "loss": 0.5987,
+      "step": 15065
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.156088567745799e-05,
+      "loss": 0.542,
+      "step": 15066
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.154992345909848e-05,
+      "loss": 0.4744,
+      "step": 15067
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.153896134301309e-05,
+      "loss": 0.5845,
+      "step": 15068
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.152799932933446e-05,
+      "loss": 0.5303,
+      "step": 15069
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.151703741819529e-05,
+      "loss": 0.6674,
+      "step": 15070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.150607560972825e-05,
+      "loss": 0.5233,
+      "step": 15071
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.149511390406598e-05,
+      "loss": 0.5471,
+      "step": 15072
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.148415230134121e-05,
+      "loss": 0.6328,
+      "step": 15073
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.147319080168655e-05,
+      "loss": 0.5411,
+      "step": 15074
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.146222940523474e-05,
+      "loss": 0.4594,
+      "step": 15075
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.145126811211842e-05,
+      "loss": 0.494,
+      "step": 15076
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.144030692247022e-05,
+      "loss": 0.4449,
+      "step": 15077
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.142934583642286e-05,
+      "loss": 0.5511,
+      "step": 15078
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.141838485410897e-05,
+      "loss": 0.5953,
+      "step": 15079
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.140742397566126e-05,
+      "loss": 0.5398,
+      "step": 15080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.139646320121235e-05,
+      "loss": 0.578,
+      "step": 15081
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.138550253089489e-05,
+      "loss": 0.5101,
+      "step": 15082
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.137454196484159e-05,
+      "loss": 0.4935,
+      "step": 15083
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.13635815031851e-05,
+      "loss": 0.5676,
+      "step": 15084
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.135262114605801e-05,
+      "loss": 0.5122,
+      "step": 15085
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.134166089359307e-05,
+      "loss": 0.5605,
+      "step": 15086
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.133070074592288e-05,
+      "loss": 0.6075,
+      "step": 15087
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.131974070318012e-05,
+      "loss": 0.6153,
+      "step": 15088
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.130878076549743e-05,
+      "loss": 0.5636,
+      "step": 15089
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.129782093300746e-05,
+      "loss": 0.5974,
+      "step": 15090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.128686120584288e-05,
+      "loss": 0.5452,
+      "step": 15091
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.12759015841363e-05,
+      "loss": 0.5537,
+      "step": 15092
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.126494206802041e-05,
+      "loss": 0.6546,
+      "step": 15093
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.125398265762784e-05,
+      "loss": 0.4968,
+      "step": 15094
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.124302335309122e-05,
+      "loss": 0.4255,
+      "step": 15095
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.123206415454322e-05,
+      "loss": 0.5182,
+      "step": 15096
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.122110506211645e-05,
+      "loss": 0.5694,
+      "step": 15097
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.121014607594361e-05,
+      "loss": 0.5272,
+      "step": 15098
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.119918719615728e-05,
+      "loss": 0.6009,
+      "step": 15099
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.118822842289011e-05,
+      "loss": 0.5699,
+      "step": 15100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.117726975627477e-05,
+      "loss": 0.532,
+      "step": 15101
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.116631119644389e-05,
+      "loss": 0.5343,
+      "step": 15102
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.115535274353005e-05,
+      "loss": 0.6168,
+      "step": 15103
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.114439439766595e-05,
+      "loss": 0.634,
+      "step": 15104
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.113343615898418e-05,
+      "loss": 0.6036,
+      "step": 15105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.112247802761742e-05,
+      "loss": 0.4925,
+      "step": 15106
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.111152000369825e-05,
+      "loss": 0.5761,
+      "step": 15107
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.110056208735931e-05,
+      "loss": 0.5818,
+      "step": 15108
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.108960427873326e-05,
+      "loss": 0.5148,
+      "step": 15109
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.107864657795268e-05,
+      "loss": 0.5849,
+      "step": 15110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.106768898515022e-05,
+      "loss": 0.5146,
+      "step": 15111
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.10567315004585e-05,
+      "loss": 0.4846,
+      "step": 15112
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.104577412401013e-05,
+      "loss": 0.5204,
+      "step": 15113
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.103481685593776e-05,
+      "loss": 0.6438,
+      "step": 15114
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.1023859696374e-05,
+      "loss": 0.5393,
+      "step": 15115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.101290264545141e-05,
+      "loss": 0.5259,
+      "step": 15116
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.10019457033027e-05,
+      "loss": 0.5358,
+      "step": 15117
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.099098887006042e-05,
+      "loss": 0.4805,
+      "step": 15118
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.098003214585722e-05,
+      "loss": 0.5801,
+      "step": 15119
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.09690755308257e-05,
+      "loss": 0.5217,
+      "step": 15120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.095811902509846e-05,
+      "loss": 0.5281,
+      "step": 15121
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.094716262880812e-05,
+      "loss": 0.6277,
+      "step": 15122
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.093620634208728e-05,
+      "loss": 0.4751,
+      "step": 15123
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.092525016506858e-05,
+      "loss": 0.5901,
+      "step": 15124
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.09142940978846e-05,
+      "loss": 0.657,
+      "step": 15125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.090333814066793e-05,
+      "loss": 0.5697,
+      "step": 15126
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.089238229355121e-05,
+      "loss": 0.5282,
+      "step": 15127
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.088142655666701e-05,
+      "loss": 0.5451,
+      "step": 15128
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.087047093014796e-05,
+      "loss": 0.5424,
+      "step": 15129
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.085951541412664e-05,
+      "loss": 0.6267,
+      "step": 15130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.084856000873564e-05,
+      "loss": 0.5894,
+      "step": 15131
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.083760471410759e-05,
+      "loss": 0.5936,
+      "step": 15132
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.082664953037506e-05,
+      "loss": 0.5556,
+      "step": 15133
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.081569445767063e-05,
+      "loss": 0.5103,
+      "step": 15134
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.080473949612694e-05,
+      "loss": 0.5167,
+      "step": 15135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.079378464587651e-05,
+      "loss": 0.6426,
+      "step": 15136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.078282990705203e-05,
+      "loss": 0.5532,
+      "step": 15137
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.077187527978602e-05,
+      "loss": 0.4933,
+      "step": 15138
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.076092076421104e-05,
+      "loss": 0.4484,
+      "step": 15139
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.074996636045976e-05,
+      "loss": 0.5428,
+      "step": 15140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.073901206866468e-05,
+      "loss": 0.7118,
+      "step": 15141
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.072805788895847e-05,
+      "loss": 0.494,
+      "step": 15142
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.071710382147365e-05,
+      "loss": 0.5597,
+      "step": 15143
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.07061498663428e-05,
+      "loss": 0.5315,
+      "step": 15144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.069519602369856e-05,
+      "loss": 0.4744,
+      "step": 15145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.068424229367342e-05,
+      "loss": 0.4764,
+      "step": 15146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.067328867640002e-05,
+      "loss": 0.4855,
+      "step": 15147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.066233517201093e-05,
+      "loss": 0.4996,
+      "step": 15148
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.065138178063869e-05,
+      "loss": 0.5025,
+      "step": 15149
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.064042850241591e-05,
+      "loss": 0.5403,
+      "step": 15150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.062947533747514e-05,
+      "loss": 0.541,
+      "step": 15151
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.061852228594895e-05,
+      "loss": 0.5282,
+      "step": 15152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.060756934796993e-05,
+      "loss": 0.505,
+      "step": 15153
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.05966165236706e-05,
+      "loss": 0.5174,
+      "step": 15154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.058566381318358e-05,
+      "loss": 0.5411,
+      "step": 15155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.057471121664141e-05,
+      "loss": 0.5458,
+      "step": 15156
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.056375873417662e-05,
+      "loss": 0.4931,
+      "step": 15157
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.055280636592184e-05,
+      "loss": 0.5195,
+      "step": 15158
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.054185411200957e-05,
+      "loss": 0.4865,
+      "step": 15159
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.053090197257241e-05,
+      "loss": 0.5474,
+      "step": 15160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.051994994774291e-05,
+      "loss": 0.4535,
+      "step": 15161
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.050899803765359e-05,
+      "loss": 0.6639,
+      "step": 15162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.049804624243706e-05,
+      "loss": 0.4803,
+      "step": 15163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.048709456222581e-05,
+      "loss": 0.5494,
+      "step": 15164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.047614299715246e-05,
+      "loss": 0.5456,
+      "step": 15165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.046519154734952e-05,
+      "loss": 0.5395,
+      "step": 15166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.045424021294953e-05,
+      "loss": 0.5072,
+      "step": 15167
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.044328899408508e-05,
+      "loss": 0.5282,
+      "step": 15168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.043233789088868e-05,
+      "loss": 0.5878,
+      "step": 15169
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.042138690349287e-05,
+      "loss": 0.5794,
+      "step": 15170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.041043603203022e-05,
+      "loss": 0.547,
+      "step": 15171
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.039948527663325e-05,
+      "loss": 0.6236,
+      "step": 15172
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.038853463743452e-05,
+      "loss": 0.4676,
+      "step": 15173
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.037758411456659e-05,
+      "loss": 0.5318,
+      "step": 15174
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.036663370816194e-05,
+      "loss": 0.6772,
+      "step": 15175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.035568341835312e-05,
+      "loss": 0.5921,
+      "step": 15176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.03447332452727e-05,
+      "loss": 0.5866,
+      "step": 15177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.033378318905319e-05,
+      "loss": 0.5334,
+      "step": 15178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.032283324982711e-05,
+      "loss": 0.557,
+      "step": 15179
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.031188342772703e-05,
+      "loss": 0.6382,
+      "step": 15180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.030093372288541e-05,
+      "loss": 0.5602,
+      "step": 15181
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.028998413543488e-05,
+      "loss": 0.5339,
+      "step": 15182
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.027903466550788e-05,
+      "loss": 0.6185,
+      "step": 15183
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.026808531323695e-05,
+      "loss": 0.5512,
+      "step": 15184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.025713607875467e-05,
+      "loss": 0.5285,
+      "step": 15185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.02461869621935e-05,
+      "loss": 0.5558,
+      "step": 15186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.023523796368597e-05,
+      "loss": 0.5862,
+      "step": 15187
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.022428908336464e-05,
+      "loss": 0.5293,
+      "step": 15188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.021334032136196e-05,
+      "loss": 0.5318,
+      "step": 15189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.020239167781052e-05,
+      "loss": 0.4631,
+      "step": 15190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.019144315284278e-05,
+      "loss": 0.5144,
+      "step": 15191
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.018049474659128e-05,
+      "loss": 0.5521,
+      "step": 15192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.016954645918852e-05,
+      "loss": 0.6249,
+      "step": 15193
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.015859829076701e-05,
+      "loss": 0.4734,
+      "step": 15194
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.014765024145929e-05,
+      "loss": 0.632,
+      "step": 15195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.013670231139785e-05,
+      "loss": 0.4646,
+      "step": 15196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.012575450071517e-05,
+      "loss": 0.6163,
+      "step": 15197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.011480680954379e-05,
+      "loss": 0.5145,
+      "step": 15198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.010385923801616e-05,
+      "loss": 0.5255,
+      "step": 15199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.009291178626486e-05,
+      "loss": 0.5464,
+      "step": 15200
+    },
+    {
+      "epoch": 0.53,
+      "eval_loss": 0.5304505228996277,
+      "eval_runtime": 845.7643,
+      "eval_samples_per_second": 28.448,
+      "eval_steps_per_second": 14.224,
+      "step": 15200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.008196445442237e-05,
+      "loss": 0.5681,
+      "step": 15201
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.007101724262112e-05,
+      "loss": 0.454,
+      "step": 15202
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.006007015099369e-05,
+      "loss": 0.4976,
+      "step": 15203
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.004912317967256e-05,
+      "loss": 0.5033,
+      "step": 15204
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.003817632879017e-05,
+      "loss": 0.4336,
+      "step": 15205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.002722959847907e-05,
+      "loss": 0.5366,
+      "step": 15206
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.001628298887174e-05,
+      "loss": 0.5985,
+      "step": 15207
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.000533650010068e-05,
+      "loss": 0.4748,
+      "step": 15208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.999439013229835e-05,
+      "loss": 0.5282,
+      "step": 15209
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.998344388559725e-05,
+      "loss": 0.5697,
+      "step": 15210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.997249776012987e-05,
+      "loss": 0.4908,
+      "step": 15211
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.996155175602868e-05,
+      "loss": 0.5364,
+      "step": 15212
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.995060587342621e-05,
+      "loss": 0.5389,
+      "step": 15213
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.99396601124549e-05,
+      "loss": 0.4944,
+      "step": 15214
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.99287144732472e-05,
+      "loss": 0.5886,
+      "step": 15215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.991776895593567e-05,
+      "loss": 0.6493,
+      "step": 15216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.990682356065274e-05,
+      "loss": 0.5046,
+      "step": 15217
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.989587828753087e-05,
+      "loss": 0.6143,
+      "step": 15218
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.988493313670256e-05,
+      "loss": 0.4794,
+      "step": 15219
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.987398810830027e-05,
+      "loss": 0.5365,
+      "step": 15220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.98630432024565e-05,
+      "loss": 0.602,
+      "step": 15221
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.985209841930369e-05,
+      "loss": 0.548,
+      "step": 15222
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.984115375897431e-05,
+      "loss": 0.5641,
+      "step": 15223
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.983020922160084e-05,
+      "loss": 0.5307,
+      "step": 15224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.981926480731573e-05,
+      "loss": 0.4726,
+      "step": 15225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.980832051625147e-05,
+      "loss": 0.508,
+      "step": 15226
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.97973763485405e-05,
+      "loss": 0.5308,
+      "step": 15227
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.978643230431527e-05,
+      "loss": 0.5919,
+      "step": 15228
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.977548838370827e-05,
+      "loss": 0.4792,
+      "step": 15229
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.976454458685192e-05,
+      "loss": 0.5666,
+      "step": 15230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.975360091387873e-05,
+      "loss": 0.5956,
+      "step": 15231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.974265736492113e-05,
+      "loss": 0.5225,
+      "step": 15232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.973171394011154e-05,
+      "loss": 0.6233,
+      "step": 15233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.972077063958246e-05,
+      "loss": 0.5875,
+      "step": 15234
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.970982746346632e-05,
+      "loss": 0.5895,
+      "step": 15235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.969888441189555e-05,
+      "loss": 0.543,
+      "step": 15236
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.968794148500264e-05,
+      "loss": 0.609,
+      "step": 15237
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.967699868291999e-05,
+      "loss": 0.5678,
+      "step": 15238
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.966605600578008e-05,
+      "loss": 0.4584,
+      "step": 15239
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.965511345371536e-05,
+      "loss": 0.6256,
+      "step": 15240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.964417102685821e-05,
+      "loss": 0.5338,
+      "step": 15241
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.963322872534114e-05,
+      "loss": 0.6478,
+      "step": 15242
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.962228654929654e-05,
+      "loss": 0.5949,
+      "step": 15243
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.961134449885688e-05,
+      "loss": 0.5794,
+      "step": 15244
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.960040257415459e-05,
+      "loss": 0.4969,
+      "step": 15245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.958946077532207e-05,
+      "loss": 0.4386,
+      "step": 15246
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.957851910249181e-05,
+      "loss": 0.5729,
+      "step": 15247
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.956757755579618e-05,
+      "loss": 0.5654,
+      "step": 15248
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.955663613536766e-05,
+      "loss": 0.5249,
+      "step": 15249
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.954569484133864e-05,
+      "loss": 0.5159,
+      "step": 15250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.953475367384156e-05,
+      "loss": 0.4991,
+      "step": 15251
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.952381263300885e-05,
+      "loss": 0.5895,
+      "step": 15252
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.951287171897295e-05,
+      "loss": 0.645,
+      "step": 15253
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.950193093186622e-05,
+      "loss": 0.4969,
+      "step": 15254
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.949099027182115e-05,
+      "loss": 0.6093,
+      "step": 15255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.948004973897011e-05,
+      "loss": 0.5565,
+      "step": 15256
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.946910933344554e-05,
+      "loss": 0.4413,
+      "step": 15257
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.945816905537987e-05,
+      "loss": 0.5291,
+      "step": 15258
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.944722890490547e-05,
+      "loss": 0.4985,
+      "step": 15259
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.943628888215481e-05,
+      "loss": 0.5374,
+      "step": 15260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.942534898726022e-05,
+      "loss": 0.5348,
+      "step": 15261
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.94144092203542e-05,
+      "loss": 0.5276,
+      "step": 15262
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.940346958156912e-05,
+      "loss": 0.5526,
+      "step": 15263
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.939253007103734e-05,
+      "loss": 0.5554,
+      "step": 15264
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.938159068889133e-05,
+      "loss": 0.603,
+      "step": 15265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.937065143526347e-05,
+      "loss": 0.5224,
+      "step": 15266
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.935971231028615e-05,
+      "loss": 0.5747,
+      "step": 15267
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.93487733140918e-05,
+      "loss": 0.5803,
+      "step": 15268
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.933783444681276e-05,
+      "loss": 0.4411,
+      "step": 15269
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.932689570858149e-05,
+      "loss": 0.4977,
+      "step": 15270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.931595709953038e-05,
+      "loss": 0.5822,
+      "step": 15271
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.930501861979176e-05,
+      "loss": 0.558,
+      "step": 15272
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.929408026949811e-05,
+      "loss": 0.5088,
+      "step": 15273
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.928314204878173e-05,
+      "loss": 0.6166,
+      "step": 15274
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.927220395777509e-05,
+      "loss": 0.5197,
+      "step": 15275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.926126599661055e-05,
+      "loss": 0.484,
+      "step": 15276
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.925032816542044e-05,
+      "loss": 0.5769,
+      "step": 15277
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.923939046433725e-05,
+      "loss": 0.4688,
+      "step": 15278
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.922845289349325e-05,
+      "loss": 0.6142,
+      "step": 15279
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.921751545302092e-05,
+      "loss": 0.4754,
+      "step": 15280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.920657814305258e-05,
+      "loss": 0.4614,
+      "step": 15281
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.919564096372062e-05,
+      "loss": 0.5754,
+      "step": 15282
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.918470391515742e-05,
+      "loss": 0.4881,
+      "step": 15283
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.917376699749539e-05,
+      "loss": 0.5494,
+      "step": 15284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.916283021086682e-05,
+      "loss": 0.6074,
+      "step": 15285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.915189355540415e-05,
+      "loss": 0.5378,
+      "step": 15286
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.914095703123973e-05,
+      "loss": 0.5041,
+      "step": 15287
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.913002063850593e-05,
+      "loss": 0.4864,
+      "step": 15288
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.911908437733511e-05,
+      "loss": 0.5537,
+      "step": 15289
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.910814824785964e-05,
+      "loss": 0.518,
+      "step": 15290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.909721225021188e-05,
+      "loss": 0.5341,
+      "step": 15291
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.908627638452419e-05,
+      "loss": 0.5406,
+      "step": 15292
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.907534065092896e-05,
+      "loss": 0.5492,
+      "step": 15293
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.906440504955851e-05,
+      "loss": 0.4778,
+      "step": 15294
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.90534695805452e-05,
+      "loss": 0.431,
+      "step": 15295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.904253424402142e-05,
+      "loss": 0.5027,
+      "step": 15296
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.903159904011948e-05,
+      "loss": 0.577,
+      "step": 15297
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.902066396897177e-05,
+      "loss": 0.6079,
+      "step": 15298
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.900972903071062e-05,
+      "loss": 0.5032,
+      "step": 15299
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.899879422546838e-05,
+      "loss": 0.536,
+      "step": 15300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.898785955337743e-05,
+      "loss": 0.484,
+      "step": 15301
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.897692501457006e-05,
+      "loss": 0.4986,
+      "step": 15302
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.896599060917864e-05,
+      "loss": 0.5536,
+      "step": 15303
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.895505633733554e-05,
+      "loss": 0.6207,
+      "step": 15304
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.894412219917304e-05,
+      "loss": 0.5419,
+      "step": 15305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.893318819482355e-05,
+      "loss": 0.4713,
+      "step": 15306
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.892225432441937e-05,
+      "loss": 0.5318,
+      "step": 15307
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.89113205880928e-05,
+      "loss": 0.5531,
+      "step": 15308
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.890038698597628e-05,
+      "loss": 0.6339,
+      "step": 15309
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.888945351820205e-05,
+      "loss": 0.6809,
+      "step": 15310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.887852018490246e-05,
+      "loss": 0.4982,
+      "step": 15311
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.886758698620984e-05,
+      "loss": 0.53,
+      "step": 15312
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.885665392225654e-05,
+      "loss": 0.5356,
+      "step": 15313
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.884572099317485e-05,
+      "loss": 0.6849,
+      "step": 15314
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.883478819909715e-05,
+      "loss": 0.5179,
+      "step": 15315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.882385554015571e-05,
+      "loss": 0.5444,
+      "step": 15316
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.881292301648287e-05,
+      "loss": 0.4961,
+      "step": 15317
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.880199062821098e-05,
+      "loss": 0.5776,
+      "step": 15318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.879105837547233e-05,
+      "loss": 0.4674,
+      "step": 15319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.87801262583992e-05,
+      "loss": 0.4611,
+      "step": 15320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.876919427712397e-05,
+      "loss": 0.6104,
+      "step": 15321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.875826243177891e-05,
+      "loss": 0.5181,
+      "step": 15322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.874733072249637e-05,
+      "loss": 0.5338,
+      "step": 15323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.873639914940865e-05,
+      "loss": 0.5967,
+      "step": 15324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.8725467712648e-05,
+      "loss": 0.5069,
+      "step": 15325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.871453641234681e-05,
+      "loss": 0.4952,
+      "step": 15326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.870360524863734e-05,
+      "loss": 0.4897,
+      "step": 15327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.86926742216519e-05,
+      "loss": 0.5515,
+      "step": 15328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.86817433315228e-05,
+      "loss": 0.6932,
+      "step": 15329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.867081257838232e-05,
+      "loss": 0.5731,
+      "step": 15330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.86598819623628e-05,
+      "loss": 0.4854,
+      "step": 15331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.864895148359647e-05,
+      "loss": 0.5303,
+      "step": 15332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.86380211422157e-05,
+      "loss": 0.4892,
+      "step": 15333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.862709093835274e-05,
+      "loss": 0.4944,
+      "step": 15334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.861616087213988e-05,
+      "loss": 0.5733,
+      "step": 15335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.860523094370944e-05,
+      "loss": 0.546,
+      "step": 15336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.859430115319368e-05,
+      "loss": 0.5057,
+      "step": 15337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.858337150072489e-05,
+      "loss": 0.577,
+      "step": 15338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.857244198643537e-05,
+      "loss": 0.5329,
+      "step": 15339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.856151261045737e-05,
+      "loss": 0.5045,
+      "step": 15340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.855058337292324e-05,
+      "loss": 0.5547,
+      "step": 15341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.85396542739652e-05,
+      "loss": 0.5674,
+      "step": 15342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.852872531371552e-05,
+      "loss": 0.5365,
+      "step": 15343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.851779649230654e-05,
+      "loss": 0.5046,
+      "step": 15344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.850686780987047e-05,
+      "loss": 0.5021,
+      "step": 15345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.849593926653963e-05,
+      "loss": 0.573,
+      "step": 15346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.848501086244627e-05,
+      "loss": 0.5689,
+      "step": 15347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.847408259772265e-05,
+      "loss": 0.5028,
+      "step": 15348
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.846315447250106e-05,
+      "loss": 0.5226,
+      "step": 15349
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.845222648691376e-05,
+      "loss": 0.537,
+      "step": 15350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.844129864109302e-05,
+      "loss": 0.4681,
+      "step": 15351
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.84303709351711e-05,
+      "loss": 0.5033,
+      "step": 15352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.841944336928024e-05,
+      "loss": 0.5624,
+      "step": 15353
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.840851594355273e-05,
+      "loss": 0.6204,
+      "step": 15354
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.839758865812081e-05,
+      "loss": 0.5444,
+      "step": 15355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.838666151311674e-05,
+      "loss": 0.5843,
+      "step": 15356
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.837573450867279e-05,
+      "loss": 0.4802,
+      "step": 15357
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.836480764492119e-05,
+      "loss": 0.5904,
+      "step": 15358
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.835388092199422e-05,
+      "loss": 0.5732,
+      "step": 15359
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.83429543400241e-05,
+      "loss": 0.5716,
+      "step": 15360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.833202789914307e-05,
+      "loss": 0.4208,
+      "step": 15361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.832110159948343e-05,
+      "loss": 0.6005,
+      "step": 15362
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.831017544117737e-05,
+      "loss": 0.5362,
+      "step": 15363
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.829924942435717e-05,
+      "loss": 0.5944,
+      "step": 15364
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.828832354915507e-05,
+      "loss": 0.5325,
+      "step": 15365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.827739781570327e-05,
+      "loss": 0.5147,
+      "step": 15366
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.826647222413405e-05,
+      "loss": 0.463,
+      "step": 15367
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.825554677457962e-05,
+      "loss": 0.4754,
+      "step": 15368
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.82446214671722e-05,
+      "loss": 0.5264,
+      "step": 15369
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.823369630204409e-05,
+      "loss": 0.6132,
+      "step": 15370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.822277127932743e-05,
+      "loss": 0.5238,
+      "step": 15371
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.821184639915454e-05,
+      "loss": 0.6267,
+      "step": 15372
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.820092166165759e-05,
+      "loss": 0.4807,
+      "step": 15373
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.81899970669688e-05,
+      "loss": 0.6018,
+      "step": 15374
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.817907261522045e-05,
+      "loss": 0.5819,
+      "step": 15375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.816814830654468e-05,
+      "loss": 0.5812,
+      "step": 15376
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.815722414107379e-05,
+      "loss": 0.5507,
+      "step": 15377
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.814630011893997e-05,
+      "loss": 0.517,
+      "step": 15378
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.813537624027541e-05,
+      "loss": 0.5076,
+      "step": 15379
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.812445250521236e-05,
+      "loss": 0.6041,
+      "step": 15380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.8113528913883e-05,
+      "loss": 0.441,
+      "step": 15381
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.81026054664196e-05,
+      "loss": 0.5532,
+      "step": 15382
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.809168216295432e-05,
+      "loss": 0.5115,
+      "step": 15383
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.808075900361936e-05,
+      "loss": 0.4353,
+      "step": 15384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.806983598854697e-05,
+      "loss": 0.4847,
+      "step": 15385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.805891311786934e-05,
+      "loss": 0.5739,
+      "step": 15386
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.804799039171863e-05,
+      "loss": 0.5777,
+      "step": 15387
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.803706781022712e-05,
+      "loss": 0.5655,
+      "step": 15388
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.802614537352692e-05,
+      "loss": 0.515,
+      "step": 15389
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.801522308175031e-05,
+      "loss": 0.5002,
+      "step": 15390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.800430093502947e-05,
+      "loss": 0.5936,
+      "step": 15391
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.799337893349653e-05,
+      "loss": 0.5517,
+      "step": 15392
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.798245707728376e-05,
+      "loss": 0.5234,
+      "step": 15393
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.797153536652329e-05,
+      "loss": 0.4231,
+      "step": 15394
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.796061380134737e-05,
+      "loss": 0.591,
+      "step": 15395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.794969238188815e-05,
+      "loss": 0.5775,
+      "step": 15396
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.79387711082778e-05,
+      "loss": 0.6013,
+      "step": 15397
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.792784998064854e-05,
+      "loss": 0.5771,
+      "step": 15398
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.791692899913253e-05,
+      "loss": 0.5725,
+      "step": 15399
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.790600816386197e-05,
+      "loss": 0.5519,
+      "step": 15400
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 0.5259846448898315,
+      "eval_runtime": 849.9561,
+      "eval_samples_per_second": 28.307,
+      "eval_steps_per_second": 14.154,
+      "step": 15400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.789508747496903e-05,
+      "loss": 0.5868,
+      "step": 15401
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.788416693258585e-05,
+      "loss": 0.5125,
+      "step": 15402
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.787324653684468e-05,
+      "loss": 0.5297,
+      "step": 15403
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.786232628787762e-05,
+      "loss": 0.5566,
+      "step": 15404
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.785140618581688e-05,
+      "loss": 0.4896,
+      "step": 15405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.784048623079464e-05,
+      "loss": 0.5664,
+      "step": 15406
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.782956642294302e-05,
+      "loss": 0.585,
+      "step": 15407
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.781864676239424e-05,
+      "loss": 0.4519,
+      "step": 15408
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.780772724928043e-05,
+      "loss": 0.4948,
+      "step": 15409
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.779680788373376e-05,
+      "loss": 0.5096,
+      "step": 15410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.778588866588639e-05,
+      "loss": 0.58,
+      "step": 15411
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.777496959587049e-05,
+      "loss": 0.5392,
+      "step": 15412
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.77640506738182e-05,
+      "loss": 0.4817,
+      "step": 15413
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.77531318998617e-05,
+      "loss": 0.4229,
+      "step": 15414
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.774221327413312e-05,
+      "loss": 0.5633,
+      "step": 15415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.773129479676463e-05,
+      "loss": 0.5079,
+      "step": 15416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.772037646788836e-05,
+      "loss": 0.5004,
+      "step": 15417
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.770945828763646e-05,
+      "loss": 0.567,
+      "step": 15418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.76985402561411e-05,
+      "loss": 0.5075,
+      "step": 15419
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.76876223735344e-05,
+      "loss": 0.5988,
+      "step": 15420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.767670463994851e-05,
+      "loss": 0.538,
+      "step": 15421
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.766578705551559e-05,
+      "loss": 0.5109,
+      "step": 15422
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.765486962036774e-05,
+      "loss": 0.5615,
+      "step": 15423
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.764395233463714e-05,
+      "loss": 0.4861,
+      "step": 15424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.763303519845588e-05,
+      "loss": 0.5225,
+      "step": 15425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.762211821195614e-05,
+      "loss": 0.5908,
+      "step": 15426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.761120137527004e-05,
+      "loss": 0.7036,
+      "step": 15427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.760028468852967e-05,
+      "loss": 0.5886,
+      "step": 15428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.75893681518672e-05,
+      "loss": 0.5076,
+      "step": 15429
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.757845176541474e-05,
+      "loss": 0.5508,
+      "step": 15430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.756753552930445e-05,
+      "loss": 0.6401,
+      "step": 15431
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.75566194436684e-05,
+      "loss": 0.655,
+      "step": 15432
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.754570350863873e-05,
+      "loss": 0.5768,
+      "step": 15433
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.753478772434758e-05,
+      "loss": 0.6621,
+      "step": 15434
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.752387209092706e-05,
+      "loss": 0.6664,
+      "step": 15435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.751295660850923e-05,
+      "loss": 0.595,
+      "step": 15436
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.750204127722628e-05,
+      "loss": 0.4242,
+      "step": 15437
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.749112609721029e-05,
+      "loss": 0.5703,
+      "step": 15438
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.748021106859337e-05,
+      "loss": 0.5148,
+      "step": 15439
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.746929619150763e-05,
+      "loss": 0.5909,
+      "step": 15440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.745838146608518e-05,
+      "loss": 0.5953,
+      "step": 15441
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.744746689245812e-05,
+      "loss": 0.6138,
+      "step": 15442
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.743655247075858e-05,
+      "loss": 0.4775,
+      "step": 15443
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.742563820111862e-05,
+      "loss": 0.6006,
+      "step": 15444
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.741472408367033e-05,
+      "loss": 0.5509,
+      "step": 15445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.740381011854584e-05,
+      "loss": 0.5084,
+      "step": 15446
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.739289630587723e-05,
+      "loss": 0.5458,
+      "step": 15447
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.738198264579662e-05,
+      "loss": 0.5664,
+      "step": 15448
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.737106913843608e-05,
+      "loss": 0.4286,
+      "step": 15449
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.736015578392767e-05,
+      "loss": 0.5492,
+      "step": 15450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.734924258240354e-05,
+      "loss": 0.5344,
+      "step": 15451
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.733832953399572e-05,
+      "loss": 0.4574,
+      "step": 15452
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.732741663883635e-05,
+      "loss": 0.6067,
+      "step": 15453
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.731650389705748e-05,
+      "loss": 0.5312,
+      "step": 15454
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.730559130879116e-05,
+      "loss": 0.7229,
+      "step": 15455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.729467887416953e-05,
+      "loss": 0.4592,
+      "step": 15456
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.728376659332465e-05,
+      "loss": 0.5054,
+      "step": 15457
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.727285446638854e-05,
+      "loss": 0.5414,
+      "step": 15458
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.726194249349336e-05,
+      "loss": 0.5617,
+      "step": 15459
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.72510306747711e-05,
+      "loss": 0.6728,
+      "step": 15460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.724011901035388e-05,
+      "loss": 0.6544,
+      "step": 15461
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.722920750037378e-05,
+      "loss": 0.5459,
+      "step": 15462
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.721829614496279e-05,
+      "loss": 0.5301,
+      "step": 15463
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.720738494425306e-05,
+      "loss": 0.5148,
+      "step": 15464
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.719647389837659e-05,
+      "loss": 0.4811,
+      "step": 15465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.71855630074655e-05,
+      "loss": 0.647,
+      "step": 15466
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.71746522716518e-05,
+      "loss": 0.4922,
+      "step": 15467
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.716374169106756e-05,
+      "loss": 0.5329,
+      "step": 15468
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.715283126584484e-05,
+      "loss": 0.4962,
+      "step": 15469
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.714192099611567e-05,
+      "loss": 0.468,
+      "step": 15470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.713101088201212e-05,
+      "loss": 0.5253,
+      "step": 15471
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.712010092366625e-05,
+      "loss": 0.4837,
+      "step": 15472
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.710919112121008e-05,
+      "loss": 0.5407,
+      "step": 15473
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.709828147477568e-05,
+      "loss": 0.4844,
+      "step": 15474
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.708737198449507e-05,
+      "loss": 0.5621,
+      "step": 15475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.707646265050031e-05,
+      "loss": 0.5329,
+      "step": 15476
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.706555347292344e-05,
+      "loss": 0.5587,
+      "step": 15477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.705464445189647e-05,
+      "loss": 0.5752,
+      "step": 15478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.704373558755148e-05,
+      "loss": 0.5934,
+      "step": 15479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.703282688002046e-05,
+      "loss": 0.6745,
+      "step": 15480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.702191832943545e-05,
+      "loss": 0.5616,
+      "step": 15481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.701100993592852e-05,
+      "loss": 0.5998,
+      "step": 15482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.700010169963163e-05,
+      "loss": 0.5222,
+      "step": 15483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.698919362067688e-05,
+      "loss": 0.593,
+      "step": 15484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.697828569919625e-05,
+      "loss": 0.541,
+      "step": 15485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.696737793532175e-05,
+      "loss": 0.5476,
+      "step": 15486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.695647032918544e-05,
+      "loss": 0.5527,
+      "step": 15487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.694556288091933e-05,
+      "loss": 0.5868,
+      "step": 15488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.693465559065539e-05,
+      "loss": 0.6218,
+      "step": 15489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.692374845852569e-05,
+      "loss": 0.4097,
+      "step": 15490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.691284148466221e-05,
+      "loss": 0.6584,
+      "step": 15491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.690193466919699e-05,
+      "loss": 0.557,
+      "step": 15492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.689102801226203e-05,
+      "loss": 0.5754,
+      "step": 15493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.68801215139893e-05,
+      "loss": 0.5245,
+      "step": 15494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.686921517451087e-05,
+      "loss": 0.5386,
+      "step": 15495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.685830899395868e-05,
+      "loss": 0.5673,
+      "step": 15496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.684740297246477e-05,
+      "loss": 0.4916,
+      "step": 15497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.683649711016113e-05,
+      "loss": 0.4603,
+      "step": 15498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.682559140717974e-05,
+      "loss": 0.5868,
+      "step": 15499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.681468586365263e-05,
+      "loss": 0.5304,
+      "step": 15500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.680378047971173e-05,
+      "loss": 0.5378,
+      "step": 15501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.679287525548912e-05,
+      "loss": 0.4945,
+      "step": 15502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.678197019111673e-05,
+      "loss": 0.561,
+      "step": 15503
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.677106528672656e-05,
+      "loss": 0.5628,
+      "step": 15504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.676016054245062e-05,
+      "loss": 0.5416,
+      "step": 15505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.674925595842085e-05,
+      "loss": 0.499,
+      "step": 15506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.673835153476925e-05,
+      "loss": 0.5083,
+      "step": 15507
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.672744727162781e-05,
+      "loss": 0.5337,
+      "step": 15508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.671654316912849e-05,
+      "loss": 0.5517,
+      "step": 15509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.670563922740331e-05,
+      "loss": 0.5293,
+      "step": 15510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.66947354465842e-05,
+      "loss": 0.47,
+      "step": 15511
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.668383182680311e-05,
+      "loss": 0.5789,
+      "step": 15512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.667292836819208e-05,
+      "loss": 0.6644,
+      "step": 15513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.666202507088302e-05,
+      "loss": 0.4383,
+      "step": 15514
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.665112193500794e-05,
+      "loss": 0.5704,
+      "step": 15515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.664021896069879e-05,
+      "loss": 0.4753,
+      "step": 15516
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.662931614808749e-05,
+      "loss": 0.5996,
+      "step": 15517
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.661841349730605e-05,
+      "loss": 0.5263,
+      "step": 15518
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.660751100848641e-05,
+      "loss": 0.4824,
+      "step": 15519
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.659660868176053e-05,
+      "loss": 0.583,
+      "step": 15520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.658570651726038e-05,
+      "loss": 0.5218,
+      "step": 15521
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.657480451511788e-05,
+      "loss": 0.5772,
+      "step": 15522
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.6563902675465e-05,
+      "loss": 0.4709,
+      "step": 15523
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.655300099843371e-05,
+      "loss": 0.4887,
+      "step": 15524
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.65420994841559e-05,
+      "loss": 0.5534,
+      "step": 15525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.653119813276356e-05,
+      "loss": 0.4993,
+      "step": 15526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.652029694438861e-05,
+      "loss": 0.4869,
+      "step": 15527
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.650939591916302e-05,
+      "loss": 0.5242,
+      "step": 15528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.649849505721871e-05,
+      "loss": 0.5652,
+      "step": 15529
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.648759435868758e-05,
+      "loss": 0.4282,
+      "step": 15530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.647669382370165e-05,
+      "loss": 0.532,
+      "step": 15531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.646579345239276e-05,
+      "loss": 0.4883,
+      "step": 15532
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.645489324489292e-05,
+      "loss": 0.5437,
+      "step": 15533
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.644399320133401e-05,
+      "loss": 0.5667,
+      "step": 15534
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.643309332184797e-05,
+      "loss": 0.5424,
+      "step": 15535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.642219360656674e-05,
+      "loss": 0.5817,
+      "step": 15536
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.641129405562221e-05,
+      "loss": 0.5511,
+      "step": 15537
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.640039466914631e-05,
+      "loss": 0.6682,
+      "step": 15538
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.6389495447271e-05,
+      "loss": 0.5159,
+      "step": 15539
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.637859639012812e-05,
+      "loss": 0.5554,
+      "step": 15540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.636769749784967e-05,
+      "loss": 0.5028,
+      "step": 15541
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.635679877056752e-05,
+      "loss": 0.4371,
+      "step": 15542
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.634590020841356e-05,
+      "loss": 0.578,
+      "step": 15543
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.633500181151974e-05,
+      "loss": 0.5183,
+      "step": 15544
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.632410358001794e-05,
+      "loss": 0.4969,
+      "step": 15545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.63132055140401e-05,
+      "loss": 0.4981,
+      "step": 15546
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.630230761371808e-05,
+      "loss": 0.4894,
+      "step": 15547
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.629140987918378e-05,
+      "loss": 0.5275,
+      "step": 15548
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.628051231056915e-05,
+      "loss": 0.5434,
+      "step": 15549
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.626961490800602e-05,
+      "loss": 0.53,
+      "step": 15550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.625871767162634e-05,
+      "loss": 0.5185,
+      "step": 15551
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.6247820601562e-05,
+      "loss": 0.5453,
+      "step": 15552
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.623692369794484e-05,
+      "loss": 0.5718,
+      "step": 15553
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.62260269609068e-05,
+      "loss": 0.6139,
+      "step": 15554
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.621513039057975e-05,
+      "loss": 0.518,
+      "step": 15555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.620423398709554e-05,
+      "loss": 0.4896,
+      "step": 15556
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.619333775058613e-05,
+      "loss": 0.5271,
+      "step": 15557
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.618244168118332e-05,
+      "loss": 0.503,
+      "step": 15558
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.617154577901906e-05,
+      "loss": 0.485,
+      "step": 15559
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.616065004422518e-05,
+      "loss": 0.5042,
+      "step": 15560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.614975447693357e-05,
+      "loss": 0.5731,
+      "step": 15561
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.613885907727611e-05,
+      "loss": 0.5713,
+      "step": 15562
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.612796384538462e-05,
+      "loss": 0.5331,
+      "step": 15563
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.611706878139104e-05,
+      "loss": 0.4693,
+      "step": 15564
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.610617388542721e-05,
+      "loss": 0.5745,
+      "step": 15565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.609527915762496e-05,
+      "loss": 0.6089,
+      "step": 15566
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.608438459811622e-05,
+      "loss": 0.5104,
+      "step": 15567
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.607349020703278e-05,
+      "loss": 0.5611,
+      "step": 15568
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.606259598450657e-05,
+      "loss": 0.5041,
+      "step": 15569
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.60517019306694e-05,
+      "loss": 0.5036,
+      "step": 15570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.60408080456531e-05,
+      "loss": 0.6352,
+      "step": 15571
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.602991432958958e-05,
+      "loss": 0.4454,
+      "step": 15572
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.601902078261068e-05,
+      "loss": 0.6043,
+      "step": 15573
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.600812740484818e-05,
+      "loss": 0.5716,
+      "step": 15574
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.599723419643403e-05,
+      "loss": 0.518,
+      "step": 15575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.59863411575e-05,
+      "loss": 0.5658,
+      "step": 15576
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.597544828817796e-05,
+      "loss": 0.5837,
+      "step": 15577
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.596455558859978e-05,
+      "loss": 0.5709,
+      "step": 15578
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.595366305889725e-05,
+      "loss": 0.6029,
+      "step": 15579
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.594277069920218e-05,
+      "loss": 0.4866,
+      "step": 15580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.593187850964647e-05,
+      "loss": 0.5234,
+      "step": 15581
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.592098649036194e-05,
+      "loss": 0.4937,
+      "step": 15582
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.591009464148037e-05,
+      "loss": 0.5116,
+      "step": 15583
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.589920296313366e-05,
+      "loss": 0.517,
+      "step": 15584
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.588831145545355e-05,
+      "loss": 0.5405,
+      "step": 15585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.587742011857195e-05,
+      "loss": 0.5662,
+      "step": 15586
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.586652895262064e-05,
+      "loss": 0.5511,
+      "step": 15587
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.585563795773143e-05,
+      "loss": 0.5555,
+      "step": 15588
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.584474713403616e-05,
+      "loss": 0.3928,
+      "step": 15589
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.583385648166664e-05,
+      "loss": 0.5427,
+      "step": 15590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.582296600075465e-05,
+      "loss": 0.5473,
+      "step": 15591
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.581207569143206e-05,
+      "loss": 0.5914,
+      "step": 15592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.580118555383063e-05,
+      "loss": 0.4571,
+      "step": 15593
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.57902955880822e-05,
+      "loss": 0.4761,
+      "step": 15594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.577940579431856e-05,
+      "loss": 0.6003,
+      "step": 15595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.57685161726715e-05,
+      "loss": 0.4654,
+      "step": 15596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.575762672327286e-05,
+      "loss": 0.5487,
+      "step": 15597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.574673744625437e-05,
+      "loss": 0.5056,
+      "step": 15598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.573584834174792e-05,
+      "loss": 0.6108,
+      "step": 15599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.572495940988523e-05,
+      "loss": 0.4209,
+      "step": 15600
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 0.5225046277046204,
+      "eval_runtime": 848.7105,
+      "eval_samples_per_second": 28.349,
+      "eval_steps_per_second": 14.174,
+      "step": 15600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.57140706507981e-05,
+      "loss": 0.5025,
+      "step": 15601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.570318206461836e-05,
+      "loss": 0.5855,
+      "step": 15602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.569229365147776e-05,
+      "loss": 0.5214,
+      "step": 15603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.568140541150812e-05,
+      "loss": 0.4443,
+      "step": 15604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.56705173448412e-05,
+      "loss": 0.532,
+      "step": 15605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.565962945160877e-05,
+      "loss": 0.549,
+      "step": 15606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.564874173194263e-05,
+      "loss": 0.5546,
+      "step": 15607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.563785418597458e-05,
+      "loss": 0.4596,
+      "step": 15608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.562696681383633e-05,
+      "loss": 0.5351,
+      "step": 15609
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.561607961565973e-05,
+      "loss": 0.4731,
+      "step": 15610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.56051925915765e-05,
+      "loss": 0.4791,
+      "step": 15611
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.559430574171842e-05,
+      "loss": 0.524,
+      "step": 15612
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.558341906621726e-05,
+      "loss": 0.509,
+      "step": 15613
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.557253256520479e-05,
+      "loss": 0.6594,
+      "step": 15614
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.556164623881278e-05,
+      "loss": 0.4516,
+      "step": 15615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.555076008717295e-05,
+      "loss": 0.6282,
+      "step": 15616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.55398741104171e-05,
+      "loss": 0.6831,
+      "step": 15617
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.5528988308677e-05,
+      "loss": 0.5275,
+      "step": 15618
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.551810268208436e-05,
+      "loss": 0.6217,
+      "step": 15619
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.550721723077095e-05,
+      "loss": 0.6552,
+      "step": 15620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.549633195486855e-05,
+      "loss": 0.5767,
+      "step": 15621
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.548544685450884e-05,
+      "loss": 0.4618,
+      "step": 15622
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.547456192982364e-05,
+      "loss": 0.5557,
+      "step": 15623
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.546367718094462e-05,
+      "loss": 0.4147,
+      "step": 15624
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.54527926080036e-05,
+      "loss": 0.5455,
+      "step": 15625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.544190821113227e-05,
+      "loss": 0.5437,
+      "step": 15626
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.543102399046237e-05,
+      "loss": 0.4666,
+      "step": 15627
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.542013994612567e-05,
+      "loss": 0.6055,
+      "step": 15628
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.540925607825384e-05,
+      "loss": 0.5071,
+      "step": 15629
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.539837238697867e-05,
+      "loss": 0.5246,
+      "step": 15630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.538748887243189e-05,
+      "loss": 0.497,
+      "step": 15631
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.537660553474517e-05,
+      "loss": 0.548,
+      "step": 15632
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.53657223740503e-05,
+      "loss": 0.6261,
+      "step": 15633
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.535483939047893e-05,
+      "loss": 0.4937,
+      "step": 15634
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.534395658416286e-05,
+      "loss": 0.5815,
+      "step": 15635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.533307395523376e-05,
+      "loss": 0.4875,
+      "step": 15636
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.532219150382335e-05,
+      "loss": 0.4865,
+      "step": 15637
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.531130923006335e-05,
+      "loss": 0.5552,
+      "step": 15638
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.530042713408549e-05,
+      "loss": 0.5489,
+      "step": 15639
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.528954521602144e-05,
+      "loss": 0.4988,
+      "step": 15640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.527866347600294e-05,
+      "loss": 0.4212,
+      "step": 15641
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.526778191416167e-05,
+      "loss": 0.5154,
+      "step": 15642
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.525690053062937e-05,
+      "loss": 0.5101,
+      "step": 15643
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.524601932553771e-05,
+      "loss": 0.5004,
+      "step": 15644
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.523513829901839e-05,
+      "loss": 0.5058,
+      "step": 15645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.522425745120312e-05,
+      "loss": 0.6161,
+      "step": 15646
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.521337678222358e-05,
+      "loss": 0.5691,
+      "step": 15647
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.52024962922115e-05,
+      "loss": 0.5631,
+      "step": 15648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.519161598129854e-05,
+      "loss": 0.448,
+      "step": 15649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.518073584961636e-05,
+      "loss": 0.4674,
+      "step": 15650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.516985589729669e-05,
+      "loss": 0.5622,
+      "step": 15651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.51589761244712e-05,
+      "loss": 0.5574,
+      "step": 15652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.514809653127156e-05,
+      "loss": 0.4409,
+      "step": 15653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.513721711782949e-05,
+      "loss": 0.5859,
+      "step": 15654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.51263378842766e-05,
+      "loss": 0.5232,
+      "step": 15655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.511545883074463e-05,
+      "loss": 0.5384,
+      "step": 15656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.510457995736522e-05,
+      "loss": 0.5305,
+      "step": 15657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.509370126427002e-05,
+      "loss": 0.4121,
+      "step": 15658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.508282275159076e-05,
+      "loss": 0.5765,
+      "step": 15659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.507194441945903e-05,
+      "loss": 0.6302,
+      "step": 15660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.506106626800656e-05,
+      "loss": 0.5697,
+      "step": 15661
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.5050188297365e-05,
+      "loss": 0.6174,
+      "step": 15662
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.503931050766598e-05,
+      "loss": 0.5581,
+      "step": 15663
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.502843289904117e-05,
+      "loss": 0.5623,
+      "step": 15664
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.501755547162222e-05,
+      "loss": 0.5142,
+      "step": 15665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.500667822554081e-05,
+      "loss": 0.502,
+      "step": 15666
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.499580116092858e-05,
+      "loss": 0.5002,
+      "step": 15667
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.498492427791714e-05,
+      "loss": 0.4297,
+      "step": 15668
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.49740475766382e-05,
+      "loss": 0.6158,
+      "step": 15669
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.496317105722336e-05,
+      "loss": 0.5707,
+      "step": 15670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.495229471980428e-05,
+      "loss": 0.5401,
+      "step": 15671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.494141856451258e-05,
+      "loss": 0.5008,
+      "step": 15672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.493054259147993e-05,
+      "loss": 0.5133,
+      "step": 15673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.491966680083794e-05,
+      "loss": 0.5308,
+      "step": 15674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.490879119271826e-05,
+      "loss": 0.5334,
+      "step": 15675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.489791576725249e-05,
+      "loss": 0.6245,
+      "step": 15676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.488704052457231e-05,
+      "loss": 0.5653,
+      "step": 15677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.487616546480927e-05,
+      "loss": 0.4797,
+      "step": 15678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.48652905880951e-05,
+      "loss": 0.6053,
+      "step": 15679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.485441589456134e-05,
+      "loss": 0.5491,
+      "step": 15680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.484354138433964e-05,
+      "loss": 0.4225,
+      "step": 15681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.48326670575616e-05,
+      "loss": 0.5039,
+      "step": 15682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.482179291435884e-05,
+      "loss": 0.5213,
+      "step": 15683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.481091895486301e-05,
+      "loss": 0.4608,
+      "step": 15684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.480004517920569e-05,
+      "loss": 0.6023,
+      "step": 15685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.478917158751846e-05,
+      "loss": 0.5206,
+      "step": 15686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.4778298179933e-05,
+      "loss": 0.4065,
+      "step": 15687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.476742495658085e-05,
+      "loss": 0.5058,
+      "step": 15688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.475655191759363e-05,
+      "loss": 0.602,
+      "step": 15689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.474567906310295e-05,
+      "loss": 0.4239,
+      "step": 15690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.473480639324041e-05,
+      "loss": 0.5342,
+      "step": 15691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.47239339081376e-05,
+      "loss": 0.5056,
+      "step": 15692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.47130616079261e-05,
+      "loss": 0.5333,
+      "step": 15693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.470218949273752e-05,
+      "loss": 0.4739,
+      "step": 15694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.469131756270345e-05,
+      "loss": 0.4336,
+      "step": 15695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.468044581795544e-05,
+      "loss": 0.5875,
+      "step": 15696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.466957425862513e-05,
+      "loss": 0.5444,
+      "step": 15697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.465870288484407e-05,
+      "loss": 0.6394,
+      "step": 15698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.464783169674383e-05,
+      "loss": 0.5103,
+      "step": 15699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.463696069445601e-05,
+      "loss": 0.5699,
+      "step": 15700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.462608987811217e-05,
+      "loss": 0.5327,
+      "step": 15701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.461521924784391e-05,
+      "loss": 0.4887,
+      "step": 15702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.460434880378279e-05,
+      "loss": 0.648,
+      "step": 15703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.459347854606032e-05,
+      "loss": 0.5607,
+      "step": 15704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.458260847480816e-05,
+      "loss": 0.5573,
+      "step": 15705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.457173859015784e-05,
+      "loss": 0.502,
+      "step": 15706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.456086889224088e-05,
+      "loss": 0.5076,
+      "step": 15707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.454999938118888e-05,
+      "loss": 0.5823,
+      "step": 15708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.453913005713338e-05,
+      "loss": 0.5606,
+      "step": 15709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.452826092020597e-05,
+      "loss": 0.5049,
+      "step": 15710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.451739197053817e-05,
+      "loss": 0.4814,
+      "step": 15711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.450652320826151e-05,
+      "loss": 0.6657,
+      "step": 15712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.449565463350762e-05,
+      "loss": 0.4943,
+      "step": 15713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.448478624640797e-05,
+      "loss": 0.4716,
+      "step": 15714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.447391804709413e-05,
+      "loss": 0.4936,
+      "step": 15715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.446305003569761e-05,
+      "loss": 0.4706,
+      "step": 15716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.445218221235e-05,
+      "loss": 0.5679,
+      "step": 15717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.444131457718278e-05,
+      "loss": 0.5384,
+      "step": 15718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.443044713032754e-05,
+      "loss": 0.5734,
+      "step": 15719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.44195798719158e-05,
+      "loss": 0.5312,
+      "step": 15720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.440871280207906e-05,
+      "loss": 0.5195,
+      "step": 15721
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.439784592094889e-05,
+      "loss": 0.5115,
+      "step": 15722
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.43869792286568e-05,
+      "loss": 0.4901,
+      "step": 15723
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.437611272533425e-05,
+      "loss": 0.5294,
+      "step": 15724
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.436524641111288e-05,
+      "loss": 0.6447,
+      "step": 15725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.43543802861241e-05,
+      "loss": 0.5236,
+      "step": 15726
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.434351435049949e-05,
+      "loss": 0.4603,
+      "step": 15727
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.433264860437056e-05,
+      "loss": 0.5718,
+      "step": 15728
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.43217830478688e-05,
+      "loss": 0.4696,
+      "step": 15729
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.431091768112573e-05,
+      "loss": 0.4832,
+      "step": 15730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.430005250427284e-05,
+      "loss": 0.4837,
+      "step": 15731
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.428918751744166e-05,
+      "loss": 0.5957,
+      "step": 15732
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.427832272076371e-05,
+      "loss": 0.5472,
+      "step": 15733
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.42674581143704e-05,
+      "loss": 0.5178,
+      "step": 15734
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.425659369839333e-05,
+      "loss": 0.6207,
+      "step": 15735
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.424572947296396e-05,
+      "loss": 0.6004,
+      "step": 15736
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.423486543821378e-05,
+      "loss": 0.4744,
+      "step": 15737
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.422400159427427e-05,
+      "loss": 0.5057,
+      "step": 15738
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.421313794127692e-05,
+      "loss": 0.5058,
+      "step": 15739
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.420227447935323e-05,
+      "loss": 0.568,
+      "step": 15740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.419141120863471e-05,
+      "loss": 0.5826,
+      "step": 15741
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.418054812925277e-05,
+      "loss": 0.545,
+      "step": 15742
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.416968524133896e-05,
+      "loss": 0.5751,
+      "step": 15743
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.415882254502469e-05,
+      "loss": 0.5599,
+      "step": 15744
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.414796004044152e-05,
+      "loss": 0.5187,
+      "step": 15745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.413709772772087e-05,
+      "loss": 0.4717,
+      "step": 15746
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.412623560699417e-05,
+      "loss": 0.5727,
+      "step": 15747
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.411537367839297e-05,
+      "loss": 0.5412,
+      "step": 15748
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.41045119420487e-05,
+      "loss": 0.6433,
+      "step": 15749
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.409365039809281e-05,
+      "loss": 0.5199,
+      "step": 15750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.408278904665679e-05,
+      "loss": 0.5293,
+      "step": 15751
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.407192788787206e-05,
+      "loss": 0.5798,
+      "step": 15752
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.406106692187011e-05,
+      "loss": 0.5655,
+      "step": 15753
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.405020614878237e-05,
+      "loss": 0.4809,
+      "step": 15754
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.403934556874033e-05,
+      "loss": 0.5216,
+      "step": 15755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.402848518187539e-05,
+      "loss": 0.5848,
+      "step": 15756
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.401762498831902e-05,
+      "loss": 0.475,
+      "step": 15757
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.400676498820268e-05,
+      "loss": 0.5219,
+      "step": 15758
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.39959051816578e-05,
+      "loss": 0.4977,
+      "step": 15759
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.398504556881577e-05,
+      "loss": 0.474,
+      "step": 15760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.397418614980813e-05,
+      "loss": 0.498,
+      "step": 15761
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.396332692476621e-05,
+      "loss": 0.4753,
+      "step": 15762
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.395246789382155e-05,
+      "loss": 0.5176,
+      "step": 15763
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.394160905710549e-05,
+      "loss": 0.4831,
+      "step": 15764
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.393075041474947e-05,
+      "loss": 0.5226,
+      "step": 15765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.391989196688496e-05,
+      "loss": 0.5415,
+      "step": 15766
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.390903371364334e-05,
+      "loss": 0.5198,
+      "step": 15767
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.389817565515607e-05,
+      "loss": 0.6164,
+      "step": 15768
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.388731779155456e-05,
+      "loss": 0.5339,
+      "step": 15769
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.38764601229702e-05,
+      "loss": 0.4774,
+      "step": 15770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.386560264953441e-05,
+      "loss": 0.4843,
+      "step": 15771
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.385474537137863e-05,
+      "loss": 0.5637,
+      "step": 15772
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.384388828863423e-05,
+      "loss": 0.5227,
+      "step": 15773
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.383303140143266e-05,
+      "loss": 0.5418,
+      "step": 15774
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.382217470990527e-05,
+      "loss": 0.5395,
+      "step": 15775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.381131821418352e-05,
+      "loss": 0.5573,
+      "step": 15776
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.380046191439878e-05,
+      "loss": 0.565,
+      "step": 15777
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.378960581068244e-05,
+      "loss": 0.496,
+      "step": 15778
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.377874990316593e-05,
+      "loss": 0.4656,
+      "step": 15779
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.376789419198059e-05,
+      "loss": 0.4895,
+      "step": 15780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.375703867725787e-05,
+      "loss": 0.5043,
+      "step": 15781
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.374618335912912e-05,
+      "loss": 0.618,
+      "step": 15782
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.373532823772569e-05,
+      "loss": 0.5162,
+      "step": 15783
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.372447331317906e-05,
+      "loss": 0.4878,
+      "step": 15784
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.371361858562053e-05,
+      "loss": 0.5753,
+      "step": 15785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.370276405518152e-05,
+      "loss": 0.528,
+      "step": 15786
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.369190972199338e-05,
+      "loss": 0.6031,
+      "step": 15787
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.368105558618748e-05,
+      "loss": 0.5502,
+      "step": 15788
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.367020164789523e-05,
+      "loss": 0.582,
+      "step": 15789
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.365934790724798e-05,
+      "loss": 0.5056,
+      "step": 15790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.364849436437707e-05,
+      "loss": 0.5495,
+      "step": 15791
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.36376410194139e-05,
+      "loss": 0.5725,
+      "step": 15792
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.362678787248979e-05,
+      "loss": 0.5893,
+      "step": 15793
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.361593492373614e-05,
+      "loss": 0.5579,
+      "step": 15794
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.360508217328431e-05,
+      "loss": 0.5248,
+      "step": 15795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.359422962126562e-05,
+      "loss": 0.6301,
+      "step": 15796
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.358337726781145e-05,
+      "loss": 0.5023,
+      "step": 15797
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.357252511305313e-05,
+      "loss": 0.5618,
+      "step": 15798
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.356167315712203e-05,
+      "loss": 0.588,
+      "step": 15799
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.355082140014947e-05,
+      "loss": 0.4759,
+      "step": 15800
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 0.5172915458679199,
+      "eval_runtime": 851.8143,
+      "eval_samples_per_second": 28.246,
+      "eval_steps_per_second": 14.123,
+      "step": 15800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.35399698422668e-05,
+      "loss": 0.4652,
+      "step": 15801
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.352911848360538e-05,
+      "loss": 0.5096,
+      "step": 15802
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.35182673242965e-05,
+      "loss": 0.5054,
+      "step": 15803
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.350741636447154e-05,
+      "loss": 0.5111,
+      "step": 15804
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.349656560426185e-05,
+      "loss": 0.4939,
+      "step": 15805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.348571504379868e-05,
+      "loss": 0.5498,
+      "step": 15806
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.347486468321345e-05,
+      "loss": 0.5332,
+      "step": 15807
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.346401452263743e-05,
+      "loss": 0.5089,
+      "step": 15808
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.345316456220193e-05,
+      "loss": 0.521,
+      "step": 15809
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.344231480203834e-05,
+      "loss": 0.521,
+      "step": 15810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.34314652422779e-05,
+      "loss": 0.5629,
+      "step": 15811
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.342061588305198e-05,
+      "loss": 0.5064,
+      "step": 15812
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.340976672449188e-05,
+      "loss": 0.5273,
+      "step": 15813
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.339891776672888e-05,
+      "loss": 0.5314,
+      "step": 15814
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.338806900989434e-05,
+      "loss": 0.5873,
+      "step": 15815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.33772204541195e-05,
+      "loss": 0.5627,
+      "step": 15816
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.336637209953576e-05,
+      "loss": 0.4894,
+      "step": 15817
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.335552394627432e-05,
+      "loss": 0.4688,
+      "step": 15818
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.334467599446653e-05,
+      "loss": 0.4733,
+      "step": 15819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.333382824424371e-05,
+      "loss": 0.5244,
+      "step": 15820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.33229806957371e-05,
+      "loss": 0.4938,
+      "step": 15821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.331213334907801e-05,
+      "loss": 0.4849,
+      "step": 15822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.330128620439777e-05,
+      "loss": 0.5379,
+      "step": 15823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.329043926182758e-05,
+      "loss": 0.4461,
+      "step": 15824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.327959252149881e-05,
+      "loss": 0.5589,
+      "step": 15825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.326874598354271e-05,
+      "loss": 0.4444,
+      "step": 15826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.325789964809053e-05,
+      "loss": 0.561,
+      "step": 15827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.32470535152736e-05,
+      "loss": 0.5915,
+      "step": 15828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.323620758522314e-05,
+      "loss": 0.5517,
+      "step": 15829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.322536185807049e-05,
+      "loss": 0.5458,
+      "step": 15830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.321451633394685e-05,
+      "loss": 0.5829,
+      "step": 15831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.320367101298351e-05,
+      "loss": 0.5248,
+      "step": 15832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.319282589531176e-05,
+      "loss": 0.5288,
+      "step": 15833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.318198098106282e-05,
+      "loss": 0.5219,
+      "step": 15834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.317113627036799e-05,
+      "loss": 0.552,
+      "step": 15835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.31602917633585e-05,
+      "loss": 0.4695,
+      "step": 15836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.314944746016561e-05,
+      "loss": 0.5337,
+      "step": 15837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.313860336092061e-05,
+      "loss": 0.5006,
+      "step": 15838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.31277594657547e-05,
+      "loss": 0.468,
+      "step": 15839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.311691577479911e-05,
+      "loss": 0.61,
+      "step": 15840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.310607228818515e-05,
+      "loss": 0.4628,
+      "step": 15841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.309522900604402e-05,
+      "loss": 0.4589,
+      "step": 15842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.308438592850699e-05,
+      "loss": 0.5068,
+      "step": 15843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.307354305570524e-05,
+      "loss": 0.5414,
+      "step": 15844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.306270038777006e-05,
+      "loss": 0.5065,
+      "step": 15845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.305185792483267e-05,
+      "loss": 0.5836,
+      "step": 15846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.304101566702428e-05,
+      "loss": 0.4914,
+      "step": 15847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.30301736144762e-05,
+      "loss": 0.4495,
+      "step": 15848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.301933176731954e-05,
+      "loss": 0.542,
+      "step": 15849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.300849012568555e-05,
+      "loss": 0.5126,
+      "step": 15850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.299764868970547e-05,
+      "loss": 0.5872,
+      "step": 15851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.298680745951054e-05,
+      "loss": 0.5576,
+      "step": 15852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.297596643523196e-05,
+      "loss": 0.5308,
+      "step": 15853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.29651256170009e-05,
+      "loss": 0.4755,
+      "step": 15854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.295428500494863e-05,
+      "loss": 0.5765,
+      "step": 15855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.294344459920631e-05,
+      "loss": 0.5758,
+      "step": 15856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.293260439990519e-05,
+      "loss": 0.5762,
+      "step": 15857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.292176440717645e-05,
+      "loss": 0.5566,
+      "step": 15858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.291092462115127e-05,
+      "loss": 0.4517,
+      "step": 15859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.290008504196088e-05,
+      "loss": 0.5379,
+      "step": 15860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.288924566973648e-05,
+      "loss": 0.5867,
+      "step": 15861
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.28784065046092e-05,
+      "loss": 0.5131,
+      "step": 15862
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.286756754671032e-05,
+      "loss": 0.613,
+      "step": 15863
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.285672879617096e-05,
+      "loss": 0.5379,
+      "step": 15864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.284589025312234e-05,
+      "loss": 0.5106,
+      "step": 15865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.283505191769562e-05,
+      "loss": 0.6036,
+      "step": 15866
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.282421379002199e-05,
+      "loss": 0.5321,
+      "step": 15867
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.281337587023264e-05,
+      "loss": 0.5057,
+      "step": 15868
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.280253815845871e-05,
+      "loss": 0.5247,
+      "step": 15869
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.279170065483142e-05,
+      "loss": 0.5492,
+      "step": 15870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.27808633594819e-05,
+      "loss": 0.5576,
+      "step": 15871
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.277002627254133e-05,
+      "loss": 0.4657,
+      "step": 15872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.275918939414089e-05,
+      "loss": 0.4856,
+      "step": 15873
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.274835272441173e-05,
+      "loss": 0.556,
+      "step": 15874
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.273751626348501e-05,
+      "loss": 0.5596,
+      "step": 15875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.272668001149188e-05,
+      "loss": 0.5009,
+      "step": 15876
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.271584396856349e-05,
+      "loss": 0.5637,
+      "step": 15877
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.270500813483101e-05,
+      "loss": 0.4488,
+      "step": 15878
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.269417251042561e-05,
+      "loss": 0.5786,
+      "step": 15879
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.268333709547837e-05,
+      "loss": 0.5374,
+      "step": 15880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.267250189012051e-05,
+      "loss": 0.4858,
+      "step": 15881
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.266166689448312e-05,
+      "loss": 0.6158,
+      "step": 15882
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.265083210869737e-05,
+      "loss": 0.4905,
+      "step": 15883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.263999753289437e-05,
+      "loss": 0.5399,
+      "step": 15884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.262916316720526e-05,
+      "loss": 0.5293,
+      "step": 15885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.261832901176121e-05,
+      "loss": 0.4297,
+      "step": 15886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.260749506669329e-05,
+      "loss": 0.524,
+      "step": 15887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.259666133213267e-05,
+      "loss": 0.5836,
+      "step": 15888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.258582780821047e-05,
+      "loss": 0.4957,
+      "step": 15889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.257499449505778e-05,
+      "loss": 0.495,
+      "step": 15890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.256416139280577e-05,
+      "loss": 0.5162,
+      "step": 15891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.255332850158552e-05,
+      "loss": 0.5443,
+      "step": 15892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.254249582152813e-05,
+      "loss": 0.4401,
+      "step": 15893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.253166335276477e-05,
+      "loss": 0.4918,
+      "step": 15894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.252083109542648e-05,
+      "loss": 0.4539,
+      "step": 15895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.250999904964441e-05,
+      "loss": 0.6391,
+      "step": 15896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.249916721554967e-05,
+      "loss": 0.5294,
+      "step": 15897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.248833559327333e-05,
+      "loss": 0.4905,
+      "step": 15898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.247750418294651e-05,
+      "loss": 0.5107,
+      "step": 15899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.246667298470028e-05,
+      "loss": 0.5699,
+      "step": 15900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.245584199866579e-05,
+      "loss": 0.4525,
+      "step": 15901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.244501122497406e-05,
+      "loss": 0.5664,
+      "step": 15902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.243418066375622e-05,
+      "loss": 0.5215,
+      "step": 15903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.242335031514337e-05,
+      "loss": 0.4914,
+      "step": 15904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.241252017926653e-05,
+      "loss": 0.5723,
+      "step": 15905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.240169025625687e-05,
+      "loss": 0.439,
+      "step": 15906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.23908605462454e-05,
+      "loss": 0.4838,
+      "step": 15907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.238003104936322e-05,
+      "loss": 0.5963,
+      "step": 15908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.23692017657414e-05,
+      "loss": 0.5081,
+      "step": 15909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.235837269551101e-05,
+      "loss": 0.5609,
+      "step": 15910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.234754383880311e-05,
+      "loss": 0.5054,
+      "step": 15911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.233671519574878e-05,
+      "loss": 0.5153,
+      "step": 15912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.232588676647906e-05,
+      "loss": 0.5043,
+      "step": 15913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.231505855112506e-05,
+      "loss": 0.5788,
+      "step": 15914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.230423054981778e-05,
+      "loss": 0.5887,
+      "step": 15915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.229340276268829e-05,
+      "loss": 0.5544,
+      "step": 15916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.228257518986768e-05,
+      "loss": 0.6062,
+      "step": 15917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.227174783148693e-05,
+      "loss": 0.605,
+      "step": 15918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.226092068767717e-05,
+      "loss": 0.4959,
+      "step": 15919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.225009375856938e-05,
+      "loss": 0.5401,
+      "step": 15920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.223926704429462e-05,
+      "loss": 0.5945,
+      "step": 15921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.222844054498396e-05,
+      "loss": 0.6247,
+      "step": 15922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.221761426076838e-05,
+      "loss": 0.522,
+      "step": 15923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.220678819177894e-05,
+      "loss": 0.637,
+      "step": 15924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.21959623381467e-05,
+      "loss": 0.5724,
+      "step": 15925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.218513670000264e-05,
+      "loss": 0.4864,
+      "step": 15926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.217431127747784e-05,
+      "loss": 0.515,
+      "step": 15927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.216348607070329e-05,
+      "loss": 0.4815,
+      "step": 15928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.215266107980999e-05,
+      "loss": 0.5368,
+      "step": 15929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.2141836304929e-05,
+      "loss": 0.5032,
+      "step": 15930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.213101174619132e-05,
+      "loss": 0.5837,
+      "step": 15931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.212018740372796e-05,
+      "loss": 0.6413,
+      "step": 15932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.210936327766995e-05,
+      "loss": 0.5298,
+      "step": 15933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.209853936814825e-05,
+      "loss": 0.4478,
+      "step": 15934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.208771567529392e-05,
+      "loss": 0.4746,
+      "step": 15935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.207689219923793e-05,
+      "loss": 0.5872,
+      "step": 15936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.20660689401113e-05,
+      "loss": 0.6477,
+      "step": 15937
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.205524589804502e-05,
+      "loss": 0.4654,
+      "step": 15938
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.204442307317006e-05,
+      "loss": 0.5094,
+      "step": 15939
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.203360046561746e-05,
+      "loss": 0.5128,
+      "step": 15940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.20227780755182e-05,
+      "loss": 0.5243,
+      "step": 15941
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.201195590300321e-05,
+      "loss": 0.5262,
+      "step": 15942
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.200113394820354e-05,
+      "loss": 0.5093,
+      "step": 15943
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.199031221125012e-05,
+      "loss": 0.5774,
+      "step": 15944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.197949069227399e-05,
+      "loss": 0.5182,
+      "step": 15945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.196866939140609e-05,
+      "loss": 0.5723,
+      "step": 15946
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.195784830877736e-05,
+      "loss": 0.4718,
+      "step": 15947
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.194702744451886e-05,
+      "loss": 0.5131,
+      "step": 15948
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.193620679876146e-05,
+      "loss": 0.4433,
+      "step": 15949
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.192538637163621e-05,
+      "loss": 0.4968,
+      "step": 15950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.191456616327402e-05,
+      "loss": 0.5011,
+      "step": 15951
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.190374617380585e-05,
+      "loss": 0.5024,
+      "step": 15952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.189292640336271e-05,
+      "loss": 0.5328,
+      "step": 15953
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.188210685207547e-05,
+      "loss": 0.4226,
+      "step": 15954
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.187128752007517e-05,
+      "loss": 0.5532,
+      "step": 15955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.186046840749273e-05,
+      "loss": 0.512,
+      "step": 15956
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.184964951445906e-05,
+      "loss": 0.5009,
+      "step": 15957
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.183883084110516e-05,
+      "loss": 0.5302,
+      "step": 15958
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.182801238756194e-05,
+      "loss": 0.4344,
+      "step": 15959
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.181719415396032e-05,
+      "loss": 0.5614,
+      "step": 15960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.18063761404313e-05,
+      "loss": 0.5321,
+      "step": 15961
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.179555834710575e-05,
+      "loss": 0.5086,
+      "step": 15962
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.178474077411463e-05,
+      "loss": 0.5144,
+      "step": 15963
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.177392342158889e-05,
+      "loss": 0.511,
+      "step": 15964
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.17631062896594e-05,
+      "loss": 0.5127,
+      "step": 15965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.175228937845714e-05,
+      "loss": 0.5963,
+      "step": 15966
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.174147268811298e-05,
+      "loss": 0.4966,
+      "step": 15967
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.173065621875789e-05,
+      "loss": 0.5981,
+      "step": 15968
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.171983997052275e-05,
+      "loss": 0.5337,
+      "step": 15969
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.170902394353848e-05,
+      "loss": 0.5667,
+      "step": 15970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.1698208137936e-05,
+      "loss": 0.5085,
+      "step": 15971
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.168739255384619e-05,
+      "loss": 0.497,
+      "step": 15972
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.167657719139999e-05,
+      "loss": 0.567,
+      "step": 15973
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.166576205072829e-05,
+      "loss": 0.5774,
+      "step": 15974
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.165494713196199e-05,
+      "loss": 0.4971,
+      "step": 15975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.164413243523196e-05,
+      "loss": 0.4641,
+      "step": 15976
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.163331796066914e-05,
+      "loss": 0.4658,
+      "step": 15977
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.162250370840437e-05,
+      "loss": 0.5076,
+      "step": 15978
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.16116896785686e-05,
+      "loss": 0.4812,
+      "step": 15979
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.160087587129264e-05,
+      "loss": 0.5481,
+      "step": 15980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.159006228670745e-05,
+      "loss": 0.5138,
+      "step": 15981
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.157924892494389e-05,
+      "loss": 0.5543,
+      "step": 15982
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.15684357861328e-05,
+      "loss": 0.5626,
+      "step": 15983
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.155762287040505e-05,
+      "loss": 0.6327,
+      "step": 15984
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.154681017789158e-05,
+      "loss": 0.4163,
+      "step": 15985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.153599770872321e-05,
+      "loss": 0.4988,
+      "step": 15986
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.15251854630308e-05,
+      "loss": 0.604,
+      "step": 15987
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.151437344094523e-05,
+      "loss": 0.5777,
+      "step": 15988
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.150356164259737e-05,
+      "loss": 0.4708,
+      "step": 15989
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.149275006811809e-05,
+      "loss": 0.5278,
+      "step": 15990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.148193871763821e-05,
+      "loss": 0.4823,
+      "step": 15991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.147112759128859e-05,
+      "loss": 0.4754,
+      "step": 15992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.146031668920011e-05,
+      "loss": 0.5041,
+      "step": 15993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.14495060115036e-05,
+      "loss": 0.547,
+      "step": 15994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.143869555832987e-05,
+      "loss": 0.6138,
+      "step": 15995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.142788532980983e-05,
+      "loss": 0.5109,
+      "step": 15996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.141707532607427e-05,
+      "loss": 0.4659,
+      "step": 15997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.140626554725405e-05,
+      "loss": 0.5751,
+      "step": 15998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.139545599348002e-05,
+      "loss": 0.5059,
+      "step": 15999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.138464666488296e-05,
+      "loss": 0.5357,
+      "step": 16000
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 0.5129332542419434,
+      "eval_runtime": 847.7068,
+      "eval_samples_per_second": 28.382,
+      "eval_steps_per_second": 14.191,
+      "step": 16000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.137383756159375e-05,
+      "loss": 0.5455,
+      "step": 16001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.136302868374317e-05,
+      "loss": 0.5452,
+      "step": 16002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.135222003146209e-05,
+      "loss": 0.5291,
+      "step": 16003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.134141160488132e-05,
+      "loss": 0.4943,
+      "step": 16004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.133060340413163e-05,
+      "loss": 0.4472,
+      "step": 16005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.131979542934389e-05,
+      "loss": 0.5117,
+      "step": 16006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.130898768064888e-05,
+      "loss": 0.5495,
+      "step": 16007
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.129818015817745e-05,
+      "loss": 0.5534,
+      "step": 16008
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.128737286206038e-05,
+      "loss": 0.5394,
+      "step": 16009
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.127656579242846e-05,
+      "loss": 0.5298,
+      "step": 16010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.126575894941251e-05,
+      "loss": 0.4628,
+      "step": 16011
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.125495233314335e-05,
+      "loss": 0.6152,
+      "step": 16012
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.124414594375171e-05,
+      "loss": 0.5194,
+      "step": 16013
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.123333978136847e-05,
+      "loss": 0.5097,
+      "step": 16014
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.122253384612435e-05,
+      "loss": 0.607,
+      "step": 16015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.121172813815016e-05,
+      "loss": 0.5245,
+      "step": 16016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.120092265757672e-05,
+      "loss": 0.562,
+      "step": 16017
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.119011740453474e-05,
+      "loss": 0.452,
+      "step": 16018
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.117931237915508e-05,
+      "loss": 0.5287,
+      "step": 16019
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.116850758156844e-05,
+      "loss": 0.6148,
+      "step": 16020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.115770301190566e-05,
+      "loss": 0.5372,
+      "step": 16021
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.114689867029749e-05,
+      "loss": 0.4751,
+      "step": 16022
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.113609455687466e-05,
+      "loss": 0.5768,
+      "step": 16023
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.1125290671768e-05,
+      "loss": 0.5884,
+      "step": 16024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.111448701510824e-05,
+      "loss": 0.5102,
+      "step": 16025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.110368358702611e-05,
+      "loss": 0.4304,
+      "step": 16026
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.109288038765243e-05,
+      "loss": 0.5715,
+      "step": 16027
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.108207741711789e-05,
+      "loss": 0.4505,
+      "step": 16028
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.107127467555331e-05,
+      "loss": 0.6283,
+      "step": 16029
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.10604721630894e-05,
+      "loss": 0.6094,
+      "step": 16030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.104966987985688e-05,
+      "loss": 0.4682,
+      "step": 16031
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.103886782598655e-05,
+      "loss": 0.4361,
+      "step": 16032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.102806600160911e-05,
+      "loss": 0.6252,
+      "step": 16033
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.101726440685532e-05,
+      "loss": 0.5044,
+      "step": 16034
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.100646304185593e-05,
+      "loss": 0.5445,
+      "step": 16035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.099566190674162e-05,
+      "loss": 0.5295,
+      "step": 16036
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.098486100164317e-05,
+      "loss": 0.5013,
+      "step": 16037
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.097406032669128e-05,
+      "loss": 0.4183,
+      "step": 16038
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.096325988201668e-05,
+      "loss": 0.5027,
+      "step": 16039
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.095245966775012e-05,
+      "loss": 0.4867,
+      "step": 16040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.094165968402225e-05,
+      "loss": 0.5601,
+      "step": 16041
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.093085993096385e-05,
+      "loss": 0.4879,
+      "step": 16042
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.092006040870562e-05,
+      "loss": 0.5617,
+      "step": 16043
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.090926111737823e-05,
+      "loss": 0.4672,
+      "step": 16044
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.089846205711244e-05,
+      "loss": 0.5637,
+      "step": 16045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.088766322803892e-05,
+      "loss": 0.4741,
+      "step": 16046
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.08768646302884e-05,
+      "loss": 0.5173,
+      "step": 16047
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.086606626399158e-05,
+      "loss": 0.5887,
+      "step": 16048
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.08552681292791e-05,
+      "loss": 0.6075,
+      "step": 16049
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.084447022628173e-05,
+      "loss": 0.5363,
+      "step": 16050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.08336725551301e-05,
+      "loss": 0.5349,
+      "step": 16051
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.082287511595493e-05,
+      "loss": 0.5211,
+      "step": 16052
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.08120779088869e-05,
+      "loss": 0.5478,
+      "step": 16053
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.080128093405667e-05,
+      "loss": 0.4843,
+      "step": 16054
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.079048419159496e-05,
+      "loss": 0.5244,
+      "step": 16055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.077968768163242e-05,
+      "loss": 0.6191,
+      "step": 16056
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.076889140429972e-05,
+      "loss": 0.4861,
+      "step": 16057
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.075809535972758e-05,
+      "loss": 0.6012,
+      "step": 16058
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.074729954804657e-05,
+      "loss": 0.4361,
+      "step": 16059
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.073650396938746e-05,
+      "loss": 0.6175,
+      "step": 16060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.072570862388084e-05,
+      "loss": 0.576,
+      "step": 16061
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.071491351165739e-05,
+      "loss": 0.5102,
+      "step": 16062
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.070411863284779e-05,
+      "loss": 0.5404,
+      "step": 16063
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.069332398758267e-05,
+      "loss": 0.5493,
+      "step": 16064
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.068252957599269e-05,
+      "loss": 0.5114,
+      "step": 16065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.06717353982085e-05,
+      "loss": 0.5009,
+      "step": 16066
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.066094145436072e-05,
+      "loss": 0.5934,
+      "step": 16067
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.065014774458003e-05,
+      "loss": 0.4527,
+      "step": 16068
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.063935426899706e-05,
+      "loss": 0.5678,
+      "step": 16069
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.062856102774244e-05,
+      "loss": 0.5988,
+      "step": 16070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.06177680209468e-05,
+      "loss": 0.524,
+      "step": 16071
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.060697524874077e-05,
+      "loss": 0.5345,
+      "step": 16072
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.059618271125499e-05,
+      "loss": 0.6059,
+      "step": 16073
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.05853904086201e-05,
+      "loss": 0.5121,
+      "step": 16074
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.057459834096668e-05,
+      "loss": 0.5401,
+      "step": 16075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.056380650842539e-05,
+      "loss": 0.5231,
+      "step": 16076
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.05530149111268e-05,
+      "loss": 0.519,
+      "step": 16077
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.054222354920159e-05,
+      "loss": 0.5771,
+      "step": 16078
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.053143242278033e-05,
+      "loss": 0.4781,
+      "step": 16079
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.05206415319936e-05,
+      "loss": 0.5299,
+      "step": 16080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.05098508769721e-05,
+      "loss": 0.569,
+      "step": 16081
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.049906045784632e-05,
+      "loss": 0.5082,
+      "step": 16082
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.048827027474695e-05,
+      "loss": 0.612,
+      "step": 16083
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.047748032780453e-05,
+      "loss": 0.484,
+      "step": 16084
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.046669061714968e-05,
+      "loss": 0.4539,
+      "step": 16085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.045590114291299e-05,
+      "loss": 0.571,
+      "step": 16086
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.044511190522503e-05,
+      "loss": 0.4772,
+      "step": 16087
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.043432290421644e-05,
+      "loss": 0.5675,
+      "step": 16088
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.042353414001774e-05,
+      "loss": 0.5531,
+      "step": 16089
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.041274561275952e-05,
+      "loss": 0.4868,
+      "step": 16090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.04019573225724e-05,
+      "loss": 0.4957,
+      "step": 16091
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.039116926958692e-05,
+      "loss": 0.4261,
+      "step": 16092
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.038038145393363e-05,
+      "loss": 0.6108,
+      "step": 16093
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.036959387574315e-05,
+      "loss": 0.5628,
+      "step": 16094
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.035880653514601e-05,
+      "loss": 0.535,
+      "step": 16095
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.034801943227279e-05,
+      "loss": 0.5504,
+      "step": 16096
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.033723256725406e-05,
+      "loss": 0.4492,
+      "step": 16097
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.032644594022031e-05,
+      "loss": 0.5029,
+      "step": 16098
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.03156595513022e-05,
+      "loss": 0.4881,
+      "step": 16099
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.03048734006302e-05,
+      "loss": 0.4899,
+      "step": 16100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.02940874883349e-05,
+      "loss": 0.5219,
+      "step": 16101
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.028330181454681e-05,
+      "loss": 0.5068,
+      "step": 16102
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.02725163793965e-05,
+      "loss": 0.5332,
+      "step": 16103
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.026173118301451e-05,
+      "loss": 0.5294,
+      "step": 16104
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.025094622553136e-05,
+      "loss": 0.5288,
+      "step": 16105
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.02401615070776e-05,
+      "loss": 0.5049,
+      "step": 16106
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.022937702778374e-05,
+      "loss": 0.5107,
+      "step": 16107
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.021859278778034e-05,
+      "loss": 0.5991,
+      "step": 16108
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.02078087871979e-05,
+      "loss": 0.558,
+      "step": 16109
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.019702502616696e-05,
+      "loss": 0.5901,
+      "step": 16110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.018624150481801e-05,
+      "loss": 0.4961,
+      "step": 16111
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.01754582232816e-05,
+      "loss": 0.5194,
+      "step": 16112
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.016467518168821e-05,
+      "loss": 0.5295,
+      "step": 16113
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.01538923801684e-05,
+      "loss": 0.504,
+      "step": 16114
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.014310981885264e-05,
+      "loss": 0.5349,
+      "step": 16115
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.013232749787142e-05,
+      "loss": 0.5385,
+      "step": 16116
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.012154541735531e-05,
+      "loss": 0.5833,
+      "step": 16117
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.011076357743474e-05,
+      "loss": 0.458,
+      "step": 16118
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.009998197824025e-05,
+      "loss": 0.4673,
+      "step": 16119
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.008920061990227e-05,
+      "loss": 0.5367,
+      "step": 16120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.007841950255138e-05,
+      "loss": 0.5367,
+      "step": 16121
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.006763862631797e-05,
+      "loss": 0.5309,
+      "step": 16122
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.005685799133261e-05,
+      "loss": 0.5254,
+      "step": 16123
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.004607759772576e-05,
+      "loss": 0.5518,
+      "step": 16124
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.003529744562785e-05,
+      "loss": 0.4347,
+      "step": 16125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.00245175351694e-05,
+      "loss": 0.5569,
+      "step": 16126
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.00137378664809e-05,
+      "loss": 0.5538,
+      "step": 16127
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.000295843969276e-05,
+      "loss": 0.4919,
+      "step": 16128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.999217925493552e-05,
+      "loss": 0.5201,
+      "step": 16129
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.998140031233957e-05,
+      "loss": 0.5809,
+      "step": 16130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.997062161203544e-05,
+      "loss": 0.5153,
+      "step": 16131
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.995984315415355e-05,
+      "loss": 0.6043,
+      "step": 16132
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.994906493882433e-05,
+      "loss": 0.5748,
+      "step": 16133
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.993828696617828e-05,
+      "loss": 0.5166,
+      "step": 16134
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.992750923634585e-05,
+      "loss": 0.4613,
+      "step": 16135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.991673174945745e-05,
+      "loss": 0.4842,
+      "step": 16136
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.990595450564355e-05,
+      "loss": 0.6461,
+      "step": 16137
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.989517750503456e-05,
+      "loss": 0.5221,
+      "step": 16138
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.988440074776098e-05,
+      "loss": 0.488,
+      "step": 16139
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.987362423395317e-05,
+      "loss": 0.5085,
+      "step": 16140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.986284796374162e-05,
+      "loss": 0.5982,
+      "step": 16141
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.985207193725675e-05,
+      "loss": 0.5483,
+      "step": 16142
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.984129615462893e-05,
+      "loss": 0.492,
+      "step": 16143
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.983052061598866e-05,
+      "loss": 0.4657,
+      "step": 16144
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.981974532146632e-05,
+      "loss": 0.5583,
+      "step": 16145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.980897027119231e-05,
+      "loss": 0.569,
+      "step": 16146
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.979819546529708e-05,
+      "loss": 0.4646,
+      "step": 16147
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.978742090391101e-05,
+      "loss": 0.4733,
+      "step": 16148
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.977664658716455e-05,
+      "loss": 0.4634,
+      "step": 16149
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.976587251518808e-05,
+      "loss": 0.4794,
+      "step": 16150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.975509868811197e-05,
+      "loss": 0.4745,
+      "step": 16151
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.974432510606668e-05,
+      "loss": 0.5067,
+      "step": 16152
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.973355176918255e-05,
+      "loss": 0.5137,
+      "step": 16153
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.972277867759004e-05,
+      "loss": 0.526,
+      "step": 16154
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.971200583141948e-05,
+      "loss": 0.5091,
+      "step": 16155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.970123323080126e-05,
+      "loss": 0.5732,
+      "step": 16156
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.969046087586582e-05,
+      "loss": 0.4891,
+      "step": 16157
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.967968876674348e-05,
+      "loss": 0.4684,
+      "step": 16158
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.966891690356465e-05,
+      "loss": 0.5286,
+      "step": 16159
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.96581452864597e-05,
+      "loss": 0.5145,
+      "step": 16160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.964737391555899e-05,
+      "loss": 0.5127,
+      "step": 16161
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.963660279099292e-05,
+      "loss": 0.6557,
+      "step": 16162
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.962583191289183e-05,
+      "loss": 0.5972,
+      "step": 16163
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.961506128138607e-05,
+      "loss": 0.4517,
+      "step": 16164
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.960429089660604e-05,
+      "loss": 0.5083,
+      "step": 16165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.959352075868206e-05,
+      "loss": 0.4131,
+      "step": 16166
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.958275086774453e-05,
+      "loss": 0.5717,
+      "step": 16167
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.957198122392377e-05,
+      "loss": 0.5007,
+      "step": 16168
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.95612118273501e-05,
+      "loss": 0.6055,
+      "step": 16169
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.955044267815394e-05,
+      "loss": 0.4791,
+      "step": 16170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.953967377646555e-05,
+      "loss": 0.493,
+      "step": 16171
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.952890512241536e-05,
+      "loss": 0.5994,
+      "step": 16172
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.951813671613363e-05,
+      "loss": 0.6385,
+      "step": 16173
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.950736855775071e-05,
+      "loss": 0.5411,
+      "step": 16174
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.949660064739696e-05,
+      "loss": 0.5392,
+      "step": 16175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.94858329852027e-05,
+      "loss": 0.5484,
+      "step": 16176
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.94750655712982e-05,
+      "loss": 0.5205,
+      "step": 16177
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.946429840581386e-05,
+      "loss": 0.5289,
+      "step": 16178
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.945353148887995e-05,
+      "loss": 0.4839,
+      "step": 16179
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.944276482062681e-05,
+      "loss": 0.4654,
+      "step": 16180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.943199840118474e-05,
+      "loss": 0.4936,
+      "step": 16181
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.942123223068405e-05,
+      "loss": 0.5893,
+      "step": 16182
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.941046630925504e-05,
+      "loss": 0.5521,
+      "step": 16183
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.939970063702803e-05,
+      "loss": 0.5407,
+      "step": 16184
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.938893521413331e-05,
+      "loss": 0.4817,
+      "step": 16185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.93781700407012e-05,
+      "loss": 0.4968,
+      "step": 16186
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.936740511686193e-05,
+      "loss": 0.4784,
+      "step": 16187
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.935664044274589e-05,
+      "loss": 0.5689,
+      "step": 16188
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.934587601848326e-05,
+      "loss": 0.5423,
+      "step": 16189
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.933511184420443e-05,
+      "loss": 0.5266,
+      "step": 16190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.932434792003961e-05,
+      "loss": 0.548,
+      "step": 16191
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.931358424611909e-05,
+      "loss": 0.5489,
+      "step": 16192
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.930282082257318e-05,
+      "loss": 0.6846,
+      "step": 16193
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.929205764953212e-05,
+      "loss": 0.4805,
+      "step": 16194
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.928129472712617e-05,
+      "loss": 0.5037,
+      "step": 16195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.927053205548565e-05,
+      "loss": 0.5608,
+      "step": 16196
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.925976963474078e-05,
+      "loss": 0.5508,
+      "step": 16197
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.924900746502185e-05,
+      "loss": 0.5806,
+      "step": 16198
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.923824554645908e-05,
+      "loss": 0.5718,
+      "step": 16199
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.922748387918274e-05,
+      "loss": 0.6064,
+      "step": 16200
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 0.5091128945350647,
+      "eval_runtime": 855.7608,
+      "eval_samples_per_second": 28.115,
+      "eval_steps_per_second": 14.058,
+      "step": 16200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.921672246332312e-05,
+      "loss": 0.5176,
+      "step": 16201
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.92059612990104e-05,
+      "loss": 0.4529,
+      "step": 16202
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.91952003863749e-05,
+      "loss": 0.4966,
+      "step": 16203
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.918443972554682e-05,
+      "loss": 0.5549,
+      "step": 16204
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.917367931665638e-05,
+      "loss": 0.3978,
+      "step": 16205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.916291915983388e-05,
+      "loss": 0.5151,
+      "step": 16206
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.915215925520949e-05,
+      "loss": 0.5685,
+      "step": 16207
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.91413996029135e-05,
+      "loss": 0.625,
+      "step": 16208
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.913064020307608e-05,
+      "loss": 0.5981,
+      "step": 16209
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.911988105582747e-05,
+      "loss": 0.5551,
+      "step": 16210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.910912216129793e-05,
+      "loss": 0.5671,
+      "step": 16211
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.909836351961764e-05,
+      "loss": 0.5823,
+      "step": 16212
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.908760513091682e-05,
+      "loss": 0.5518,
+      "step": 16213
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.90768469953257e-05,
+      "loss": 0.5509,
+      "step": 16214
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.906608911297445e-05,
+      "loss": 0.5513,
+      "step": 16215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.905533148399334e-05,
+      "loss": 0.5136,
+      "step": 16216
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.904457410851253e-05,
+      "loss": 0.5261,
+      "step": 16217
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.90338169866622e-05,
+      "loss": 0.4777,
+      "step": 16218
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.90230601185726e-05,
+      "loss": 0.646,
+      "step": 16219
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.901230350437389e-05,
+      "loss": 0.5166,
+      "step": 16220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.900154714419628e-05,
+      "loss": 0.5325,
+      "step": 16221
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.899079103816993e-05,
+      "loss": 0.5208,
+      "step": 16222
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.898003518642504e-05,
+      "loss": 0.5716,
+      "step": 16223
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.896927958909181e-05,
+      "loss": 0.5546,
+      "step": 16224
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.895852424630038e-05,
+      "loss": 0.5102,
+      "step": 16225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.894776915818097e-05,
+      "loss": 0.5328,
+      "step": 16226
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.893701432486373e-05,
+      "loss": 0.5185,
+      "step": 16227
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.89262597464788e-05,
+      "loss": 0.5808,
+      "step": 16228
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.891550542315641e-05,
+      "loss": 0.4469,
+      "step": 16229
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.890475135502669e-05,
+      "loss": 0.5747,
+      "step": 16230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.889399754221976e-05,
+      "loss": 0.5172,
+      "step": 16231
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.888324398486584e-05,
+      "loss": 0.5719,
+      "step": 16232
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.887249068309504e-05,
+      "loss": 0.5422,
+      "step": 16233
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.886173763703757e-05,
+      "loss": 0.5184,
+      "step": 16234
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.885098484682351e-05,
+      "loss": 0.574,
+      "step": 16235
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.884023231258304e-05,
+      "loss": 0.5122,
+      "step": 16236
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.882948003444629e-05,
+      "loss": 0.4832,
+      "step": 16237
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.88187280125434e-05,
+      "loss": 0.4523,
+      "step": 16238
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.880797624700451e-05,
+      "loss": 0.5437,
+      "step": 16239
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.879722473795977e-05,
+      "loss": 0.4724,
+      "step": 16240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.878647348553924e-05,
+      "loss": 0.5583,
+      "step": 16241
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.877572248987315e-05,
+      "loss": 0.5256,
+      "step": 16242
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.876497175109155e-05,
+      "loss": 0.5867,
+      "step": 16243
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.875422126932456e-05,
+      "loss": 0.5103,
+      "step": 16244
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.874347104470234e-05,
+      "loss": 0.6124,
+      "step": 16245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.873272107735495e-05,
+      "loss": 0.5166,
+      "step": 16246
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.872197136741255e-05,
+      "loss": 0.5646,
+      "step": 16247
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.871122191500522e-05,
+      "loss": 0.4972,
+      "step": 16248
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.870047272026306e-05,
+      "loss": 0.4031,
+      "step": 16249
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.868972378331619e-05,
+      "loss": 0.525,
+      "step": 16250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.86789751042947e-05,
+      "loss": 0.4807,
+      "step": 16251
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.866822668332871e-05,
+      "loss": 0.4887,
+      "step": 16252
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.865747852054826e-05,
+      "loss": 0.4823,
+      "step": 16253
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.864673061608347e-05,
+      "loss": 0.5151,
+      "step": 16254
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.863598297006439e-05,
+      "loss": 0.5133,
+      "step": 16255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.862523558262116e-05,
+      "loss": 0.5559,
+      "step": 16256
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.861448845388384e-05,
+      "loss": 0.5363,
+      "step": 16257
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.860374158398247e-05,
+      "loss": 0.6491,
+      "step": 16258
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.859299497304716e-05,
+      "loss": 0.5418,
+      "step": 16259
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.858224862120796e-05,
+      "loss": 0.5581,
+      "step": 16260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.857150252859498e-05,
+      "loss": 0.5259,
+      "step": 16261
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.856075669533822e-05,
+      "loss": 0.5333,
+      "step": 16262
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.855001112156776e-05,
+      "loss": 0.5223,
+      "step": 16263
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.85392658074137e-05,
+      "loss": 0.4699,
+      "step": 16264
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.852852075300607e-05,
+      "loss": 0.4975,
+      "step": 16265
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.851777595847488e-05,
+      "loss": 0.4817,
+      "step": 16266
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.850703142395023e-05,
+      "loss": 0.5392,
+      "step": 16267
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.849628714956212e-05,
+      "loss": 0.5194,
+      "step": 16268
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.848554313544064e-05,
+      "loss": 0.6074,
+      "step": 16269
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.847479938171582e-05,
+      "loss": 0.5507,
+      "step": 16270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.846405588851764e-05,
+      "loss": 0.4828,
+      "step": 16271
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.845331265597621e-05,
+      "loss": 0.4043,
+      "step": 16272
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.844256968422149e-05,
+      "loss": 0.4928,
+      "step": 16273
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.843182697338357e-05,
+      "loss": 0.5229,
+      "step": 16274
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.842108452359243e-05,
+      "loss": 0.5731,
+      "step": 16275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.841034233497808e-05,
+      "loss": 0.5335,
+      "step": 16276
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.839960040767057e-05,
+      "loss": 0.4926,
+      "step": 16277
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.838885874179991e-05,
+      "loss": 0.5349,
+      "step": 16278
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.837811733749607e-05,
+      "loss": 0.5144,
+      "step": 16279
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.836737619488911e-05,
+      "loss": 0.5349,
+      "step": 16280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.835663531410899e-05,
+      "loss": 0.5871,
+      "step": 16281
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.834589469528575e-05,
+      "loss": 0.579,
+      "step": 16282
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.833515433854936e-05,
+      "loss": 0.5933,
+      "step": 16283
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.83244142440298e-05,
+      "loss": 0.5607,
+      "step": 16284
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.831367441185712e-05,
+      "loss": 0.4146,
+      "step": 16285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.830293484216124e-05,
+      "loss": 0.58,
+      "step": 16286
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.829219553507219e-05,
+      "loss": 0.5546,
+      "step": 16287
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.828145649071994e-05,
+      "loss": 0.4819,
+      "step": 16288
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.827071770923445e-05,
+      "loss": 0.4723,
+      "step": 16289
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.825997919074571e-05,
+      "loss": 0.4935,
+      "step": 16290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.82492409353837e-05,
+      "loss": 0.4921,
+      "step": 16291
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.823850294327837e-05,
+      "loss": 0.4319,
+      "step": 16292
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.822776521455971e-05,
+      "loss": 0.5052,
+      "step": 16293
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.821702774935765e-05,
+      "loss": 0.5526,
+      "step": 16294
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.820629054780218e-05,
+      "loss": 0.4584,
+      "step": 16295
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.819555361002325e-05,
+      "loss": 0.4229,
+      "step": 16296
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.818481693615079e-05,
+      "loss": 0.4742,
+      "step": 16297
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.817408052631477e-05,
+      "loss": 0.6128,
+      "step": 16298
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.816334438064513e-05,
+      "loss": 0.5184,
+      "step": 16299
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.815260849927181e-05,
+      "loss": 0.5447,
+      "step": 16300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.814187288232477e-05,
+      "loss": 0.5729,
+      "step": 16301
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.81311375299339e-05,
+      "loss": 0.4521,
+      "step": 16302
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.812040244222919e-05,
+      "loss": 0.5097,
+      "step": 16303
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.810966761934053e-05,
+      "loss": 0.4831,
+      "step": 16304
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.809893306139788e-05,
+      "loss": 0.6057,
+      "step": 16305
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.808819876853114e-05,
+      "loss": 0.5007,
+      "step": 16306
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.807746474087022e-05,
+      "loss": 0.5395,
+      "step": 16307
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.806673097854508e-05,
+      "loss": 0.5048,
+      "step": 16308
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.805599748168558e-05,
+      "loss": 0.4422,
+      "step": 16309
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.804526425042169e-05,
+      "loss": 0.57,
+      "step": 16310
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.803453128488328e-05,
+      "loss": 0.4472,
+      "step": 16311
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.802379858520024e-05,
+      "loss": 0.4259,
+      "step": 16312
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.801306615150253e-05,
+      "loss": 0.5265,
+      "step": 16313
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.800233398391999e-05,
+      "loss": 0.4422,
+      "step": 16314
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.799160208258253e-05,
+      "loss": 0.5174,
+      "step": 16315
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.798087044762007e-05,
+      "loss": 0.5629,
+      "step": 16316
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.797013907916244e-05,
+      "loss": 0.5313,
+      "step": 16317
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.79594079773396e-05,
+      "loss": 0.5496,
+      "step": 16318
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.79486771422814e-05,
+      "loss": 0.4913,
+      "step": 16319
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.793794657411768e-05,
+      "loss": 0.4901,
+      "step": 16320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.792721627297837e-05,
+      "loss": 0.472,
+      "step": 16321
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.79164862389933e-05,
+      "loss": 0.523,
+      "step": 16322
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.79057564722924e-05,
+      "loss": 0.4895,
+      "step": 16323
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.789502697300547e-05,
+      "loss": 0.6155,
+      "step": 16324
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.788429774126239e-05,
+      "loss": 0.5636,
+      "step": 16325
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.787356877719305e-05,
+      "loss": 0.5249,
+      "step": 16326
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.786284008092729e-05,
+      "loss": 0.5859,
+      "step": 16327
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.785211165259492e-05,
+      "loss": 0.6229,
+      "step": 16328
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.784138349232587e-05,
+      "loss": 0.5466,
+      "step": 16329
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.783065560024992e-05,
+      "loss": 0.4729,
+      "step": 16330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.781992797649695e-05,
+      "loss": 0.466,
+      "step": 16331
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.780920062119679e-05,
+      "loss": 0.4796,
+      "step": 16332
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.779847353447925e-05,
+      "loss": 0.5311,
+      "step": 16333
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.778774671647422e-05,
+      "loss": 0.535,
+      "step": 16334
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.777702016731148e-05,
+      "loss": 0.5639,
+      "step": 16335
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.776629388712088e-05,
+      "loss": 0.5174,
+      "step": 16336
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.775556787603224e-05,
+      "loss": 0.5077,
+      "step": 16337
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.774484213417536e-05,
+      "loss": 0.5204,
+      "step": 16338
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.773411666168011e-05,
+      "loss": 0.4944,
+      "step": 16339
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.772339145867624e-05,
+      "loss": 0.52,
+      "step": 16340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.771266652529362e-05,
+      "loss": 0.5305,
+      "step": 16341
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.770194186166202e-05,
+      "loss": 0.4147,
+      "step": 16342
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.769121746791123e-05,
+      "loss": 0.4592,
+      "step": 16343
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.768049334417111e-05,
+      "loss": 0.4895,
+      "step": 16344
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.76697694905714e-05,
+      "loss": 0.6529,
+      "step": 16345
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.765904590724191e-05,
+      "loss": 0.4791,
+      "step": 16346
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.764832259431245e-05,
+      "loss": 0.5742,
+      "step": 16347
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.763759955191278e-05,
+      "loss": 0.4236,
+      "step": 16348
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.762687678017272e-05,
+      "loss": 0.4749,
+      "step": 16349
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.761615427922203e-05,
+      "loss": 0.4624,
+      "step": 16350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.760543204919046e-05,
+      "loss": 0.5705,
+      "step": 16351
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.759471009020784e-05,
+      "loss": 0.5071,
+      "step": 16352
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.75839884024039e-05,
+      "loss": 0.5194,
+      "step": 16353
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.757326698590845e-05,
+      "loss": 0.6257,
+      "step": 16354
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.756254584085121e-05,
+      "loss": 0.5737,
+      "step": 16355
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.755182496736194e-05,
+      "loss": 0.5893,
+      "step": 16356
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.754110436557045e-05,
+      "loss": 0.5162,
+      "step": 16357
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.753038403560643e-05,
+      "loss": 0.5267,
+      "step": 16358
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.751966397759968e-05,
+      "loss": 0.5125,
+      "step": 16359
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.750894419167995e-05,
+      "loss": 0.5385,
+      "step": 16360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.749822467797694e-05,
+      "loss": 0.5508,
+      "step": 16361
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.748750543662044e-05,
+      "loss": 0.6066,
+      "step": 16362
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.747678646774018e-05,
+      "loss": 0.5106,
+      "step": 16363
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.746606777146585e-05,
+      "loss": 0.5223,
+      "step": 16364
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.745534934792724e-05,
+      "loss": 0.5296,
+      "step": 16365
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.744463119725404e-05,
+      "loss": 0.5141,
+      "step": 16366
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.7433913319576e-05,
+      "loss": 0.5304,
+      "step": 16367
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.742319571502284e-05,
+      "loss": 0.4798,
+      "step": 16368
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.741247838372425e-05,
+      "loss": 0.6636,
+      "step": 16369
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.740176132580997e-05,
+      "loss": 0.554,
+      "step": 16370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.73910445414097e-05,
+      "loss": 0.4377,
+      "step": 16371
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.738032803065319e-05,
+      "loss": 0.5145,
+      "step": 16372
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.736961179367008e-05,
+      "loss": 0.5352,
+      "step": 16373
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.73588958305901e-05,
+      "loss": 0.5149,
+      "step": 16374
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.734818014154299e-05,
+      "loss": 0.5553,
+      "step": 16375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.733746472665837e-05,
+      "loss": 0.4605,
+      "step": 16376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.7326749586066e-05,
+      "loss": 0.5722,
+      "step": 16377
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.731603471989554e-05,
+      "loss": 0.53,
+      "step": 16378
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.730532012827665e-05,
+      "loss": 0.537,
+      "step": 16379
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.729460581133906e-05,
+      "loss": 0.5016,
+      "step": 16380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.728389176921243e-05,
+      "loss": 0.5516,
+      "step": 16381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.72731780020264e-05,
+      "loss": 0.4698,
+      "step": 16382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.726246450991071e-05,
+      "loss": 0.4899,
+      "step": 16383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.725175129299498e-05,
+      "loss": 0.4868,
+      "step": 16384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.72410383514089e-05,
+      "loss": 0.5083,
+      "step": 16385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.723032568528213e-05,
+      "loss": 0.4359,
+      "step": 16386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.721961329474432e-05,
+      "loss": 0.516,
+      "step": 16387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.72089011799251e-05,
+      "loss": 0.5477,
+      "step": 16388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.719818934095416e-05,
+      "loss": 0.5489,
+      "step": 16389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.718747777796115e-05,
+      "loss": 0.4892,
+      "step": 16390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.717676649107568e-05,
+      "loss": 0.6923,
+      "step": 16391
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.716605548042743e-05,
+      "loss": 0.4815,
+      "step": 16392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.715534474614601e-05,
+      "loss": 0.5186,
+      "step": 16393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.714463428836108e-05,
+      "loss": 0.561,
+      "step": 16394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.713392410720227e-05,
+      "loss": 0.5681,
+      "step": 16395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.712321420279919e-05,
+      "loss": 0.5771,
+      "step": 16396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.711250457528149e-05,
+      "loss": 0.5098,
+      "step": 16397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.710179522477877e-05,
+      "loss": 0.4929,
+      "step": 16398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.709108615142064e-05,
+      "loss": 0.493,
+      "step": 16399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.708037735533676e-05,
+      "loss": 0.4835,
+      "step": 16400
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 0.5048079490661621,
+      "eval_runtime": 854.406,
+      "eval_samples_per_second": 28.16,
+      "eval_steps_per_second": 14.08,
+      "step": 16400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.706966883665669e-05,
+      "loss": 0.5429,
+      "step": 16401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.70589605955101e-05,
+      "loss": 0.5375,
+      "step": 16402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.704825263202654e-05,
+      "loss": 0.5409,
+      "step": 16403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.703754494633562e-05,
+      "loss": 0.6105,
+      "step": 16404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.702683753856696e-05,
+      "loss": 0.5473,
+      "step": 16405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.701613040885014e-05,
+      "loss": 0.4377,
+      "step": 16406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.700542355731478e-05,
+      "loss": 0.5017,
+      "step": 16407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.699471698409044e-05,
+      "loss": 0.5488,
+      "step": 16408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.698401068930668e-05,
+      "loss": 0.5275,
+      "step": 16409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.697330467309315e-05,
+      "loss": 0.4413,
+      "step": 16410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.696259893557937e-05,
+      "loss": 0.4977,
+      "step": 16411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.695189347689496e-05,
+      "loss": 0.4585,
+      "step": 16412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.694118829716945e-05,
+      "loss": 0.4818,
+      "step": 16413
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.693048339653242e-05,
+      "loss": 0.598,
+      "step": 16414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.691977877511346e-05,
+      "loss": 0.5399,
+      "step": 16415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.690907443304212e-05,
+      "loss": 0.4596,
+      "step": 16416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.689837037044791e-05,
+      "loss": 0.4595,
+      "step": 16417
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.688766658746047e-05,
+      "loss": 0.477,
+      "step": 16418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.687696308420928e-05,
+      "loss": 0.5074,
+      "step": 16419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.686625986082393e-05,
+      "loss": 0.4202,
+      "step": 16420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.685555691743396e-05,
+      "loss": 0.4978,
+      "step": 16421
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.684485425416888e-05,
+      "loss": 0.5547,
+      "step": 16422
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.683415187115827e-05,
+      "loss": 0.5791,
+      "step": 16423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.682344976853162e-05,
+      "loss": 0.4661,
+      "step": 16424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.68127479464185e-05,
+      "loss": 0.4868,
+      "step": 16425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.680204640494845e-05,
+      "loss": 0.4724,
+      "step": 16426
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.679134514425092e-05,
+      "loss": 0.5421,
+      "step": 16427
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.678064416445552e-05,
+      "loss": 0.4133,
+      "step": 16428
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.67699434656917e-05,
+      "loss": 0.4728,
+      "step": 16429
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.675924304808901e-05,
+      "loss": 0.5308,
+      "step": 16430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.674854291177695e-05,
+      "loss": 0.5388,
+      "step": 16431
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.673784305688501e-05,
+      "loss": 0.5277,
+      "step": 16432
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.672714348354275e-05,
+      "loss": 0.4654,
+      "step": 16433
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.671644419187962e-05,
+      "loss": 0.617,
+      "step": 16434
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.67057451820251e-05,
+      "loss": 0.47,
+      "step": 16435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.669504645410874e-05,
+      "loss": 0.4653,
+      "step": 16436
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.668434800826e-05,
+      "loss": 0.4983,
+      "step": 16437
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.667364984460837e-05,
+      "loss": 0.6124,
+      "step": 16438
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.666295196328334e-05,
+      "loss": 0.5101,
+      "step": 16439
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.665225436441436e-05,
+      "loss": 0.5042,
+      "step": 16440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.664155704813096e-05,
+      "loss": 0.5717,
+      "step": 16441
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.663086001456257e-05,
+      "loss": 0.5796,
+      "step": 16442
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.662016326383868e-05,
+      "loss": 0.5591,
+      "step": 16443
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.660946679608876e-05,
+      "loss": 0.4412,
+      "step": 16444
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.659877061144223e-05,
+      "loss": 0.5388,
+      "step": 16445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.658807471002862e-05,
+      "loss": 0.5316,
+      "step": 16446
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.657737909197733e-05,
+      "loss": 0.5169,
+      "step": 16447
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.656668375741782e-05,
+      "loss": 0.5035,
+      "step": 16448
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.655598870647958e-05,
+      "loss": 0.6289,
+      "step": 16449
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.654529393929198e-05,
+      "loss": 0.6063,
+      "step": 16450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.653459945598456e-05,
+      "loss": 0.498,
+      "step": 16451
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.65239052566867e-05,
+      "loss": 0.5104,
+      "step": 16452
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.65132113415278e-05,
+      "loss": 0.4486,
+      "step": 16453
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.650251771063736e-05,
+      "loss": 0.5387,
+      "step": 16454
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.649182436414479e-05,
+      "loss": 0.517,
+      "step": 16455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.64811313021795e-05,
+      "loss": 0.5266,
+      "step": 16456
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.647043852487093e-05,
+      "loss": 0.4935,
+      "step": 16457
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.645974603234847e-05,
+      "loss": 0.457,
+      "step": 16458
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.644905382474158e-05,
+      "loss": 0.568,
+      "step": 16459
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.643836190217961e-05,
+      "loss": 0.4863,
+      "step": 16460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.642767026479202e-05,
+      "loss": 0.4493,
+      "step": 16461
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.641697891270823e-05,
+      "loss": 0.4632,
+      "step": 16462
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.640628784605755e-05,
+      "loss": 0.5535,
+      "step": 16463
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.639559706496948e-05,
+      "loss": 0.4448,
+      "step": 16464
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.638490656957335e-05,
+      "loss": 0.4769,
+      "step": 16465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.637421635999856e-05,
+      "loss": 0.4737,
+      "step": 16466
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.636352643637453e-05,
+      "loss": 0.6289,
+      "step": 16467
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.63528367988306e-05,
+      "loss": 0.6232,
+      "step": 16468
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.63421474474962e-05,
+      "loss": 0.4728,
+      "step": 16469
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.633145838250067e-05,
+      "loss": 0.5767,
+      "step": 16470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.632076960397336e-05,
+      "loss": 0.5561,
+      "step": 16471
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.631008111204371e-05,
+      "loss": 0.4252,
+      "step": 16472
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.629939290684101e-05,
+      "loss": 0.5294,
+      "step": 16473
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.628870498849469e-05,
+      "loss": 0.6928,
+      "step": 16474
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.627801735713409e-05,
+      "loss": 0.4825,
+      "step": 16475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.626733001288851e-05,
+      "loss": 0.5793,
+      "step": 16476
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.62566429558874e-05,
+      "loss": 0.5711,
+      "step": 16477
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.624595618626004e-05,
+      "loss": 0.5846,
+      "step": 16478
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.623526970413577e-05,
+      "loss": 0.6042,
+      "step": 16479
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.622458350964397e-05,
+      "loss": 0.4996,
+      "step": 16480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.621389760291396e-05,
+      "loss": 0.6627,
+      "step": 16481
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.62032119840751e-05,
+      "loss": 0.5162,
+      "step": 16482
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.61925266532567e-05,
+      "loss": 0.4911,
+      "step": 16483
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.618184161058805e-05,
+      "loss": 0.5476,
+      "step": 16484
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.617115685619855e-05,
+      "loss": 0.6318,
+      "step": 16485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.616047239021745e-05,
+      "loss": 0.5614,
+      "step": 16486
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.614978821277414e-05,
+      "loss": 0.464,
+      "step": 16487
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.613910432399788e-05,
+      "loss": 0.5448,
+      "step": 16488
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.612842072401798e-05,
+      "loss": 0.4617,
+      "step": 16489
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.61177374129638e-05,
+      "loss": 0.5489,
+      "step": 16490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.610705439096456e-05,
+      "loss": 0.4623,
+      "step": 16491
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.609637165814965e-05,
+      "loss": 0.4935,
+      "step": 16492
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.608568921464831e-05,
+      "loss": 0.4949,
+      "step": 16493
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.607500706058984e-05,
+      "loss": 0.5165,
+      "step": 16494
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.606432519610356e-05,
+      "loss": 0.4996,
+      "step": 16495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.605364362131873e-05,
+      "loss": 0.4826,
+      "step": 16496
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.60429623363646e-05,
+      "loss": 0.5433,
+      "step": 16497
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.603228134137052e-05,
+      "loss": 0.5665,
+      "step": 16498
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.60216006364657e-05,
+      "loss": 0.556,
+      "step": 16499
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.601092022177949e-05,
+      "loss": 0.4912,
+      "step": 16500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.600024009744109e-05,
+      "loss": 0.5623,
+      "step": 16501
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.598956026357976e-05,
+      "loss": 0.6234,
+      "step": 16502
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.597888072032482e-05,
+      "loss": 0.551,
+      "step": 16503
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.596820146780548e-05,
+      "loss": 0.4881,
+      "step": 16504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.595752250615102e-05,
+      "loss": 0.4831,
+      "step": 16505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.59468438354907e-05,
+      "loss": 0.4882,
+      "step": 16506
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.59361654559537e-05,
+      "loss": 0.5793,
+      "step": 16507
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.592548736766936e-05,
+      "loss": 0.5465,
+      "step": 16508
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.591480957076685e-05,
+      "loss": 0.504,
+      "step": 16509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.590413206537544e-05,
+      "loss": 0.551,
+      "step": 16510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.589345485162436e-05,
+      "loss": 0.5714,
+      "step": 16511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.588277792964281e-05,
+      "loss": 0.4556,
+      "step": 16512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.587210129956007e-05,
+      "loss": 0.5416,
+      "step": 16513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.586142496150533e-05,
+      "loss": 0.5157,
+      "step": 16514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.585074891560779e-05,
+      "loss": 0.4973,
+      "step": 16515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.58400731619967e-05,
+      "loss": 0.5091,
+      "step": 16516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.582939770080127e-05,
+      "loss": 0.5568,
+      "step": 16517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.581872253215069e-05,
+      "loss": 0.5285,
+      "step": 16518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.580804765617419e-05,
+      "loss": 0.4957,
+      "step": 16519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.579737307300092e-05,
+      "loss": 0.453,
+      "step": 16520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.578669878276017e-05,
+      "loss": 0.4171,
+      "step": 16521
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.577602478558106e-05,
+      "loss": 0.6152,
+      "step": 16522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.57653510815928e-05,
+      "loss": 0.4896,
+      "step": 16523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.575467767092454e-05,
+      "loss": 0.4313,
+      "step": 16524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.574400455370553e-05,
+      "loss": 0.5332,
+      "step": 16525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.573333173006489e-05,
+      "loss": 0.5608,
+      "step": 16526
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.572265920013186e-05,
+      "loss": 0.5339,
+      "step": 16527
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.571198696403558e-05,
+      "loss": 0.4548,
+      "step": 16528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.570131502190521e-05,
+      "loss": 0.484,
+      "step": 16529
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.569064337386994e-05,
+      "loss": 0.4975,
+      "step": 16530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.567997202005892e-05,
+      "loss": 0.4156,
+      "step": 16531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.566930096060127e-05,
+      "loss": 0.4783,
+      "step": 16532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.565863019562622e-05,
+      "loss": 0.5082,
+      "step": 16533
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.564795972526286e-05,
+      "loss": 0.4277,
+      "step": 16534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.56372895496404e-05,
+      "loss": 0.5073,
+      "step": 16535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.562661966888794e-05,
+      "loss": 0.5,
+      "step": 16536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.56159500831346e-05,
+      "loss": 0.5216,
+      "step": 16537
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.560528079250957e-05,
+      "loss": 0.4986,
+      "step": 16538
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.559461179714194e-05,
+      "loss": 0.4499,
+      "step": 16539
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.558394309716088e-05,
+      "loss": 0.5009,
+      "step": 16540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.55732746926955e-05,
+      "loss": 0.5603,
+      "step": 16541
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.55626065838749e-05,
+      "loss": 0.4934,
+      "step": 16542
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.555193877082825e-05,
+      "loss": 0.5257,
+      "step": 16543
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.554127125368462e-05,
+      "loss": 0.5285,
+      "step": 16544
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.553060403257315e-05,
+      "loss": 0.4224,
+      "step": 16545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.551993710762295e-05,
+      "loss": 0.4411,
+      "step": 16546
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.550927047896308e-05,
+      "loss": 0.4503,
+      "step": 16547
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.54986041467227e-05,
+      "loss": 0.4788,
+      "step": 16548
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.548793811103089e-05,
+      "loss": 0.4729,
+      "step": 16549
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.54772723720167e-05,
+      "loss": 0.497,
+      "step": 16550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.54666069298093e-05,
+      "loss": 0.5303,
+      "step": 16551
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.545594178453771e-05,
+      "loss": 0.5518,
+      "step": 16552
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.544527693633107e-05,
+      "loss": 0.5503,
+      "step": 16553
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.543461238531841e-05,
+      "loss": 0.5212,
+      "step": 16554
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.542394813162883e-05,
+      "loss": 0.5333,
+      "step": 16555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.541328417539141e-05,
+      "loss": 0.4443,
+      "step": 16556
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.540262051673518e-05,
+      "loss": 0.484,
+      "step": 16557
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.539195715578928e-05,
+      "loss": 0.5787,
+      "step": 16558
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.53812940926827e-05,
+      "loss": 0.5548,
+      "step": 16559
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.537063132754453e-05,
+      "loss": 0.512,
+      "step": 16560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.535996886050382e-05,
+      "loss": 0.547,
+      "step": 16561
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.534930669168962e-05,
+      "loss": 0.4611,
+      "step": 16562
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5338644821231e-05,
+      "loss": 0.4607,
+      "step": 16563
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.532798324925699e-05,
+      "loss": 0.5147,
+      "step": 16564
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.531732197589658e-05,
+      "loss": 0.5377,
+      "step": 16565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.53066610012789e-05,
+      "loss": 0.4705,
+      "step": 16566
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.529600032553293e-05,
+      "loss": 0.5619,
+      "step": 16567
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.528533994878767e-05,
+      "loss": 0.5146,
+      "step": 16568
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.527467987117222e-05,
+      "loss": 0.5533,
+      "step": 16569
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.526402009281553e-05,
+      "loss": 0.4895,
+      "step": 16570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.525336061384669e-05,
+      "loss": 0.4903,
+      "step": 16571
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.524270143439467e-05,
+      "loss": 0.5431,
+      "step": 16572
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.523204255458846e-05,
+      "loss": 0.4443,
+      "step": 16573
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.522138397455712e-05,
+      "loss": 0.6255,
+      "step": 16574
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.521072569442963e-05,
+      "loss": 0.423,
+      "step": 16575
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.520006771433499e-05,
+      "loss": 0.5008,
+      "step": 16576
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.518941003440223e-05,
+      "loss": 0.3986,
+      "step": 16577
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.517875265476027e-05,
+      "loss": 0.4623,
+      "step": 16578
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.516809557553816e-05,
+      "loss": 0.5805,
+      "step": 16579
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.515743879686488e-05,
+      "loss": 0.4734,
+      "step": 16580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.514678231886938e-05,
+      "loss": 0.4332,
+      "step": 16581
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.513612614168069e-05,
+      "loss": 0.5544,
+      "step": 16582
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.512547026542771e-05,
+      "loss": 0.5762,
+      "step": 16583
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.51148146902395e-05,
+      "loss": 0.4298,
+      "step": 16584
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.510415941624498e-05,
+      "loss": 0.4511,
+      "step": 16585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.50935044435731e-05,
+      "loss": 0.5635,
+      "step": 16586
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.508284977235286e-05,
+      "loss": 0.3673,
+      "step": 16587
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.507219540271318e-05,
+      "loss": 0.5283,
+      "step": 16588
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.506154133478305e-05,
+      "loss": 0.5027,
+      "step": 16589
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.50508875686914e-05,
+      "loss": 0.4666,
+      "step": 16590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.504023410456715e-05,
+      "loss": 0.4966,
+      "step": 16591
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.50295809425393e-05,
+      "loss": 0.5682,
+      "step": 16592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.501892808273674e-05,
+      "loss": 0.5019,
+      "step": 16593
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.500827552528844e-05,
+      "loss": 0.5507,
+      "step": 16594
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.499762327032332e-05,
+      "loss": 0.5104,
+      "step": 16595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.498697131797027e-05,
+      "loss": 0.5372,
+      "step": 16596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.497631966835828e-05,
+      "loss": 0.5198,
+      "step": 16597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.496566832161623e-05,
+      "loss": 0.5638,
+      "step": 16598
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.495501727787302e-05,
+      "loss": 0.4908,
+      "step": 16599
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.494436653725761e-05,
+      "loss": 0.4951,
+      "step": 16600
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 0.5016784071922302,
+      "eval_runtime": 845.5774,
+      "eval_samples_per_second": 28.454,
+      "eval_steps_per_second": 14.227,
+      "step": 16600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.493371609989887e-05,
+      "loss": 0.5129,
+      "step": 16601
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.492306596592573e-05,
+      "loss": 0.397,
+      "step": 16602
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.49124161354671e-05,
+      "loss": 0.6718,
+      "step": 16603
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.490176660865185e-05,
+      "loss": 0.468,
+      "step": 16604
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.489111738560888e-05,
+      "loss": 0.495,
+      "step": 16605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.488046846646709e-05,
+      "loss": 0.4466,
+      "step": 16606
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.486981985135537e-05,
+      "loss": 0.4888,
+      "step": 16607
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.485917154040263e-05,
+      "loss": 0.5604,
+      "step": 16608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.484852353373767e-05,
+      "loss": 0.4948,
+      "step": 16609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.483787583148945e-05,
+      "loss": 0.5538,
+      "step": 16610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.482722843378677e-05,
+      "loss": 0.4846,
+      "step": 16611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.48165813407586e-05,
+      "loss": 0.4615,
+      "step": 16612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.480593455253371e-05,
+      "loss": 0.483,
+      "step": 16613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.479528806924099e-05,
+      "loss": 0.4946,
+      "step": 16614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.47846418910093e-05,
+      "loss": 0.5085,
+      "step": 16615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.477399601796752e-05,
+      "loss": 0.5663,
+      "step": 16616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.476335045024446e-05,
+      "loss": 0.5846,
+      "step": 16617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.475270518796899e-05,
+      "loss": 0.3846,
+      "step": 16618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.474206023126993e-05,
+      "loss": 0.6151,
+      "step": 16619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.473141558027617e-05,
+      "loss": 0.5836,
+      "step": 16620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.472077123511649e-05,
+      "loss": 0.5794,
+      "step": 16621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.471012719591974e-05,
+      "loss": 0.516,
+      "step": 16622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.469948346281477e-05,
+      "loss": 0.5499,
+      "step": 16623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.468884003593037e-05,
+      "loss": 0.524,
+      "step": 16624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.46781969153954e-05,
+      "loss": 0.4377,
+      "step": 16625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.466755410133866e-05,
+      "loss": 0.5418,
+      "step": 16626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.465691159388893e-05,
+      "loss": 0.5816,
+      "step": 16627
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.464626939317509e-05,
+      "loss": 0.5452,
+      "step": 16628
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.463562749932586e-05,
+      "loss": 0.4431,
+      "step": 16629
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.462498591247011e-05,
+      "loss": 0.4663,
+      "step": 16630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.461434463273664e-05,
+      "loss": 0.5492,
+      "step": 16631
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.460370366025419e-05,
+      "loss": 0.4791,
+      "step": 16632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.45930629951516e-05,
+      "loss": 0.555,
+      "step": 16633
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.458242263755764e-05,
+      "loss": 0.4596,
+      "step": 16634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.457178258760107e-05,
+      "loss": 0.6127,
+      "step": 16635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.456114284541073e-05,
+      "loss": 0.568,
+      "step": 16636
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.455050341111532e-05,
+      "loss": 0.5083,
+      "step": 16637
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.453986428484368e-05,
+      "loss": 0.4366,
+      "step": 16638
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.452922546672456e-05,
+      "loss": 0.5752,
+      "step": 16639
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.45185869568867e-05,
+      "loss": 0.6,
+      "step": 16640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.450794875545889e-05,
+      "loss": 0.5539,
+      "step": 16641
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.449731086256986e-05,
+      "loss": 0.5754,
+      "step": 16642
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.44866732783484e-05,
+      "loss": 0.62,
+      "step": 16643
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.447603600292323e-05,
+      "loss": 0.6322,
+      "step": 16644
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.446539903642311e-05,
+      "loss": 0.5371,
+      "step": 16645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.445476237897678e-05,
+      "loss": 0.5043,
+      "step": 16646
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.444412603071299e-05,
+      "loss": 0.5566,
+      "step": 16647
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.443348999176045e-05,
+      "loss": 0.5209,
+      "step": 16648
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.442285426224791e-05,
+      "loss": 0.5357,
+      "step": 16649
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.441221884230407e-05,
+      "loss": 0.5543,
+      "step": 16650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.440158373205771e-05,
+      "loss": 0.4981,
+      "step": 16651
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.439094893163752e-05,
+      "loss": 0.4195,
+      "step": 16652
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.43803144411722e-05,
+      "loss": 0.5423,
+      "step": 16653
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.436968026079048e-05,
+      "loss": 0.4641,
+      "step": 16654
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.435904639062105e-05,
+      "loss": 0.5169,
+      "step": 16655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.43484128307927e-05,
+      "loss": 0.4861,
+      "step": 16656
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4337779581434e-05,
+      "loss": 0.5642,
+      "step": 16657
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.432714664267373e-05,
+      "loss": 0.6082,
+      "step": 16658
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.431651401464054e-05,
+      "loss": 0.5046,
+      "step": 16659
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.430588169746317e-05,
+      "loss": 0.5061,
+      "step": 16660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.429524969127027e-05,
+      "loss": 0.5472,
+      "step": 16661
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.428461799619053e-05,
+      "loss": 0.4815,
+      "step": 16662
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.427398661235266e-05,
+      "loss": 0.511,
+      "step": 16663
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.426335553988526e-05,
+      "loss": 0.5648,
+      "step": 16664
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.425272477891709e-05,
+      "loss": 0.5993,
+      "step": 16665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.424209432957677e-05,
+      "loss": 0.5045,
+      "step": 16666
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.423146419199296e-05,
+      "loss": 0.4981,
+      "step": 16667
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.422083436629435e-05,
+      "loss": 0.4751,
+      "step": 16668
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.421020485260956e-05,
+      "loss": 0.5596,
+      "step": 16669
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.419957565106726e-05,
+      "loss": 0.6118,
+      "step": 16670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.41889467617961e-05,
+      "loss": 0.533,
+      "step": 16671
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.417831818492473e-05,
+      "loss": 0.4621,
+      "step": 16672
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.416768992058179e-05,
+      "loss": 0.5643,
+      "step": 16673
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.415706196889592e-05,
+      "loss": 0.3765,
+      "step": 16674
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.41464343299957e-05,
+      "loss": 0.5767,
+      "step": 16675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.413580700400986e-05,
+      "loss": 0.4564,
+      "step": 16676
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.412517999106694e-05,
+      "loss": 0.5076,
+      "step": 16677
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.411455329129562e-05,
+      "loss": 0.5378,
+      "step": 16678
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.410392690482448e-05,
+      "loss": 0.4892,
+      "step": 16679
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.409330083178214e-05,
+      "loss": 0.4971,
+      "step": 16680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.408267507229723e-05,
+      "loss": 0.5599,
+      "step": 16681
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.407204962649834e-05,
+      "loss": 0.5843,
+      "step": 16682
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.406142449451408e-05,
+      "loss": 0.5461,
+      "step": 16683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.405079967647305e-05,
+      "loss": 0.5984,
+      "step": 16684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.404017517250384e-05,
+      "loss": 0.6702,
+      "step": 16685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.402955098273506e-05,
+      "loss": 0.5129,
+      "step": 16686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.401892710729528e-05,
+      "loss": 0.5409,
+      "step": 16687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.400830354631306e-05,
+      "loss": 0.5583,
+      "step": 16688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.399768029991704e-05,
+      "loss": 0.505,
+      "step": 16689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.398705736823574e-05,
+      "loss": 0.4793,
+      "step": 16690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.397643475139779e-05,
+      "loss": 0.5091,
+      "step": 16691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.396581244953171e-05,
+      "loss": 0.5683,
+      "step": 16692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.395519046276609e-05,
+      "loss": 0.5315,
+      "step": 16693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.394456879122948e-05,
+      "loss": 0.429,
+      "step": 16694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.393394743505043e-05,
+      "loss": 0.618,
+      "step": 16695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.392332639435752e-05,
+      "loss": 0.519,
+      "step": 16696
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.39127056692793e-05,
+      "loss": 0.4729,
+      "step": 16697
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.390208525994427e-05,
+      "loss": 0.492,
+      "step": 16698
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.389146516648105e-05,
+      "loss": 0.5156,
+      "step": 16699
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.38808453890181e-05,
+      "loss": 0.4718,
+      "step": 16700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.387022592768397e-05,
+      "loss": 0.5296,
+      "step": 16701
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.385960678260725e-05,
+      "loss": 0.5506,
+      "step": 16702
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.384898795391639e-05,
+      "loss": 0.5031,
+      "step": 16703
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.383836944173997e-05,
+      "loss": 0.5236,
+      "step": 16704
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.38277512462065e-05,
+      "loss": 0.4506,
+      "step": 16705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.381713336744445e-05,
+      "loss": 0.568,
+      "step": 16706
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.38065158055824e-05,
+      "loss": 0.4992,
+      "step": 16707
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.379589856074879e-05,
+      "loss": 0.4471,
+      "step": 16708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.378528163307218e-05,
+      "loss": 0.4894,
+      "step": 16709
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.377466502268105e-05,
+      "loss": 0.5199,
+      "step": 16710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.376404872970387e-05,
+      "loss": 0.5343,
+      "step": 16711
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.375343275426918e-05,
+      "loss": 0.5369,
+      "step": 16712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.374281709650542e-05,
+      "loss": 0.4629,
+      "step": 16713
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.373220175654113e-05,
+      "loss": 0.5043,
+      "step": 16714
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.372158673450475e-05,
+      "loss": 0.4954,
+      "step": 16715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.371097203052477e-05,
+      "loss": 0.4661,
+      "step": 16716
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.370035764472966e-05,
+      "loss": 0.5433,
+      "step": 16717
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.368974357724789e-05,
+      "loss": 0.4519,
+      "step": 16718
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.367912982820793e-05,
+      "loss": 0.4034,
+      "step": 16719
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.366851639773822e-05,
+      "loss": 0.5036,
+      "step": 16720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.365790328596724e-05,
+      "loss": 0.4598,
+      "step": 16721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.364729049302347e-05,
+      "loss": 0.5359,
+      "step": 16722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.363667801903532e-05,
+      "loss": 0.483,
+      "step": 16723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.362606586413122e-05,
+      "loss": 0.4862,
+      "step": 16724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.361545402843967e-05,
+      "loss": 0.534,
+      "step": 16725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.360484251208906e-05,
+      "loss": 0.5443,
+      "step": 16726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.359423131520785e-05,
+      "loss": 0.5228,
+      "step": 16727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.358362043792446e-05,
+      "loss": 0.4698,
+      "step": 16728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.357300988036731e-05,
+      "loss": 0.5594,
+      "step": 16729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.356239964266485e-05,
+      "loss": 0.5776,
+      "step": 16730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.35517897249455e-05,
+      "loss": 0.3949,
+      "step": 16731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.354118012733761e-05,
+      "loss": 0.5099,
+      "step": 16732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.353057084996968e-05,
+      "loss": 0.6354,
+      "step": 16733
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.351996189297006e-05,
+      "loss": 0.5557,
+      "step": 16734
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.350935325646717e-05,
+      "loss": 0.5627,
+      "step": 16735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.349874494058943e-05,
+      "loss": 0.5058,
+      "step": 16736
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.348813694546519e-05,
+      "loss": 0.619,
+      "step": 16737
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.347752927122289e-05,
+      "loss": 0.4499,
+      "step": 16738
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.346692191799087e-05,
+      "loss": 0.5678,
+      "step": 16739
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.345631488589757e-05,
+      "loss": 0.5115,
+      "step": 16740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.344570817507133e-05,
+      "loss": 0.5675,
+      "step": 16741
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.343510178564053e-05,
+      "loss": 0.4752,
+      "step": 16742
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.342449571773358e-05,
+      "loss": 0.4856,
+      "step": 16743
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.341388997147878e-05,
+      "loss": 0.4273,
+      "step": 16744
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.340328454700456e-05,
+      "loss": 0.4537,
+      "step": 16745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.339267944443926e-05,
+      "loss": 0.5785,
+      "step": 16746
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.338207466391121e-05,
+      "loss": 0.6319,
+      "step": 16747
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.337147020554883e-05,
+      "loss": 0.5621,
+      "step": 16748
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.336086606948039e-05,
+      "loss": 0.5401,
+      "step": 16749
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.335026225583426e-05,
+      "loss": 0.4617,
+      "step": 16750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.333965876473881e-05,
+      "loss": 0.5117,
+      "step": 16751
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.332905559632235e-05,
+      "loss": 0.5123,
+      "step": 16752
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.331845275071325e-05,
+      "loss": 0.5551,
+      "step": 16753
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.33078502280398e-05,
+      "loss": 0.5051,
+      "step": 16754
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.329724802843033e-05,
+      "loss": 0.5269,
+      "step": 16755
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.328664615201318e-05,
+      "loss": 0.4408,
+      "step": 16756
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.327604459891664e-05,
+      "loss": 0.5562,
+      "step": 16757
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.326544336926906e-05,
+      "loss": 0.5231,
+      "step": 16758
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.325484246319873e-05,
+      "loss": 0.5018,
+      "step": 16759
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.324424188083396e-05,
+      "loss": 0.4578,
+      "step": 16760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.323364162230307e-05,
+      "loss": 0.5241,
+      "step": 16761
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.32230416877343e-05,
+      "loss": 0.5141,
+      "step": 16762
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.321244207725602e-05,
+      "loss": 0.4291,
+      "step": 16763
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.320184279099649e-05,
+      "loss": 0.5217,
+      "step": 16764
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.319124382908396e-05,
+      "loss": 0.4703,
+      "step": 16765
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.318064519164677e-05,
+      "loss": 0.4713,
+      "step": 16766
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.317004687881317e-05,
+      "loss": 0.4403,
+      "step": 16767
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.315944889071142e-05,
+      "loss": 0.4894,
+      "step": 16768
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.314885122746984e-05,
+      "loss": 0.4222,
+      "step": 16769
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.313825388921665e-05,
+      "loss": 0.5608,
+      "step": 16770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.312765687608015e-05,
+      "loss": 0.5377,
+      "step": 16771
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.311706018818855e-05,
+      "loss": 0.4652,
+      "step": 16772
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.310646382567015e-05,
+      "loss": 0.534,
+      "step": 16773
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.309586778865318e-05,
+      "loss": 0.5063,
+      "step": 16774
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.30852720772659e-05,
+      "loss": 0.4582,
+      "step": 16775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.307467669163655e-05,
+      "loss": 0.5793,
+      "step": 16776
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.306408163189336e-05,
+      "loss": 0.5295,
+      "step": 16777
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.305348689816456e-05,
+      "loss": 0.5611,
+      "step": 16778
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.304289249057841e-05,
+      "loss": 0.4592,
+      "step": 16779
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.303229840926311e-05,
+      "loss": 0.5049,
+      "step": 16780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.302170465434691e-05,
+      "loss": 0.5786,
+      "step": 16781
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3011111225958e-05,
+      "loss": 0.5341,
+      "step": 16782
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.300051812422462e-05,
+      "loss": 0.4559,
+      "step": 16783
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.298992534927498e-05,
+      "loss": 0.5879,
+      "step": 16784
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.297933290123726e-05,
+      "loss": 0.4274,
+      "step": 16785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.296874078023969e-05,
+      "loss": 0.5188,
+      "step": 16786
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.295814898641046e-05,
+      "loss": 0.5738,
+      "step": 16787
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.294755751987779e-05,
+      "loss": 0.4079,
+      "step": 16788
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.293696638076983e-05,
+      "loss": 0.5823,
+      "step": 16789
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.292637556921483e-05,
+      "loss": 0.5912,
+      "step": 16790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.291578508534091e-05,
+      "loss": 0.4873,
+      "step": 16791
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.290519492927626e-05,
+      "loss": 0.4713,
+      "step": 16792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.289460510114909e-05,
+      "loss": 0.4979,
+      "step": 16793
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.288401560108754e-05,
+      "loss": 0.4744,
+      "step": 16794
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.287342642921977e-05,
+      "loss": 0.5061,
+      "step": 16795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2862837585674e-05,
+      "loss": 0.4573,
+      "step": 16796
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.285224907057833e-05,
+      "loss": 0.5097,
+      "step": 16797
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.284166088406097e-05,
+      "loss": 0.4397,
+      "step": 16798
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.283107302625003e-05,
+      "loss": 0.495,
+      "step": 16799
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.282048549727365e-05,
+      "loss": 0.3621,
+      "step": 16800
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 0.49714845418930054,
+      "eval_runtime": 848.5565,
+      "eval_samples_per_second": 28.354,
+      "eval_steps_per_second": 14.177,
+      "step": 16800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.280989829726004e-05,
+      "loss": 0.5141,
+      "step": 16801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.279931142633727e-05,
+      "loss": 0.5644,
+      "step": 16802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.278872488463351e-05,
+      "loss": 0.4989,
+      "step": 16803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.277813867227688e-05,
+      "loss": 0.4426,
+      "step": 16804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.27675527893955e-05,
+      "loss": 0.5057,
+      "step": 16805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.275696723611754e-05,
+      "loss": 0.5825,
+      "step": 16806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.27463820125711e-05,
+      "loss": 0.604,
+      "step": 16807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.273579711888425e-05,
+      "loss": 0.4902,
+      "step": 16808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.272521255518517e-05,
+      "loss": 0.5088,
+      "step": 16809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.27146283216019e-05,
+      "loss": 0.5458,
+      "step": 16810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.270404441826261e-05,
+      "loss": 0.5292,
+      "step": 16811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.269346084529539e-05,
+      "loss": 0.5099,
+      "step": 16812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.268287760282827e-05,
+      "loss": 0.5083,
+      "step": 16813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.267229469098942e-05,
+      "loss": 0.5356,
+      "step": 16814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.266171210990688e-05,
+      "loss": 0.5219,
+      "step": 16815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.265112985970878e-05,
+      "loss": 0.4886,
+      "step": 16816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.264054794052318e-05,
+      "loss": 0.4739,
+      "step": 16817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.262996635247811e-05,
+      "loss": 0.5507,
+      "step": 16818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.261938509570172e-05,
+      "loss": 0.5845,
+      "step": 16819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.260880417032203e-05,
+      "loss": 0.6066,
+      "step": 16820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.259822357646711e-05,
+      "loss": 0.4153,
+      "step": 16821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.258764331426505e-05,
+      "loss": 0.5174,
+      "step": 16822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.257706338384385e-05,
+      "loss": 0.4647,
+      "step": 16823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.256648378533162e-05,
+      "loss": 0.526,
+      "step": 16824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.25559045188564e-05,
+      "loss": 0.4742,
+      "step": 16825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.254532558454617e-05,
+      "loss": 0.477,
+      "step": 16826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.253474698252906e-05,
+      "loss": 0.4929,
+      "step": 16827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.252416871293304e-05,
+      "loss": 0.5856,
+      "step": 16828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.25135907758862e-05,
+      "loss": 0.5137,
+      "step": 16829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.250301317151651e-05,
+      "loss": 0.4873,
+      "step": 16830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.249243589995202e-05,
+      "loss": 0.4626,
+      "step": 16831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.248185896132076e-05,
+      "loss": 0.435,
+      "step": 16832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.247128235575076e-05,
+      "loss": 0.5229,
+      "step": 16833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.246070608336996e-05,
+      "loss": 0.532,
+      "step": 16834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.245013014430645e-05,
+      "loss": 0.4595,
+      "step": 16835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.243955453868818e-05,
+      "loss": 0.5587,
+      "step": 16836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.242897926664319e-05,
+      "loss": 0.5687,
+      "step": 16837
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.241840432829946e-05,
+      "loss": 0.4849,
+      "step": 16838
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.240782972378496e-05,
+      "loss": 0.5524,
+      "step": 16839
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.239725545322772e-05,
+      "loss": 0.4732,
+      "step": 16840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.238668151675568e-05,
+      "loss": 0.4877,
+      "step": 16841
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.237610791449685e-05,
+      "loss": 0.5132,
+      "step": 16842
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.236553464657921e-05,
+      "loss": 0.5686,
+      "step": 16843
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.23549617131307e-05,
+      "loss": 0.4997,
+      "step": 16844
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.234438911427931e-05,
+      "loss": 0.4982,
+      "step": 16845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.233381685015299e-05,
+      "loss": 0.4752,
+      "step": 16846
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.232324492087974e-05,
+      "loss": 0.4828,
+      "step": 16847
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.231267332658747e-05,
+      "loss": 0.5219,
+      "step": 16848
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.230210206740414e-05,
+      "loss": 0.5775,
+      "step": 16849
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.229153114345772e-05,
+      "loss": 0.5225,
+      "step": 16850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.228096055487614e-05,
+      "loss": 0.4864,
+      "step": 16851
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.22703903017873e-05,
+      "loss": 0.457,
+      "step": 16852
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.225982038431922e-05,
+      "loss": 0.5453,
+      "step": 16853
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.224925080259975e-05,
+      "loss": 0.458,
+      "step": 16854
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.223868155675687e-05,
+      "loss": 0.4997,
+      "step": 16855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.222811264691849e-05,
+      "loss": 0.5314,
+      "step": 16856
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.22175440732125e-05,
+      "loss": 0.5466,
+      "step": 16857
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.220697583576685e-05,
+      "loss": 0.4982,
+      "step": 16858
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.219640793470943e-05,
+      "loss": 0.4948,
+      "step": 16859
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.218584037016818e-05,
+      "loss": 0.5244,
+      "step": 16860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.217527314227097e-05,
+      "loss": 0.4858,
+      "step": 16861
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.216470625114568e-05,
+      "loss": 0.6142,
+      "step": 16862
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.215413969692026e-05,
+      "loss": 0.7064,
+      "step": 16863
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.214357347972256e-05,
+      "loss": 0.5644,
+      "step": 16864
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.21330075996805e-05,
+      "loss": 0.5895,
+      "step": 16865
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.212244205692191e-05,
+      "loss": 0.5515,
+      "step": 16866
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.211187685157471e-05,
+      "loss": 0.579,
+      "step": 16867
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.210131198376678e-05,
+      "loss": 0.5202,
+      "step": 16868
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.209074745362597e-05,
+      "loss": 0.4088,
+      "step": 16869
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.208018326128012e-05,
+      "loss": 0.5199,
+      "step": 16870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.206961940685716e-05,
+      "loss": 0.4214,
+      "step": 16871
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.205905589048488e-05,
+      "loss": 0.5176,
+      "step": 16872
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.204849271229119e-05,
+      "loss": 0.5112,
+      "step": 16873
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.203792987240389e-05,
+      "loss": 0.446,
+      "step": 16874
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.202736737095086e-05,
+      "loss": 0.5376,
+      "step": 16875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.201680520805993e-05,
+      "loss": 0.4711,
+      "step": 16876
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.200624338385893e-05,
+      "loss": 0.4796,
+      "step": 16877
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.199568189847572e-05,
+      "loss": 0.5979,
+      "step": 16878
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.19851207520381e-05,
+      "loss": 0.4628,
+      "step": 16879
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.197455994467389e-05,
+      "loss": 0.5481,
+      "step": 16880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.196399947651094e-05,
+      "loss": 0.4489,
+      "step": 16881
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.195343934767705e-05,
+      "loss": 0.4175,
+      "step": 16882
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.194287955830004e-05,
+      "loss": 0.479,
+      "step": 16883
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.193232010850772e-05,
+      "loss": 0.4857,
+      "step": 16884
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.192176099842787e-05,
+      "loss": 0.5055,
+      "step": 16885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.191120222818832e-05,
+      "loss": 0.4865,
+      "step": 16886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.190064379791687e-05,
+      "loss": 0.5087,
+      "step": 16887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.189008570774127e-05,
+      "loss": 0.5241,
+      "step": 16888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.187952795778937e-05,
+      "loss": 0.5016,
+      "step": 16889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.186897054818888e-05,
+      "loss": 0.4732,
+      "step": 16890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.185841347906764e-05,
+      "loss": 0.5149,
+      "step": 16891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.184785675055341e-05,
+      "loss": 0.4762,
+      "step": 16892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.183730036277394e-05,
+      "loss": 0.4726,
+      "step": 16893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.182674431585704e-05,
+      "loss": 0.4723,
+      "step": 16894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.181618860993043e-05,
+      "loss": 0.6601,
+      "step": 16895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.180563324512189e-05,
+      "loss": 0.4955,
+      "step": 16896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.179507822155918e-05,
+      "loss": 0.4866,
+      "step": 16897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.178452353937002e-05,
+      "loss": 0.4919,
+      "step": 16898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.17739691986822e-05,
+      "loss": 0.5023,
+      "step": 16899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.176341519962345e-05,
+      "loss": 0.5407,
+      "step": 16900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.175286154232148e-05,
+      "loss": 0.4868,
+      "step": 16901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.174230822690406e-05,
+      "loss": 0.4919,
+      "step": 16902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.17317552534989e-05,
+      "loss": 0.4241,
+      "step": 16903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.172120262223374e-05,
+      "loss": 0.5169,
+      "step": 16904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.17106503332363e-05,
+      "loss": 0.4964,
+      "step": 16905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.170009838663426e-05,
+      "loss": 0.534,
+      "step": 16906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.16895467825554e-05,
+      "loss": 0.5384,
+      "step": 16907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.167899552112737e-05,
+      "loss": 0.5145,
+      "step": 16908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.166844460247793e-05,
+      "loss": 0.5539,
+      "step": 16909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.165789402673473e-05,
+      "loss": 0.5648,
+      "step": 16910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.164734379402549e-05,
+      "loss": 0.4935,
+      "step": 16911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.16367939044779e-05,
+      "loss": 0.5526,
+      "step": 16912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.162624435821965e-05,
+      "loss": 0.47,
+      "step": 16913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.161569515537845e-05,
+      "loss": 0.447,
+      "step": 16914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.160514629608195e-05,
+      "loss": 0.496,
+      "step": 16915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.15945977804578e-05,
+      "loss": 0.4529,
+      "step": 16916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.158404960863374e-05,
+      "loss": 0.5644,
+      "step": 16917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.157350178073739e-05,
+      "loss": 0.4359,
+      "step": 16918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.156295429689642e-05,
+      "loss": 0.5032,
+      "step": 16919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.155240715723852e-05,
+      "loss": 0.5791,
+      "step": 16920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.154186036189128e-05,
+      "loss": 0.4041,
+      "step": 16921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.153131391098242e-05,
+      "loss": 0.4965,
+      "step": 16922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.152076780463958e-05,
+      "loss": 0.5049,
+      "step": 16923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.151022204299034e-05,
+      "loss": 0.5779,
+      "step": 16924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.149967662616244e-05,
+      "loss": 0.4259,
+      "step": 16925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.148913155428342e-05,
+      "loss": 0.4608,
+      "step": 16926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.147858682748096e-05,
+      "loss": 0.4734,
+      "step": 16927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.146804244588267e-05,
+      "loss": 0.5259,
+      "step": 16928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.145749840961618e-05,
+      "loss": 0.5237,
+      "step": 16929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.144695471880908e-05,
+      "loss": 0.5837,
+      "step": 16930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.143641137358903e-05,
+      "loss": 0.5437,
+      "step": 16931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.142586837408362e-05,
+      "loss": 0.5254,
+      "step": 16932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.141532572042042e-05,
+      "loss": 0.4723,
+      "step": 16933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.14047834127271e-05,
+      "loss": 0.4455,
+      "step": 16934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.139424145113123e-05,
+      "loss": 0.4779,
+      "step": 16935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.138369983576035e-05,
+      "loss": 0.5315,
+      "step": 16936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.137315856674213e-05,
+      "loss": 0.4738,
+      "step": 16937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.136261764420409e-05,
+      "loss": 0.5019,
+      "step": 16938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.135207706827386e-05,
+      "loss": 0.5314,
+      "step": 16939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.134153683907899e-05,
+      "loss": 0.5109,
+      "step": 16940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.133099695674704e-05,
+      "loss": 0.5516,
+      "step": 16941
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.132045742140562e-05,
+      "loss": 0.5968,
+      "step": 16942
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.130991823318224e-05,
+      "loss": 0.4142,
+      "step": 16943
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.12993793922045e-05,
+      "loss": 0.5299,
+      "step": 16944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.128884089859995e-05,
+      "loss": 0.4931,
+      "step": 16945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.127830275249612e-05,
+      "loss": 0.52,
+      "step": 16946
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.126776495402057e-05,
+      "loss": 0.5275,
+      "step": 16947
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.125722750330084e-05,
+      "loss": 0.5395,
+      "step": 16948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.124669040046447e-05,
+      "loss": 0.564,
+      "step": 16949
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1236153645639e-05,
+      "loss": 0.5956,
+      "step": 16950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.122561723895195e-05,
+      "loss": 0.5007,
+      "step": 16951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.121508118053086e-05,
+      "loss": 0.4673,
+      "step": 16952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.120454547050324e-05,
+      "loss": 0.5329,
+      "step": 16953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.119401010899658e-05,
+      "loss": 0.5907,
+      "step": 16954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.118347509613846e-05,
+      "loss": 0.5904,
+      "step": 16955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.117294043205631e-05,
+      "loss": 0.5431,
+      "step": 16956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.11624061168777e-05,
+      "loss": 0.5917,
+      "step": 16957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.11518721507301e-05,
+      "loss": 0.5005,
+      "step": 16958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.114133853374101e-05,
+      "loss": 0.4774,
+      "step": 16959
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.113080526603792e-05,
+      "loss": 0.4891,
+      "step": 16960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.112027234774831e-05,
+      "loss": 0.5498,
+      "step": 16961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.11097397789997e-05,
+      "loss": 0.5681,
+      "step": 16962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.109920755991953e-05,
+      "loss": 0.5138,
+      "step": 16963
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.108867569063527e-05,
+      "loss": 0.5212,
+      "step": 16964
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.107814417127445e-05,
+      "loss": 0.505,
+      "step": 16965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.106761300196444e-05,
+      "loss": 0.5304,
+      "step": 16966
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.10570821828328e-05,
+      "loss": 0.5206,
+      "step": 16967
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.104655171400694e-05,
+      "loss": 0.49,
+      "step": 16968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.103602159561431e-05,
+      "loss": 0.5136,
+      "step": 16969
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.102549182778238e-05,
+      "loss": 0.4213,
+      "step": 16970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.101496241063858e-05,
+      "loss": 0.5657,
+      "step": 16971
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.100443334431033e-05,
+      "loss": 0.5218,
+      "step": 16972
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.099390462892512e-05,
+      "loss": 0.4855,
+      "step": 16973
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.098337626461034e-05,
+      "loss": 0.5384,
+      "step": 16974
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.097284825149344e-05,
+      "loss": 0.4567,
+      "step": 16975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.096232058970184e-05,
+      "loss": 0.6148,
+      "step": 16976
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.095179327936294e-05,
+      "loss": 0.4492,
+      "step": 16977
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.094126632060421e-05,
+      "loss": 0.4755,
+      "step": 16978
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.093073971355297e-05,
+      "loss": 0.5162,
+      "step": 16979
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.092021345833673e-05,
+      "loss": 0.5421,
+      "step": 16980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.090968755508282e-05,
+      "loss": 0.5389,
+      "step": 16981
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.089916200391865e-05,
+      "loss": 0.4865,
+      "step": 16982
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.088863680497165e-05,
+      "loss": 0.6022,
+      "step": 16983
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.087811195836918e-05,
+      "loss": 0.5767,
+      "step": 16984
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.08675874642386e-05,
+      "loss": 0.5849,
+      "step": 16985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.085706332270736e-05,
+      "loss": 0.4259,
+      "step": 16986
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.084653953390279e-05,
+      "loss": 0.4468,
+      "step": 16987
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.083601609795227e-05,
+      "loss": 0.5426,
+      "step": 16988
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.08254930149832e-05,
+      "loss": 0.4894,
+      "step": 16989
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.081497028512287e-05,
+      "loss": 0.543,
+      "step": 16990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.080444790849871e-05,
+      "loss": 0.4695,
+      "step": 16991
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.079392588523804e-05,
+      "loss": 0.4671,
+      "step": 16992
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.078340421546824e-05,
+      "loss": 0.5098,
+      "step": 16993
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.077288289931665e-05,
+      "loss": 0.4782,
+      "step": 16994
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.076236193691058e-05,
+      "loss": 0.5568,
+      "step": 16995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.07518413283774e-05,
+      "loss": 0.4976,
+      "step": 16996
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.074132107384442e-05,
+      "loss": 0.545,
+      "step": 16997
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.073080117343902e-05,
+      "loss": 0.5203,
+      "step": 16998
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.07202816272885e-05,
+      "loss": 0.5274,
+      "step": 16999
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.070976243552012e-05,
+      "loss": 0.5166,
+      "step": 17000
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 0.4935392439365387,
+      "eval_runtime": 845.9205,
+      "eval_samples_per_second": 28.442,
+      "eval_steps_per_second": 14.221,
+      "step": 17000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.06992435982613e-05,
+      "loss": 0.545,
+      "step": 17001
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.06887251156393e-05,
+      "loss": 0.4775,
+      "step": 17002
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.067820698778141e-05,
+      "loss": 0.5213,
+      "step": 17003
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.066768921481499e-05,
+      "loss": 0.5697,
+      "step": 17004
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.065717179686727e-05,
+      "loss": 0.5215,
+      "step": 17005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.064665473406562e-05,
+      "loss": 0.4366,
+      "step": 17006
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.063613802653728e-05,
+      "loss": 0.4996,
+      "step": 17007
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.062562167440955e-05,
+      "loss": 0.6087,
+      "step": 17008
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.061510567780972e-05,
+      "loss": 0.4865,
+      "step": 17009
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.060459003686506e-05,
+      "loss": 0.4488,
+      "step": 17010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.059407475170286e-05,
+      "loss": 0.4941,
+      "step": 17011
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.058355982245037e-05,
+      "loss": 0.4158,
+      "step": 17012
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.057304524923486e-05,
+      "loss": 0.481,
+      "step": 17013
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.05625310321836e-05,
+      "loss": 0.5728,
+      "step": 17014
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.055201717142384e-05,
+      "loss": 0.4604,
+      "step": 17015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.054150366708285e-05,
+      "loss": 0.4573,
+      "step": 17016
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.053099051928787e-05,
+      "loss": 0.4541,
+      "step": 17017
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.05204777281661e-05,
+      "loss": 0.4559,
+      "step": 17018
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.050996529384484e-05,
+      "loss": 0.4542,
+      "step": 17019
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.049945321645131e-05,
+      "loss": 0.5534,
+      "step": 17020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.048894149611271e-05,
+      "loss": 0.3839,
+      "step": 17021
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.047843013295632e-05,
+      "loss": 0.4798,
+      "step": 17022
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.046791912710933e-05,
+      "loss": 0.4572,
+      "step": 17023
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.045740847869895e-05,
+      "loss": 0.5264,
+      "step": 17024
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.044689818785243e-05,
+      "loss": 0.4057,
+      "step": 17025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.043638825469692e-05,
+      "loss": 0.5423,
+      "step": 17026
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.042587867935969e-05,
+      "loss": 0.4533,
+      "step": 17027
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.041536946196789e-05,
+      "loss": 0.4651,
+      "step": 17028
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.040486060264875e-05,
+      "loss": 0.4955,
+      "step": 17029
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.039435210152947e-05,
+      "loss": 0.435,
+      "step": 17030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.038384395873719e-05,
+      "loss": 0.4764,
+      "step": 17031
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.037333617439912e-05,
+      "loss": 0.4563,
+      "step": 17032
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.036282874864244e-05,
+      "loss": 0.5785,
+      "step": 17033
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.035232168159436e-05,
+      "loss": 0.545,
+      "step": 17034
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.034181497338199e-05,
+      "loss": 0.485,
+      "step": 17035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.033130862413252e-05,
+      "loss": 0.4736,
+      "step": 17036
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.032080263397312e-05,
+      "loss": 0.4716,
+      "step": 17037
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.031029700303096e-05,
+      "loss": 0.4635,
+      "step": 17038
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.029979173143315e-05,
+      "loss": 0.4828,
+      "step": 17039
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.028928681930689e-05,
+      "loss": 0.4345,
+      "step": 17040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.027878226677926e-05,
+      "loss": 0.4557,
+      "step": 17041
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.026827807397747e-05,
+      "loss": 0.5006,
+      "step": 17042
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.025777424102862e-05,
+      "loss": 0.4901,
+      "step": 17043
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.024727076805984e-05,
+      "loss": 0.4059,
+      "step": 17044
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.023676765519826e-05,
+      "loss": 0.5809,
+      "step": 17045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0226264902571e-05,
+      "loss": 0.4885,
+      "step": 17046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.02157625103052e-05,
+      "loss": 0.5178,
+      "step": 17047
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.020526047852797e-05,
+      "loss": 0.5831,
+      "step": 17048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.019475880736637e-05,
+      "loss": 0.5565,
+      "step": 17049
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.018425749694757e-05,
+      "loss": 0.4894,
+      "step": 17050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.017375654739866e-05,
+      "loss": 0.4486,
+      "step": 17051
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.016325595884667e-05,
+      "loss": 0.5315,
+      "step": 17052
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.015275573141877e-05,
+      "loss": 0.5249,
+      "step": 17053
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.014225586524202e-05,
+      "loss": 0.4611,
+      "step": 17054
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.01317563604435e-05,
+      "loss": 0.4887,
+      "step": 17055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.01212572171503e-05,
+      "loss": 0.483,
+      "step": 17056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.011075843548948e-05,
+      "loss": 0.4715,
+      "step": 17057
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.010026001558812e-05,
+      "loss": 0.5283,
+      "step": 17058
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.008976195757326e-05,
+      "loss": 0.6032,
+      "step": 17059
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.007926426157206e-05,
+      "loss": 0.4636,
+      "step": 17060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.006876692771144e-05,
+      "loss": 0.4955,
+      "step": 17061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.005826995611852e-05,
+      "loss": 0.5088,
+      "step": 17062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.004777334692034e-05,
+      "loss": 0.369,
+      "step": 17063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.003727710024396e-05,
+      "loss": 0.5968,
+      "step": 17064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.002678121621641e-05,
+      "loss": 0.5837,
+      "step": 17065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.00162856949647e-05,
+      "loss": 0.5052,
+      "step": 17066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.000579053661588e-05,
+      "loss": 0.5047,
+      "step": 17067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9995295741297e-05,
+      "loss": 0.4848,
+      "step": 17068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.998480130913505e-05,
+      "loss": 0.4672,
+      "step": 17069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.997430724025706e-05,
+      "loss": 0.4874,
+      "step": 17070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.996381353479004e-05,
+      "loss": 0.5377,
+      "step": 17071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.995332019286101e-05,
+      "loss": 0.5008,
+      "step": 17072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.994282721459695e-05,
+      "loss": 0.554,
+      "step": 17073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.993233460012488e-05,
+      "loss": 0.4928,
+      "step": 17074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.992184234957179e-05,
+      "loss": 0.4433,
+      "step": 17075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.991135046306467e-05,
+      "loss": 0.4894,
+      "step": 17076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.990085894073052e-05,
+      "loss": 0.5539,
+      "step": 17077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.98903677826963e-05,
+      "loss": 0.4271,
+      "step": 17078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9879876989089e-05,
+      "loss": 0.532,
+      "step": 17079
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.986938656003559e-05,
+      "loss": 0.4944,
+      "step": 17080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.985889649566305e-05,
+      "loss": 0.4475,
+      "step": 17081
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.984840679609833e-05,
+      "loss": 0.4624,
+      "step": 17082
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.98379174614684e-05,
+      "loss": 0.4477,
+      "step": 17083
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.98274284919002e-05,
+      "loss": 0.4783,
+      "step": 17084
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.981693988752071e-05,
+      "loss": 0.4749,
+      "step": 17085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.980645164845688e-05,
+      "loss": 0.4978,
+      "step": 17086
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.979596377483558e-05,
+      "loss": 0.5006,
+      "step": 17087
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.978547626678385e-05,
+      "loss": 0.4946,
+      "step": 17088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.977498912442852e-05,
+      "loss": 0.5296,
+      "step": 17089
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.976450234789663e-05,
+      "loss": 0.461,
+      "step": 17090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.975401593731505e-05,
+      "loss": 0.5445,
+      "step": 17091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.974352989281066e-05,
+      "loss": 0.5459,
+      "step": 17092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.973304421451046e-05,
+      "loss": 0.4642,
+      "step": 17093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.972255890254127e-05,
+      "loss": 0.4715,
+      "step": 17094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.971207395703007e-05,
+      "loss": 0.5493,
+      "step": 17095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.970158937810376e-05,
+      "loss": 0.5081,
+      "step": 17096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.969110516588919e-05,
+      "loss": 0.5148,
+      "step": 17097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.968062132051329e-05,
+      "loss": 0.533,
+      "step": 17098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.967013784210293e-05,
+      "loss": 0.4459,
+      "step": 17099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.965965473078503e-05,
+      "loss": 0.3342,
+      "step": 17100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.964917198668643e-05,
+      "loss": 0.4536,
+      "step": 17101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.963868960993402e-05,
+      "loss": 0.4886,
+      "step": 17102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.962820760065468e-05,
+      "loss": 0.4824,
+      "step": 17103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.961772595897526e-05,
+      "loss": 0.4583,
+      "step": 17104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.960724468502265e-05,
+      "loss": 0.5028,
+      "step": 17105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.959676377892368e-05,
+      "loss": 0.4521,
+      "step": 17106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.958628324080521e-05,
+      "loss": 0.3703,
+      "step": 17107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.95758030707941e-05,
+      "loss": 0.5132,
+      "step": 17108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.956532326901722e-05,
+      "loss": 0.6443,
+      "step": 17109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.955484383560133e-05,
+      "loss": 0.4938,
+      "step": 17110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.954436477067335e-05,
+      "loss": 0.6161,
+      "step": 17111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.953388607436007e-05,
+      "loss": 0.5215,
+      "step": 17112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.952340774678831e-05,
+      "loss": 0.3938,
+      "step": 17113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.951292978808495e-05,
+      "loss": 0.6059,
+      "step": 17114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.950245219837673e-05,
+      "loss": 0.5393,
+      "step": 17115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.949197497779052e-05,
+      "loss": 0.4522,
+      "step": 17116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.948149812645308e-05,
+      "loss": 0.4511,
+      "step": 17117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.947102164449128e-05,
+      "loss": 0.4569,
+      "step": 17118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.946054553203188e-05,
+      "loss": 0.5029,
+      "step": 17119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.945006978920166e-05,
+      "loss": 0.6552,
+      "step": 17120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.943959441612745e-05,
+      "loss": 0.504,
+      "step": 17121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.942911941293602e-05,
+      "loss": 0.5337,
+      "step": 17122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.941864477975413e-05,
+      "loss": 0.637,
+      "step": 17123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.94081705167086e-05,
+      "loss": 0.4357,
+      "step": 17124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.939769662392617e-05,
+      "loss": 0.42,
+      "step": 17125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.938722310153364e-05,
+      "loss": 0.4567,
+      "step": 17126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.937674994965775e-05,
+      "loss": 0.5507,
+      "step": 17127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.936627716842526e-05,
+      "loss": 0.4675,
+      "step": 17128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.935580475796292e-05,
+      "loss": 0.5641,
+      "step": 17129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.934533271839752e-05,
+      "loss": 0.5403,
+      "step": 17130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.933486104985578e-05,
+      "loss": 0.4406,
+      "step": 17131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.932438975246442e-05,
+      "loss": 0.5332,
+      "step": 17132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.931391882635022e-05,
+      "loss": 0.5099,
+      "step": 17133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.930344827163988e-05,
+      "loss": 0.4395,
+      "step": 17134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.929297808846015e-05,
+      "loss": 0.4493,
+      "step": 17135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.928250827693771e-05,
+      "loss": 0.4329,
+      "step": 17136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.927203883719938e-05,
+      "loss": 0.4803,
+      "step": 17137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.926156976937175e-05,
+      "loss": 0.4687,
+      "step": 17138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.925110107358161e-05,
+      "loss": 0.4644,
+      "step": 17139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.924063274995567e-05,
+      "loss": 0.5265,
+      "step": 17140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.923016479862059e-05,
+      "loss": 0.466,
+      "step": 17141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.921969721970309e-05,
+      "loss": 0.5031,
+      "step": 17142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.920923001332984e-05,
+      "loss": 0.4094,
+      "step": 17143
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.919876317962756e-05,
+      "loss": 0.5134,
+      "step": 17144
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.918829671872293e-05,
+      "loss": 0.4925,
+      "step": 17145
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.91778306307426e-05,
+      "loss": 0.5247,
+      "step": 17146
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.916736491581328e-05,
+      "loss": 0.5103,
+      "step": 17147
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.915689957406162e-05,
+      "loss": 0.4375,
+      "step": 17148
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.914643460561428e-05,
+      "loss": 0.4678,
+      "step": 17149
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.913597001059795e-05,
+      "loss": 0.5463,
+      "step": 17150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.912550578913924e-05,
+      "loss": 0.5062,
+      "step": 17151
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.911504194136484e-05,
+      "loss": 0.4904,
+      "step": 17152
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.91045784674014e-05,
+      "loss": 0.413,
+      "step": 17153
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.909411536737554e-05,
+      "loss": 0.5011,
+      "step": 17154
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.90836526414139e-05,
+      "loss": 0.5774,
+      "step": 17155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.907319028964311e-05,
+      "loss": 0.5198,
+      "step": 17156
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.906272831218982e-05,
+      "loss": 0.4884,
+      "step": 17157
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.905226670918068e-05,
+      "loss": 0.5192,
+      "step": 17158
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.904180548074222e-05,
+      "loss": 0.4965,
+      "step": 17159
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.903134462700115e-05,
+      "loss": 0.5957,
+      "step": 17160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.902088414808402e-05,
+      "loss": 0.4389,
+      "step": 17161
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.901042404411748e-05,
+      "loss": 0.462,
+      "step": 17162
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.89999643152281e-05,
+      "loss": 0.471,
+      "step": 17163
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.898950496154248e-05,
+      "loss": 0.4865,
+      "step": 17164
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.897904598318725e-05,
+      "loss": 0.5533,
+      "step": 17165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.896858738028893e-05,
+      "loss": 0.4963,
+      "step": 17166
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.895812915297418e-05,
+      "loss": 0.4611,
+      "step": 17167
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.894767130136953e-05,
+      "loss": 0.5078,
+      "step": 17168
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.893721382560156e-05,
+      "loss": 0.5708,
+      "step": 17169
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.892675672579686e-05,
+      "loss": 0.4921,
+      "step": 17170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.891630000208199e-05,
+      "loss": 0.507,
+      "step": 17171
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.890584365458348e-05,
+      "loss": 0.4659,
+      "step": 17172
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.889538768342795e-05,
+      "loss": 0.5512,
+      "step": 17173
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.888493208874187e-05,
+      "loss": 0.4625,
+      "step": 17174
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.887447687065187e-05,
+      "loss": 0.4522,
+      "step": 17175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.886402202928445e-05,
+      "loss": 0.4857,
+      "step": 17176
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.885356756476613e-05,
+      "loss": 0.5329,
+      "step": 17177
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.88431134772235e-05,
+      "loss": 0.5375,
+      "step": 17178
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.883265976678303e-05,
+      "loss": 0.4463,
+      "step": 17179
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.882220643357129e-05,
+      "loss": 0.5166,
+      "step": 17180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.881175347771478e-05,
+      "loss": 0.5055,
+      "step": 17181
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.880130089934e-05,
+      "loss": 0.5256,
+      "step": 17182
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.879084869857349e-05,
+      "loss": 0.4944,
+      "step": 17183
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.878039687554174e-05,
+      "loss": 0.4822,
+      "step": 17184
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.876994543037127e-05,
+      "loss": 0.4978,
+      "step": 17185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.875949436318855e-05,
+      "loss": 0.6368,
+      "step": 17186
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.87490436741201e-05,
+      "loss": 0.4951,
+      "step": 17187
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.873859336329238e-05,
+      "loss": 0.4421,
+      "step": 17188
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.872814343083192e-05,
+      "loss": 0.4837,
+      "step": 17189
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.871769387686512e-05,
+      "loss": 0.5631,
+      "step": 17190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.870724470151853e-05,
+      "loss": 0.4616,
+      "step": 17191
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.869679590491857e-05,
+      "loss": 0.5551,
+      "step": 17192
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.868634748719176e-05,
+      "loss": 0.4971,
+      "step": 17193
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.867589944846454e-05,
+      "loss": 0.5434,
+      "step": 17194
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.866545178886334e-05,
+      "loss": 0.5161,
+      "step": 17195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.86550045085146e-05,
+      "loss": 0.4774,
+      "step": 17196
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.864455760754482e-05,
+      "loss": 0.5092,
+      "step": 17197
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.86341110860804e-05,
+      "loss": 0.5641,
+      "step": 17198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.862366494424777e-05,
+      "loss": 0.5309,
+      "step": 17199
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.861321918217343e-05,
+      "loss": 0.5464,
+      "step": 17200
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 0.48956164717674255,
+      "eval_runtime": 851.6347,
+      "eval_samples_per_second": 28.252,
+      "eval_steps_per_second": 14.126,
+      "step": 17200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.860277379998372e-05,
+      "loss": 0.3965,
+      "step": 17201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.859232879780515e-05,
+      "loss": 0.5023,
+      "step": 17202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.858188417576407e-05,
+      "loss": 0.5849,
+      "step": 17203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.857143993398692e-05,
+      "loss": 0.5389,
+      "step": 17204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.856099607260014e-05,
+      "loss": 0.5322,
+      "step": 17205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.855055259173009e-05,
+      "loss": 0.5427,
+      "step": 17206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.854010949150318e-05,
+      "loss": 0.5303,
+      "step": 17207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.852966677204583e-05,
+      "loss": 0.5795,
+      "step": 17208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.85192244334844e-05,
+      "loss": 0.4408,
+      "step": 17209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.850878247594531e-05,
+      "loss": 0.46,
+      "step": 17210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.849834089955493e-05,
+      "loss": 0.4702,
+      "step": 17211
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.848789970443961e-05,
+      "loss": 0.4802,
+      "step": 17212
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.847745889072577e-05,
+      "loss": 0.5089,
+      "step": 17213
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.846701845853973e-05,
+      "loss": 0.4817,
+      "step": 17214
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.845657840800791e-05,
+      "loss": 0.5135,
+      "step": 17215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.844613873925664e-05,
+      "loss": 0.5349,
+      "step": 17216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.843569945241227e-05,
+      "loss": 0.547,
+      "step": 17217
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.842526054760115e-05,
+      "loss": 0.4687,
+      "step": 17218
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.84148220249496e-05,
+      "loss": 0.4607,
+      "step": 17219
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.840438388458404e-05,
+      "loss": 0.5732,
+      "step": 17220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.839394612663074e-05,
+      "loss": 0.557,
+      "step": 17221
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.838350875121603e-05,
+      "loss": 0.44,
+      "step": 17222
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.837307175846628e-05,
+      "loss": 0.4846,
+      "step": 17223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.836263514850778e-05,
+      "loss": 0.5033,
+      "step": 17224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.835219892146685e-05,
+      "loss": 0.4701,
+      "step": 17225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.834176307746983e-05,
+      "loss": 0.4985,
+      "step": 17226
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.833132761664298e-05,
+      "loss": 0.563,
+      "step": 17227
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.832089253911264e-05,
+      "loss": 0.5124,
+      "step": 17228
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.831045784500513e-05,
+      "loss": 0.5089,
+      "step": 17229
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.830002353444668e-05,
+      "loss": 0.5147,
+      "step": 17230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.828958960756363e-05,
+      "loss": 0.5622,
+      "step": 17231
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.827915606448223e-05,
+      "loss": 0.4344,
+      "step": 17232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.826872290532882e-05,
+      "loss": 0.5174,
+      "step": 17233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.825829013022962e-05,
+      "loss": 0.5047,
+      "step": 17234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.82478577393109e-05,
+      "loss": 0.4297,
+      "step": 17235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.823742573269898e-05,
+      "loss": 0.5584,
+      "step": 17236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.822699411052007e-05,
+      "loss": 0.5372,
+      "step": 17237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.821656287290044e-05,
+      "loss": 0.5377,
+      "step": 17238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.820613201996635e-05,
+      "loss": 0.4606,
+      "step": 17239
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.819570155184405e-05,
+      "loss": 0.5779,
+      "step": 17240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.818527146865978e-05,
+      "loss": 0.4613,
+      "step": 17241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.817484177053979e-05,
+      "loss": 0.4406,
+      "step": 17242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.816441245761027e-05,
+      "loss": 0.6017,
+      "step": 17243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.81539835299975e-05,
+      "loss": 0.5759,
+      "step": 17244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.814355498782767e-05,
+      "loss": 0.4946,
+      "step": 17245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.813312683122705e-05,
+      "loss": 0.5863,
+      "step": 17246
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.81226990603218e-05,
+      "loss": 0.5976,
+      "step": 17247
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.811227167523815e-05,
+      "loss": 0.4649,
+      "step": 17248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.810184467610233e-05,
+      "loss": 0.4796,
+      "step": 17249
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.80914180630405e-05,
+      "loss": 0.549,
+      "step": 17250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.808099183617889e-05,
+      "loss": 0.4773,
+      "step": 17251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.807056599564368e-05,
+      "loss": 0.4636,
+      "step": 17252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.806014054156102e-05,
+      "loss": 0.4801,
+      "step": 17253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.804971547405717e-05,
+      "loss": 0.4562,
+      "step": 17254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.803929079325826e-05,
+      "loss": 0.5118,
+      "step": 17255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.802886649929046e-05,
+      "loss": 0.4798,
+      "step": 17256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.801844259227996e-05,
+      "loss": 0.4817,
+      "step": 17257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.80080190723529e-05,
+      "loss": 0.4849,
+      "step": 17258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.799759593963547e-05,
+      "loss": 0.5106,
+      "step": 17259
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.79871731942538e-05,
+      "loss": 0.4701,
+      "step": 17260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.797675083633405e-05,
+      "loss": 0.5051,
+      "step": 17261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.796632886600236e-05,
+      "loss": 0.5541,
+      "step": 17262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.795590728338485e-05,
+      "loss": 0.4685,
+      "step": 17263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.794548608860772e-05,
+      "loss": 0.5179,
+      "step": 17264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.793506528179705e-05,
+      "loss": 0.5533,
+      "step": 17265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.792464486307897e-05,
+      "loss": 0.5447,
+      "step": 17266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.791422483257961e-05,
+      "loss": 0.4929,
+      "step": 17267
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.790380519042507e-05,
+      "loss": 0.4908,
+      "step": 17268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.78933859367415e-05,
+      "loss": 0.4671,
+      "step": 17269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.788296707165499e-05,
+      "loss": 0.5133,
+      "step": 17270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.787254859529162e-05,
+      "loss": 0.5629,
+      "step": 17271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.786213050777753e-05,
+      "loss": 0.4936,
+      "step": 17272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.785171280923877e-05,
+      "loss": 0.488,
+      "step": 17273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.784129549980144e-05,
+      "loss": 0.5227,
+      "step": 17274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.783087857959167e-05,
+      "loss": 0.4693,
+      "step": 17275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.782046204873547e-05,
+      "loss": 0.5478,
+      "step": 17276
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.781004590735897e-05,
+      "loss": 0.4521,
+      "step": 17277
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.779963015558821e-05,
+      "loss": 0.5736,
+      "step": 17278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.778921479354925e-05,
+      "loss": 0.5165,
+      "step": 17279
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.77787998213682e-05,
+      "loss": 0.5038,
+      "step": 17280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.776838523917104e-05,
+      "loss": 0.4726,
+      "step": 17281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.775797104708388e-05,
+      "loss": 0.572,
+      "step": 17282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.774755724523275e-05,
+      "loss": 0.5318,
+      "step": 17283
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.773714383374368e-05,
+      "loss": 0.4937,
+      "step": 17284
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.772673081274273e-05,
+      "loss": 0.4744,
+      "step": 17285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.771631818235588e-05,
+      "loss": 0.506,
+      "step": 17286
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.770590594270924e-05,
+      "loss": 0.4567,
+      "step": 17287
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.769549409392878e-05,
+      "loss": 0.5671,
+      "step": 17288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.768508263614051e-05,
+      "loss": 0.6048,
+      "step": 17289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.767467156947049e-05,
+      "loss": 0.4504,
+      "step": 17290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.766426089404467e-05,
+      "loss": 0.5362,
+      "step": 17291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.765385060998909e-05,
+      "loss": 0.5045,
+      "step": 17292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.764344071742976e-05,
+      "loss": 0.5368,
+      "step": 17293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.763303121649263e-05,
+      "loss": 0.4531,
+      "step": 17294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.762262210730373e-05,
+      "loss": 0.506,
+      "step": 17295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.761221338998903e-05,
+      "loss": 0.4814,
+      "step": 17296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.76018050646745e-05,
+      "loss": 0.4914,
+      "step": 17297
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.759139713148614e-05,
+      "loss": 0.5439,
+      "step": 17298
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.75809895905499e-05,
+      "loss": 0.5475,
+      "step": 17299
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.757058244199176e-05,
+      "loss": 0.506,
+      "step": 17300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.756017568593768e-05,
+      "loss": 0.5035,
+      "step": 17301
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.754976932251359e-05,
+      "loss": 0.4604,
+      "step": 17302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.753936335184547e-05,
+      "loss": 0.4885,
+      "step": 17303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.752895777405928e-05,
+      "loss": 0.4915,
+      "step": 17304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.75185525892809e-05,
+      "loss": 0.5307,
+      "step": 17305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.750814779763634e-05,
+      "loss": 0.5144,
+      "step": 17306
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.74977433992515e-05,
+      "loss": 0.5479,
+      "step": 17307
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.74873393942523e-05,
+      "loss": 0.4657,
+      "step": 17308
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.747693578276469e-05,
+      "loss": 0.5135,
+      "step": 17309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.746653256491454e-05,
+      "loss": 0.5199,
+      "step": 17310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.745612974082782e-05,
+      "loss": 0.4941,
+      "step": 17311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.744572731063038e-05,
+      "loss": 0.4499,
+      "step": 17312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.74353252744482e-05,
+      "loss": 0.4369,
+      "step": 17313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.742492363240714e-05,
+      "loss": 0.4325,
+      "step": 17314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.741452238463306e-05,
+      "loss": 0.4361,
+      "step": 17315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.74041215312519e-05,
+      "loss": 0.438,
+      "step": 17316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.73937210723895e-05,
+      "loss": 0.5193,
+      "step": 17317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.73833210081718e-05,
+      "loss": 0.5721,
+      "step": 17318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.737292133872464e-05,
+      "loss": 0.4263,
+      "step": 17319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.736252206417386e-05,
+      "loss": 0.5868,
+      "step": 17320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.735212318464539e-05,
+      "loss": 0.5096,
+      "step": 17321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.734172470026505e-05,
+      "loss": 0.5199,
+      "step": 17322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.73313266111587e-05,
+      "loss": 0.4618,
+      "step": 17323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.732092891745221e-05,
+      "loss": 0.449,
+      "step": 17324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.73105316192714e-05,
+      "loss": 0.5136,
+      "step": 17325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.730013471674213e-05,
+      "loss": 0.4747,
+      "step": 17326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.728973820999023e-05,
+      "loss": 0.3874,
+      "step": 17327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.727934209914153e-05,
+      "loss": 0.4592,
+      "step": 17328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.726894638432191e-05,
+      "loss": 0.4874,
+      "step": 17329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.725855106565712e-05,
+      "loss": 0.3892,
+      "step": 17330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7248156143273e-05,
+      "loss": 0.478,
+      "step": 17331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.723776161729535e-05,
+      "loss": 0.5381,
+      "step": 17332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.722736748785002e-05,
+      "loss": 0.624,
+      "step": 17333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.721697375506275e-05,
+      "loss": 0.4896,
+      "step": 17334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.720658041905942e-05,
+      "loss": 0.5452,
+      "step": 17335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.719618747996577e-05,
+      "loss": 0.6545,
+      "step": 17336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.718579493790759e-05,
+      "loss": 0.444,
+      "step": 17337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.717540279301069e-05,
+      "loss": 0.4631,
+      "step": 17338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.716501104540084e-05,
+      "loss": 0.6553,
+      "step": 17339
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.715461969520378e-05,
+      "loss": 0.5549,
+      "step": 17340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.714422874254534e-05,
+      "loss": 0.5561,
+      "step": 17341
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.713383818755124e-05,
+      "loss": 0.6164,
+      "step": 17342
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.712344803034727e-05,
+      "loss": 0.5816,
+      "step": 17343
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.711305827105915e-05,
+      "loss": 0.5085,
+      "step": 17344
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.710266890981265e-05,
+      "loss": 0.4585,
+      "step": 17345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.709227994673355e-05,
+      "loss": 0.4991,
+      "step": 17346
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.708189138194753e-05,
+      "loss": 0.5457,
+      "step": 17347
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.707150321558038e-05,
+      "loss": 0.52,
+      "step": 17348
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.70611154477578e-05,
+      "loss": 0.596,
+      "step": 17349
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.705072807860549e-05,
+      "loss": 0.4189,
+      "step": 17350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.704034110824924e-05,
+      "loss": 0.4657,
+      "step": 17351
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.702995453681471e-05,
+      "loss": 0.5015,
+      "step": 17352
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.701956836442766e-05,
+      "loss": 0.4575,
+      "step": 17353
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.700918259121376e-05,
+      "loss": 0.5229,
+      "step": 17354
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.699879721729871e-05,
+      "loss": 0.4949,
+      "step": 17355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.698841224280824e-05,
+      "loss": 0.4096,
+      "step": 17356
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.697802766786803e-05,
+      "loss": 0.5168,
+      "step": 17357
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.696764349260374e-05,
+      "loss": 0.4809,
+      "step": 17358
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.695725971714108e-05,
+      "loss": 0.5089,
+      "step": 17359
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.694687634160572e-05,
+      "loss": 0.5031,
+      "step": 17360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.693649336612336e-05,
+      "loss": 0.4705,
+      "step": 17361
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.692611079081962e-05,
+      "loss": 0.5339,
+      "step": 17362
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.69157286158202e-05,
+      "loss": 0.4615,
+      "step": 17363
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.690534684125076e-05,
+      "loss": 0.4454,
+      "step": 17364
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.689496546723692e-05,
+      "loss": 0.4519,
+      "step": 17365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.688458449390437e-05,
+      "loss": 0.5072,
+      "step": 17366
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.687420392137874e-05,
+      "loss": 0.4292,
+      "step": 17367
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.686382374978564e-05,
+      "loss": 0.5379,
+      "step": 17368
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.685344397925076e-05,
+      "loss": 0.5183,
+      "step": 17369
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.684306460989968e-05,
+      "loss": 0.5442,
+      "step": 17370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.683268564185805e-05,
+      "loss": 0.4785,
+      "step": 17371
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.68223070752515e-05,
+      "loss": 0.5431,
+      "step": 17372
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.68119289102056e-05,
+      "loss": 0.4471,
+      "step": 17373
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.680155114684602e-05,
+      "loss": 0.4706,
+      "step": 17374
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.679117378529833e-05,
+      "loss": 0.5472,
+      "step": 17375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.678079682568812e-05,
+      "loss": 0.4694,
+      "step": 17376
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.677042026814102e-05,
+      "loss": 0.6059,
+      "step": 17377
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.676004411278259e-05,
+      "loss": 0.5296,
+      "step": 17378
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.674966835973847e-05,
+      "loss": 0.4398,
+      "step": 17379
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.673929300913417e-05,
+      "loss": 0.4989,
+      "step": 17380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.672891806109527e-05,
+      "loss": 0.5155,
+      "step": 17381
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.671854351574742e-05,
+      "loss": 0.4303,
+      "step": 17382
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.67081693732161e-05,
+      "loss": 0.4832,
+      "step": 17383
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.669779563362692e-05,
+      "loss": 0.4709,
+      "step": 17384
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.668742229710543e-05,
+      "loss": 0.4475,
+      "step": 17385
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.667704936377716e-05,
+      "loss": 0.5378,
+      "step": 17386
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.666667683376768e-05,
+      "loss": 0.522,
+      "step": 17387
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.665630470720254e-05,
+      "loss": 0.5436,
+      "step": 17388
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.664593298420725e-05,
+      "loss": 0.4187,
+      "step": 17389
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.663556166490735e-05,
+      "loss": 0.5603,
+      "step": 17390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.662519074942837e-05,
+      "loss": 0.4861,
+      "step": 17391
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.661482023789584e-05,
+      "loss": 0.4964,
+      "step": 17392
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.66044501304353e-05,
+      "loss": 0.5012,
+      "step": 17393
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.659408042717219e-05,
+      "loss": 0.5247,
+      "step": 17394
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.65837111282321e-05,
+      "loss": 0.5477,
+      "step": 17395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.657334223374048e-05,
+      "loss": 0.5354,
+      "step": 17396
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.656297374382286e-05,
+      "loss": 0.4139,
+      "step": 17397
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.65526056586047e-05,
+      "loss": 0.4731,
+      "step": 17398
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.65422379782115e-05,
+      "loss": 0.5429,
+      "step": 17399
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.653187070276877e-05,
+      "loss": 0.5093,
+      "step": 17400
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 0.4857846200466156,
+      "eval_runtime": 850.6335,
+      "eval_samples_per_second": 28.285,
+      "eval_steps_per_second": 14.142,
+      "step": 17400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.652150383240196e-05,
+      "loss": 0.4686,
+      "step": 17401
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.651113736723655e-05,
+      "loss": 0.4352,
+      "step": 17402
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.650077130739803e-05,
+      "loss": 0.5403,
+      "step": 17403
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.649040565301181e-05,
+      "loss": 0.5128,
+      "step": 17404
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.648004040420341e-05,
+      "loss": 0.4155,
+      "step": 17405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.646967556109826e-05,
+      "loss": 0.4982,
+      "step": 17406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.64593111238218e-05,
+      "loss": 0.4056,
+      "step": 17407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.644894709249949e-05,
+      "loss": 0.594,
+      "step": 17408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.643858346725674e-05,
+      "loss": 0.522,
+      "step": 17409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.642822024821902e-05,
+      "loss": 0.5131,
+      "step": 17410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.641785743551176e-05,
+      "loss": 0.6002,
+      "step": 17411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.640749502926033e-05,
+      "loss": 0.5761,
+      "step": 17412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.639713302959023e-05,
+      "loss": 0.544,
+      "step": 17413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.63867714366268e-05,
+      "loss": 0.4913,
+      "step": 17414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.637641025049551e-05,
+      "loss": 0.5037,
+      "step": 17415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.636604947132173e-05,
+      "loss": 0.5587,
+      "step": 17416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.635568909923087e-05,
+      "loss": 0.4355,
+      "step": 17417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.634532913434832e-05,
+      "loss": 0.5238,
+      "step": 17418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.633496957679946e-05,
+      "loss": 0.4425,
+      "step": 17419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.63246104267097e-05,
+      "loss": 0.6084,
+      "step": 17420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.631425168420443e-05,
+      "loss": 0.4731,
+      "step": 17421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.630389334940898e-05,
+      "loss": 0.4933,
+      "step": 17422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.629353542244875e-05,
+      "loss": 0.4743,
+      "step": 17423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.628317790344911e-05,
+      "loss": 0.4232,
+      "step": 17424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.627282079253538e-05,
+      "loss": 0.4942,
+      "step": 17425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6262464089833e-05,
+      "loss": 0.4914,
+      "step": 17426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.62521077954672e-05,
+      "loss": 0.4612,
+      "step": 17427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.624175190956344e-05,
+      "loss": 0.4867,
+      "step": 17428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.623139643224701e-05,
+      "loss": 0.3693,
+      "step": 17429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.622104136364323e-05,
+      "loss": 0.5158,
+      "step": 17430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.621068670387747e-05,
+      "loss": 0.5899,
+      "step": 17431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.620033245307501e-05,
+      "loss": 0.4657,
+      "step": 17432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.618997861136123e-05,
+      "loss": 0.5595,
+      "step": 17433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.617962517886139e-05,
+      "loss": 0.5537,
+      "step": 17434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.616927215570082e-05,
+      "loss": 0.493,
+      "step": 17435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.615891954200485e-05,
+      "loss": 0.4658,
+      "step": 17436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.614856733789874e-05,
+      "loss": 0.4033,
+      "step": 17437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.613821554350782e-05,
+      "loss": 0.4425,
+      "step": 17438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.612786415895738e-05,
+      "loss": 0.5308,
+      "step": 17439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.611751318437266e-05,
+      "loss": 0.4693,
+      "step": 17440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6107162619879e-05,
+      "loss": 0.4509,
+      "step": 17441
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.609681246560167e-05,
+      "loss": 0.47,
+      "step": 17442
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.608646272166588e-05,
+      "loss": 0.4969,
+      "step": 17443
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.607611338819697e-05,
+      "loss": 0.5181,
+      "step": 17444
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.606576446532016e-05,
+      "loss": 0.4207,
+      "step": 17445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.605541595316073e-05,
+      "loss": 0.4565,
+      "step": 17446
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.604506785184392e-05,
+      "loss": 0.519,
+      "step": 17447
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.603472016149494e-05,
+      "loss": 0.5069,
+      "step": 17448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.602437288223911e-05,
+      "loss": 0.5791,
+      "step": 17449
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.60140260142016e-05,
+      "loss": 0.4619,
+      "step": 17450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.600367955750768e-05,
+      "loss": 0.5524,
+      "step": 17451
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.599333351228257e-05,
+      "loss": 0.5769,
+      "step": 17452
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.598298787865147e-05,
+      "loss": 0.559,
+      "step": 17453
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.597264265673962e-05,
+      "loss": 0.5078,
+      "step": 17454
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.596229784667222e-05,
+      "loss": 0.5345,
+      "step": 17455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.595195344857445e-05,
+      "loss": 0.4045,
+      "step": 17456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.594160946257158e-05,
+      "loss": 0.5877,
+      "step": 17457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.593126588878875e-05,
+      "loss": 0.5753,
+      "step": 17458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.592092272735117e-05,
+      "loss": 0.4428,
+      "step": 17459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.591057997838403e-05,
+      "loss": 0.4294,
+      "step": 17460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.59002376420125e-05,
+      "loss": 0.5378,
+      "step": 17461
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.588989571836177e-05,
+      "loss": 0.4514,
+      "step": 17462
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5879554207557e-05,
+      "loss": 0.5052,
+      "step": 17463
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.586921310972342e-05,
+      "loss": 0.484,
+      "step": 17464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.585887242498607e-05,
+      "loss": 0.5915,
+      "step": 17465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.584853215347019e-05,
+      "loss": 0.4413,
+      "step": 17466
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.583819229530089e-05,
+      "loss": 0.4948,
+      "step": 17467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.582785285060338e-05,
+      "loss": 0.4484,
+      "step": 17468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.581751381950274e-05,
+      "loss": 0.4988,
+      "step": 17469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.580717520212412e-05,
+      "loss": 0.507,
+      "step": 17470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.579683699859267e-05,
+      "loss": 0.4324,
+      "step": 17471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.57864992090335e-05,
+      "loss": 0.47,
+      "step": 17472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.577616183357173e-05,
+      "loss": 0.5344,
+      "step": 17473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.576582487233251e-05,
+      "loss": 0.4315,
+      "step": 17474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.575548832544091e-05,
+      "loss": 0.5062,
+      "step": 17475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.574515219302206e-05,
+      "loss": 0.4465,
+      "step": 17476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.573481647520108e-05,
+      "loss": 0.4578,
+      "step": 17477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.572448117210299e-05,
+      "loss": 0.46,
+      "step": 17478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.571414628385298e-05,
+      "loss": 0.5307,
+      "step": 17479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.570381181057606e-05,
+      "loss": 0.4884,
+      "step": 17480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.569347775239735e-05,
+      "loss": 0.5192,
+      "step": 17481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.568314410944192e-05,
+      "loss": 0.4518,
+      "step": 17482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.567281088183484e-05,
+      "loss": 0.5279,
+      "step": 17483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.566247806970119e-05,
+      "loss": 0.5284,
+      "step": 17484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.565214567316598e-05,
+      "loss": 0.4409,
+      "step": 17485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.564181369235433e-05,
+      "loss": 0.5044,
+      "step": 17486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.563148212739127e-05,
+      "loss": 0.4056,
+      "step": 17487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.562115097840182e-05,
+      "loss": 0.6113,
+      "step": 17488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.561082024551107e-05,
+      "loss": 0.5071,
+      "step": 17489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.560048992884403e-05,
+      "loss": 0.4378,
+      "step": 17490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.559016002852571e-05,
+      "loss": 0.4606,
+      "step": 17491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.557983054468117e-05,
+      "loss": 0.4922,
+      "step": 17492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.556950147743542e-05,
+      "loss": 0.4981,
+      "step": 17493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.555917282691347e-05,
+      "loss": 0.531,
+      "step": 17494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.554884459324033e-05,
+      "loss": 0.5337,
+      "step": 17495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.553851677654102e-05,
+      "loss": 0.5173,
+      "step": 17496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.552818937694053e-05,
+      "loss": 0.4131,
+      "step": 17497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.551786239456384e-05,
+      "loss": 0.478,
+      "step": 17498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5507535829536e-05,
+      "loss": 0.4738,
+      "step": 17499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.549720968198193e-05,
+      "loss": 0.5137,
+      "step": 17500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.548688395202662e-05,
+      "loss": 0.4871,
+      "step": 17501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.547655863979509e-05,
+      "loss": 0.485,
+      "step": 17502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.546623374541227e-05,
+      "loss": 0.5041,
+      "step": 17503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.545590926900315e-05,
+      "loss": 0.5414,
+      "step": 17504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.544558521069267e-05,
+      "loss": 0.5585,
+      "step": 17505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.543526157060578e-05,
+      "loss": 0.4989,
+      "step": 17506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.542493834886747e-05,
+      "loss": 0.4893,
+      "step": 17507
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.541461554560267e-05,
+      "loss": 0.4607,
+      "step": 17508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.540429316093627e-05,
+      "loss": 0.4557,
+      "step": 17509
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.539397119499326e-05,
+      "loss": 0.5284,
+      "step": 17510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.538364964789855e-05,
+      "loss": 0.4733,
+      "step": 17511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.537332851977711e-05,
+      "loss": 0.427,
+      "step": 17512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.53630078107538e-05,
+      "loss": 0.5184,
+      "step": 17513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.535268752095353e-05,
+      "loss": 0.5104,
+      "step": 17514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.534236765050127e-05,
+      "loss": 0.4457,
+      "step": 17515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.533204819952185e-05,
+      "loss": 0.4829,
+      "step": 17516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.532172916814025e-05,
+      "loss": 0.472,
+      "step": 17517
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.531141055648133e-05,
+      "loss": 0.4166,
+      "step": 17518
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.530109236466995e-05,
+      "loss": 0.4928,
+      "step": 17519
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.529077459283105e-05,
+      "loss": 0.4722,
+      "step": 17520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.528045724108942e-05,
+      "loss": 0.4982,
+      "step": 17521
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.527014030957004e-05,
+      "loss": 0.5482,
+      "step": 17522
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.525982379839774e-05,
+      "loss": 0.535,
+      "step": 17523
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.524950770769735e-05,
+      "loss": 0.4847,
+      "step": 17524
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.523919203759377e-05,
+      "loss": 0.5954,
+      "step": 17525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.522887678821184e-05,
+      "loss": 0.4474,
+      "step": 17526
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.52185619596764e-05,
+      "loss": 0.4955,
+      "step": 17527
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.52082475521123e-05,
+      "loss": 0.4526,
+      "step": 17528
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.519793356564439e-05,
+      "loss": 0.455,
+      "step": 17529
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.518762000039748e-05,
+      "loss": 0.5135,
+      "step": 17530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.517730685649645e-05,
+      "loss": 0.4291,
+      "step": 17531
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.516699413406604e-05,
+      "loss": 0.5107,
+      "step": 17532
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.515668183323115e-05,
+      "loss": 0.5443,
+      "step": 17533
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.514636995411654e-05,
+      "loss": 0.4796,
+      "step": 17534
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.513605849684704e-05,
+      "loss": 0.4908,
+      "step": 17535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.512574746154746e-05,
+      "loss": 0.4627,
+      "step": 17536
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.511543684834256e-05,
+      "loss": 0.4876,
+      "step": 17537
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.510512665735719e-05,
+      "loss": 0.6038,
+      "step": 17538
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.509481688871608e-05,
+      "loss": 0.4526,
+      "step": 17539
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.508450754254407e-05,
+      "loss": 0.5954,
+      "step": 17540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.507419861896591e-05,
+      "loss": 0.4477,
+      "step": 17541
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.506389011810634e-05,
+      "loss": 0.5191,
+      "step": 17542
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.505358204009017e-05,
+      "loss": 0.5705,
+      "step": 17543
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.504327438504216e-05,
+      "loss": 0.5769,
+      "step": 17544
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.503296715308704e-05,
+      "loss": 0.3939,
+      "step": 17545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.50226603443496e-05,
+      "loss": 0.4988,
+      "step": 17546
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.501235395895453e-05,
+      "loss": 0.4727,
+      "step": 17547
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.500204799702664e-05,
+      "loss": 0.4299,
+      "step": 17548
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.499174245869061e-05,
+      "loss": 0.5259,
+      "step": 17549
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.49814373440712e-05,
+      "loss": 0.5361,
+      "step": 17550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.497113265329316e-05,
+      "loss": 0.4656,
+      "step": 17551
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.496082838648114e-05,
+      "loss": 0.4698,
+      "step": 17552
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.495052454375992e-05,
+      "loss": 0.4528,
+      "step": 17553
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.49402211252542e-05,
+      "loss": 0.4995,
+      "step": 17554
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.492991813108864e-05,
+      "loss": 0.4657,
+      "step": 17555
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.4919615561388e-05,
+      "loss": 0.4671,
+      "step": 17556
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.490931341627695e-05,
+      "loss": 0.4367,
+      "step": 17557
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.489901169588018e-05,
+      "loss": 0.5074,
+      "step": 17558
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.488871040032236e-05,
+      "loss": 0.5014,
+      "step": 17559
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.487840952972819e-05,
+      "loss": 0.5277,
+      "step": 17560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.486810908422236e-05,
+      "loss": 0.5337,
+      "step": 17561
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.48578090639295e-05,
+      "loss": 0.4843,
+      "step": 17562
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.484750946897432e-05,
+      "loss": 0.4957,
+      "step": 17563
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.483721029948143e-05,
+      "loss": 0.518,
+      "step": 17564
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.48269115555755e-05,
+      "loss": 0.5193,
+      "step": 17565
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.481661323738121e-05,
+      "loss": 0.493,
+      "step": 17566
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.48063153450232e-05,
+      "loss": 0.5782,
+      "step": 17567
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.479601787862605e-05,
+      "loss": 0.4881,
+      "step": 17568
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.478572083831445e-05,
+      "loss": 0.5243,
+      "step": 17569
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.477542422421301e-05,
+      "loss": 0.471,
+      "step": 17570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.476512803644636e-05,
+      "loss": 0.6119,
+      "step": 17571
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.475483227513912e-05,
+      "loss": 0.4683,
+      "step": 17572
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.474453694041589e-05,
+      "loss": 0.5107,
+      "step": 17573
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.473424203240129e-05,
+      "loss": 0.5213,
+      "step": 17574
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.472394755121994e-05,
+      "loss": 0.4601,
+      "step": 17575
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.471365349699636e-05,
+      "loss": 0.4443,
+      "step": 17576
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.470335986985525e-05,
+      "loss": 0.4645,
+      "step": 17577
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.46930666699211e-05,
+      "loss": 0.4511,
+      "step": 17578
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.468277389731859e-05,
+      "loss": 0.5552,
+      "step": 17579
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.467248155217221e-05,
+      "loss": 0.4585,
+      "step": 17580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.466218963460655e-05,
+      "loss": 0.5323,
+      "step": 17581
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.465189814474624e-05,
+      "loss": 0.451,
+      "step": 17582
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.464160708271574e-05,
+      "loss": 0.4728,
+      "step": 17583
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.46313164486397e-05,
+      "loss": 0.5992,
+      "step": 17584
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.462102624264262e-05,
+      "loss": 0.5416,
+      "step": 17585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.461073646484906e-05,
+      "loss": 0.4486,
+      "step": 17586
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.460044711538355e-05,
+      "loss": 0.4683,
+      "step": 17587
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.459015819437062e-05,
+      "loss": 0.5259,
+      "step": 17588
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.457986970193484e-05,
+      "loss": 0.4428,
+      "step": 17589
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.45695816382007e-05,
+      "loss": 0.4752,
+      "step": 17590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.45592940032927e-05,
+      "loss": 0.4381,
+      "step": 17591
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.454900679733542e-05,
+      "loss": 0.5372,
+      "step": 17592
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.453872002045332e-05,
+      "loss": 0.5054,
+      "step": 17593
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.45284336727709e-05,
+      "loss": 0.4477,
+      "step": 17594
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.451814775441269e-05,
+      "loss": 0.4796,
+      "step": 17595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.450786226550316e-05,
+      "loss": 0.5534,
+      "step": 17596
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.449757720616683e-05,
+      "loss": 0.5928,
+      "step": 17597
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.448729257652814e-05,
+      "loss": 0.5241,
+      "step": 17598
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.447700837671164e-05,
+      "loss": 0.4637,
+      "step": 17599
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.44667246068417e-05,
+      "loss": 0.443,
+      "step": 17600
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 0.4828401505947113,
+      "eval_runtime": 853.2893,
+      "eval_samples_per_second": 28.197,
+      "eval_steps_per_second": 14.098,
+      "step": 17600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.445644126704284e-05,
+      "loss": 0.4874,
+      "step": 17601
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.444615835743955e-05,
+      "loss": 0.4997,
+      "step": 17602
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.443587587815622e-05,
+      "loss": 0.4945,
+      "step": 17603
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.442559382931739e-05,
+      "loss": 0.4683,
+      "step": 17604
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.441531221104742e-05,
+      "loss": 0.5008,
+      "step": 17605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.440503102347082e-05,
+      "loss": 0.4814,
+      "step": 17606
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.439475026671199e-05,
+      "loss": 0.3825,
+      "step": 17607
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.438446994089535e-05,
+      "loss": 0.5625,
+      "step": 17608
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.437419004614538e-05,
+      "loss": 0.4624,
+      "step": 17609
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.436391058258646e-05,
+      "loss": 0.3871,
+      "step": 17610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.435363155034299e-05,
+      "loss": 0.5345,
+      "step": 17611
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.434335294953942e-05,
+      "loss": 0.492,
+      "step": 17612
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.433307478030013e-05,
+      "loss": 0.521,
+      "step": 17613
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.432279704274953e-05,
+      "loss": 0.47,
+      "step": 17614
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.431251973701204e-05,
+      "loss": 0.5493,
+      "step": 17615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.430224286321199e-05,
+      "loss": 0.451,
+      "step": 17616
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.429196642147381e-05,
+      "loss": 0.5113,
+      "step": 17617
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.428169041192186e-05,
+      "loss": 0.4632,
+      "step": 17618
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.427141483468054e-05,
+      "loss": 0.5049,
+      "step": 17619
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.42611396898742e-05,
+      "loss": 0.6125,
+      "step": 17620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.425086497762718e-05,
+      "loss": 0.506,
+      "step": 17621
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.424059069806389e-05,
+      "loss": 0.3974,
+      "step": 17622
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.423031685130863e-05,
+      "loss": 0.5194,
+      "step": 17623
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.422004343748578e-05,
+      "loss": 0.4746,
+      "step": 17624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.420977045671969e-05,
+      "loss": 0.4435,
+      "step": 17625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.419949790913466e-05,
+      "loss": 0.4535,
+      "step": 17626
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.418922579485505e-05,
+      "loss": 0.4928,
+      "step": 17627
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.41789541140052e-05,
+      "loss": 0.4815,
+      "step": 17628
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.416868286670939e-05,
+      "loss": 0.4581,
+      "step": 17629
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.415841205309197e-05,
+      "loss": 0.4875,
+      "step": 17630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.414814167327723e-05,
+      "loss": 0.4946,
+      "step": 17631
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.413787172738949e-05,
+      "loss": 0.3939,
+      "step": 17632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.412760221555308e-05,
+      "loss": 0.4964,
+      "step": 17633
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.411733313789221e-05,
+      "loss": 0.5602,
+      "step": 17634
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.410706449453126e-05,
+      "loss": 0.4932,
+      "step": 17635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.409679628559445e-05,
+      "loss": 0.6332,
+      "step": 17636
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.40865285112061e-05,
+      "loss": 0.4687,
+      "step": 17637
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.407626117149048e-05,
+      "loss": 0.4842,
+      "step": 17638
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.406599426657183e-05,
+      "loss": 0.3958,
+      "step": 17639
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.405572779657444e-05,
+      "loss": 0.5367,
+      "step": 17640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.404546176162256e-05,
+      "loss": 0.4724,
+      "step": 17641
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.403519616184045e-05,
+      "loss": 0.5196,
+      "step": 17642
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.402493099735237e-05,
+      "loss": 0.4691,
+      "step": 17643
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.40146662682825e-05,
+      "loss": 0.528,
+      "step": 17644
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.400440197475516e-05,
+      "loss": 0.5275,
+      "step": 17645
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.399413811689455e-05,
+      "loss": 0.5925,
+      "step": 17646
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.398387469482487e-05,
+      "loss": 0.4656,
+      "step": 17647
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.397361170867039e-05,
+      "loss": 0.4237,
+      "step": 17648
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.396334915855527e-05,
+      "loss": 0.6477,
+      "step": 17649
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.395308704460377e-05,
+      "loss": 0.5345,
+      "step": 17650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.394282536694008e-05,
+      "loss": 0.5059,
+      "step": 17651
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.39325641256884e-05,
+      "loss": 0.532,
+      "step": 17652
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.392230332097291e-05,
+      "loss": 0.5409,
+      "step": 17653
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.39120429529178e-05,
+      "loss": 0.49,
+      "step": 17654
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.39017830216473e-05,
+      "loss": 0.5928,
+      "step": 17655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.389152352728556e-05,
+      "loss": 0.4665,
+      "step": 17656
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.388126446995671e-05,
+      "loss": 0.4819,
+      "step": 17657
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3871005849785e-05,
+      "loss": 0.4945,
+      "step": 17658
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.386074766689454e-05,
+      "loss": 0.4998,
+      "step": 17659
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.385048992140948e-05,
+      "loss": 0.5303,
+      "step": 17660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.384023261345402e-05,
+      "loss": 0.512,
+      "step": 17661
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.382997574315227e-05,
+      "loss": 0.6046,
+      "step": 17662
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.38197193106284e-05,
+      "loss": 0.4859,
+      "step": 17663
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.380946331600652e-05,
+      "loss": 0.4682,
+      "step": 17664
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.379920775941077e-05,
+      "loss": 0.5771,
+      "step": 17665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.378895264096529e-05,
+      "loss": 0.4786,
+      "step": 17666
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.377869796079418e-05,
+      "loss": 0.5019,
+      "step": 17667
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.376844371902157e-05,
+      "loss": 0.5721,
+      "step": 17668
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.375818991577158e-05,
+      "loss": 0.5221,
+      "step": 17669
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.374793655116828e-05,
+      "loss": 0.6556,
+      "step": 17670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.373768362533581e-05,
+      "loss": 0.5201,
+      "step": 17671
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.372743113839822e-05,
+      "loss": 0.4287,
+      "step": 17672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.371717909047967e-05,
+      "loss": 0.4941,
+      "step": 17673
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.370692748170418e-05,
+      "loss": 0.4585,
+      "step": 17674
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.369667631219583e-05,
+      "loss": 0.5308,
+      "step": 17675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.368642558207873e-05,
+      "loss": 0.5054,
+      "step": 17676
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.367617529147694e-05,
+      "loss": 0.5385,
+      "step": 17677
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.366592544051448e-05,
+      "loss": 0.5285,
+      "step": 17678
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.365567602931548e-05,
+      "loss": 0.4728,
+      "step": 17679
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.364542705800391e-05,
+      "loss": 0.6091,
+      "step": 17680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.363517852670389e-05,
+      "loss": 0.4302,
+      "step": 17681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.362493043553943e-05,
+      "loss": 0.4915,
+      "step": 17682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.361468278463454e-05,
+      "loss": 0.5218,
+      "step": 17683
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.360443557411331e-05,
+      "loss": 0.4402,
+      "step": 17684
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.35941888040997e-05,
+      "loss": 0.4976,
+      "step": 17685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.358394247471778e-05,
+      "loss": 0.4564,
+      "step": 17686
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.357369658609155e-05,
+      "loss": 0.4649,
+      "step": 17687
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.356345113834501e-05,
+      "loss": 0.4824,
+      "step": 17688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.355320613160218e-05,
+      "loss": 0.4233,
+      "step": 17689
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.354296156598703e-05,
+      "loss": 0.4714,
+      "step": 17690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.35327174416236e-05,
+      "loss": 0.473,
+      "step": 17691
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.352247375863583e-05,
+      "loss": 0.4978,
+      "step": 17692
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.351223051714771e-05,
+      "loss": 0.4794,
+      "step": 17693
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.350198771728326e-05,
+      "loss": 0.4873,
+      "step": 17694
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.349174535916643e-05,
+      "loss": 0.4256,
+      "step": 17695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.348150344292114e-05,
+      "loss": 0.5094,
+      "step": 17696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.347126196867144e-05,
+      "loss": 0.5045,
+      "step": 17697
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.346102093654118e-05,
+      "loss": 0.4701,
+      "step": 17698
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.345078034665441e-05,
+      "loss": 0.4429,
+      "step": 17699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.344054019913503e-05,
+      "loss": 0.5237,
+      "step": 17700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.343030049410697e-05,
+      "loss": 0.4545,
+      "step": 17701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.342006123169418e-05,
+      "loss": 0.4608,
+      "step": 17702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.340982241202058e-05,
+      "loss": 0.4837,
+      "step": 17703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.339958403521011e-05,
+      "loss": 0.5335,
+      "step": 17704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.33893461013867e-05,
+      "loss": 0.5284,
+      "step": 17705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.33791086106742e-05,
+      "loss": 0.4876,
+      "step": 17706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.336887156319658e-05,
+      "loss": 0.4558,
+      "step": 17707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.335863495907772e-05,
+      "loss": 0.5283,
+      "step": 17708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.334839879844151e-05,
+      "loss": 0.3894,
+      "step": 17709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.333816308141187e-05,
+      "loss": 0.5366,
+      "step": 17710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.332792780811266e-05,
+      "loss": 0.4394,
+      "step": 17711
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.331769297866777e-05,
+      "loss": 0.5368,
+      "step": 17712
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.330745859320106e-05,
+      "loss": 0.5447,
+      "step": 17713
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.32972246518364e-05,
+      "loss": 0.508,
+      "step": 17714
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.328699115469772e-05,
+      "loss": 0.5596,
+      "step": 17715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.327675810190879e-05,
+      "loss": 0.611,
+      "step": 17716
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.32665254935935e-05,
+      "loss": 0.5201,
+      "step": 17717
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.325629332987572e-05,
+      "loss": 0.4737,
+      "step": 17718
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.324606161087926e-05,
+      "loss": 0.4876,
+      "step": 17719
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.323583033672799e-05,
+      "loss": 0.4872,
+      "step": 17720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.322559950754569e-05,
+      "loss": 0.483,
+      "step": 17721
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.321536912345626e-05,
+      "loss": 0.493,
+      "step": 17722
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.320513918458347e-05,
+      "loss": 0.4891,
+      "step": 17723
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.319490969105114e-05,
+      "loss": 0.4699,
+      "step": 17724
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.318468064298311e-05,
+      "loss": 0.5175,
+      "step": 17725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.317445204050316e-05,
+      "loss": 0.6159,
+      "step": 17726
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.316422388373507e-05,
+      "loss": 0.5597,
+      "step": 17727
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.315399617280269e-05,
+      "loss": 0.5047,
+      "step": 17728
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.314376890782976e-05,
+      "loss": 0.5647,
+      "step": 17729
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.313354208894011e-05,
+      "loss": 0.5888,
+      "step": 17730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.31233157162575e-05,
+      "loss": 0.4314,
+      "step": 17731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.311308978990564e-05,
+      "loss": 0.4556,
+      "step": 17732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.310286431000843e-05,
+      "loss": 0.4787,
+      "step": 17733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.309263927668953e-05,
+      "loss": 0.535,
+      "step": 17734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.308241469007272e-05,
+      "loss": 0.5515,
+      "step": 17735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.307219055028173e-05,
+      "loss": 0.4252,
+      "step": 17736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.306196685744034e-05,
+      "loss": 0.4386,
+      "step": 17737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.305174361167228e-05,
+      "loss": 0.4853,
+      "step": 17738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.304152081310131e-05,
+      "loss": 0.4677,
+      "step": 17739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.303129846185114e-05,
+      "loss": 0.4731,
+      "step": 17740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.302107655804546e-05,
+      "loss": 0.491,
+      "step": 17741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.301085510180804e-05,
+      "loss": 0.5053,
+      "step": 17742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.300063409326258e-05,
+      "loss": 0.4999,
+      "step": 17743
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.299041353253277e-05,
+      "loss": 0.6012,
+      "step": 17744
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.298019341974232e-05,
+      "loss": 0.5109,
+      "step": 17745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.296997375501495e-05,
+      "loss": 0.4546,
+      "step": 17746
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.295975453847434e-05,
+      "loss": 0.4817,
+      "step": 17747
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.294953577024417e-05,
+      "loss": 0.4746,
+      "step": 17748
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.29393174504481e-05,
+      "loss": 0.5478,
+      "step": 17749
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.292909957920985e-05,
+      "loss": 0.5307,
+      "step": 17750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.291888215665305e-05,
+      "loss": 0.5483,
+      "step": 17751
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.29086651829014e-05,
+      "loss": 0.4934,
+      "step": 17752
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.289844865807855e-05,
+      "loss": 0.5105,
+      "step": 17753
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.288823258230812e-05,
+      "loss": 0.4362,
+      "step": 17754
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.28780169557138e-05,
+      "loss": 0.5465,
+      "step": 17755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.286780177841922e-05,
+      "loss": 0.5626,
+      "step": 17756
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.285758705054802e-05,
+      "loss": 0.5274,
+      "step": 17757
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.284737277222383e-05,
+      "loss": 0.4889,
+      "step": 17758
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.283715894357027e-05,
+      "loss": 0.4261,
+      "step": 17759
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.282694556471099e-05,
+      "loss": 0.5274,
+      "step": 17760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.281673263576957e-05,
+      "loss": 0.4809,
+      "step": 17761
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.280652015686961e-05,
+      "loss": 0.4258,
+      "step": 17762
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.279630812813476e-05,
+      "loss": 0.5042,
+      "step": 17763
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.278609654968859e-05,
+      "loss": 0.4709,
+      "step": 17764
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.277588542165472e-05,
+      "loss": 0.4626,
+      "step": 17765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.27656747441567e-05,
+      "loss": 0.4431,
+      "step": 17766
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.275546451731813e-05,
+      "loss": 0.4514,
+      "step": 17767
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.274525474126263e-05,
+      "loss": 0.5397,
+      "step": 17768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.273504541611367e-05,
+      "loss": 0.5446,
+      "step": 17769
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.272483654199494e-05,
+      "loss": 0.5552,
+      "step": 17770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.271462811902994e-05,
+      "loss": 0.5411,
+      "step": 17771
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.270442014734218e-05,
+      "loss": 0.5604,
+      "step": 17772
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.269421262705532e-05,
+      "loss": 0.4481,
+      "step": 17773
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.26840055582928e-05,
+      "loss": 0.4396,
+      "step": 17774
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.267379894117822e-05,
+      "loss": 0.4741,
+      "step": 17775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.266359277583511e-05,
+      "loss": 0.6265,
+      "step": 17776
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.265338706238695e-05,
+      "loss": 0.4943,
+      "step": 17777
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.264318180095733e-05,
+      "loss": 0.5723,
+      "step": 17778
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.263297699166974e-05,
+      "loss": 0.4427,
+      "step": 17779
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.262277263464768e-05,
+      "loss": 0.4876,
+      "step": 17780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.261256873001466e-05,
+      "loss": 0.5491,
+      "step": 17781
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.260236527789419e-05,
+      "loss": 0.5257,
+      "step": 17782
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.259216227840979e-05,
+      "loss": 0.5458,
+      "step": 17783
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.25819597316849e-05,
+      "loss": 0.4676,
+      "step": 17784
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.257175763784301e-05,
+      "loss": 0.4266,
+      "step": 17785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.256155599700764e-05,
+      "loss": 0.5165,
+      "step": 17786
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.255135480930224e-05,
+      "loss": 0.5315,
+      "step": 17787
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.254115407485029e-05,
+      "loss": 0.4305,
+      "step": 17788
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.253095379377523e-05,
+      "loss": 0.5094,
+      "step": 17789
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.252075396620051e-05,
+      "loss": 0.4876,
+      "step": 17790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.251055459224964e-05,
+      "loss": 0.5164,
+      "step": 17791
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.250035567204602e-05,
+      "loss": 0.4409,
+      "step": 17792
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.249015720571307e-05,
+      "loss": 0.5537,
+      "step": 17793
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.247995919337429e-05,
+      "loss": 0.5384,
+      "step": 17794
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.246976163515306e-05,
+      "loss": 0.4288,
+      "step": 17795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.245956453117282e-05,
+      "loss": 0.5031,
+      "step": 17796
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2449367881557e-05,
+      "loss": 0.4712,
+      "step": 17797
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.243917168642898e-05,
+      "loss": 0.4855,
+      "step": 17798
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.24289759459122e-05,
+      "loss": 0.4511,
+      "step": 17799
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.241878066013004e-05,
+      "loss": 0.4323,
+      "step": 17800
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 0.4787278473377228,
+      "eval_runtime": 845.1719,
+      "eval_samples_per_second": 28.468,
+      "eval_steps_per_second": 14.234,
+      "step": 17800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.240858582920594e-05,
+      "loss": 0.4144,
+      "step": 17801
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.239839145326324e-05,
+      "loss": 0.4642,
+      "step": 17802
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.238819753242534e-05,
+      "loss": 0.5507,
+      "step": 17803
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.237800406681564e-05,
+      "loss": 0.537,
+      "step": 17804
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.23678110565575e-05,
+      "loss": 0.5173,
+      "step": 17805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.235761850177428e-05,
+      "loss": 0.5125,
+      "step": 17806
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.234742640258938e-05,
+      "loss": 0.5106,
+      "step": 17807
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.23372347591261e-05,
+      "loss": 0.5658,
+      "step": 17808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.232704357150785e-05,
+      "loss": 0.4615,
+      "step": 17809
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.231685283985797e-05,
+      "loss": 0.535,
+      "step": 17810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.230666256429974e-05,
+      "loss": 0.5474,
+      "step": 17811
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.229647274495659e-05,
+      "loss": 0.5716,
+      "step": 17812
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.228628338195175e-05,
+      "loss": 0.5329,
+      "step": 17813
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.227609447540863e-05,
+      "loss": 0.5412,
+      "step": 17814
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.226590602545052e-05,
+      "loss": 0.6653,
+      "step": 17815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.225571803220072e-05,
+      "loss": 0.4384,
+      "step": 17816
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.224553049578257e-05,
+      "loss": 0.4632,
+      "step": 17817
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.223534341631932e-05,
+      "loss": 0.5109,
+      "step": 17818
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.222515679393433e-05,
+      "loss": 0.4675,
+      "step": 17819
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.221497062875086e-05,
+      "loss": 0.4739,
+      "step": 17820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.220478492089219e-05,
+      "loss": 0.4749,
+      "step": 17821
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.219459967048163e-05,
+      "loss": 0.5418,
+      "step": 17822
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.21844148776424e-05,
+      "loss": 0.5983,
+      "step": 17823
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.217423054249783e-05,
+      "loss": 0.5119,
+      "step": 17824
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.216404666517117e-05,
+      "loss": 0.4217,
+      "step": 17825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.215386324578564e-05,
+      "loss": 0.4796,
+      "step": 17826
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.214368028446456e-05,
+      "loss": 0.4066,
+      "step": 17827
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.213349778133112e-05,
+      "loss": 0.4496,
+      "step": 17828
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.212331573650857e-05,
+      "loss": 0.4517,
+      "step": 17829
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.211313415012019e-05,
+      "loss": 0.5194,
+      "step": 17830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.210295302228915e-05,
+      "loss": 0.5712,
+      "step": 17831
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.209277235313874e-05,
+      "loss": 0.4574,
+      "step": 17832
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.208259214279212e-05,
+      "loss": 0.4692,
+      "step": 17833
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.207241239137253e-05,
+      "loss": 0.4341,
+      "step": 17834
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.206223309900318e-05,
+      "loss": 0.5017,
+      "step": 17835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.205205426580728e-05,
+      "loss": 0.434,
+      "step": 17836
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.204187589190803e-05,
+      "loss": 0.5328,
+      "step": 17837
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.203169797742861e-05,
+      "loss": 0.4759,
+      "step": 17838
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.20215205224922e-05,
+      "loss": 0.4295,
+      "step": 17839
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2011343527222e-05,
+      "loss": 0.4956,
+      "step": 17840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.200116699174117e-05,
+      "loss": 0.4438,
+      "step": 17841
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.199099091617288e-05,
+      "loss": 0.5039,
+      "step": 17842
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.198081530064033e-05,
+      "loss": 0.4815,
+      "step": 17843
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.19706401452666e-05,
+      "loss": 0.5782,
+      "step": 17844
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.196046545017495e-05,
+      "loss": 0.4745,
+      "step": 17845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.195029121548845e-05,
+      "loss": 0.4352,
+      "step": 17846
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.194011744133025e-05,
+      "loss": 0.597,
+      "step": 17847
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.192994412782351e-05,
+      "loss": 0.502,
+      "step": 17848
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.191977127509135e-05,
+      "loss": 0.5887,
+      "step": 17849
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.190959888325689e-05,
+      "loss": 0.5236,
+      "step": 17850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.189942695244326e-05,
+      "loss": 0.4793,
+      "step": 17851
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.188925548277356e-05,
+      "loss": 0.5882,
+      "step": 17852
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.187908447437093e-05,
+      "loss": 0.4923,
+      "step": 17853
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.186891392735843e-05,
+      "loss": 0.498,
+      "step": 17854
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.18587438418592e-05,
+      "loss": 0.5331,
+      "step": 17855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.184857421799631e-05,
+      "loss": 0.4754,
+      "step": 17856
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.183840505589283e-05,
+      "loss": 0.616,
+      "step": 17857
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.182823635567187e-05,
+      "loss": 0.5133,
+      "step": 17858
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.18180681174565e-05,
+      "loss": 0.4779,
+      "step": 17859
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.180790034136976e-05,
+      "loss": 0.5349,
+      "step": 17860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.179773302753478e-05,
+      "loss": 0.5182,
+      "step": 17861
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.178756617607453e-05,
+      "loss": 0.494,
+      "step": 17862
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.177739978711212e-05,
+      "loss": 0.5293,
+      "step": 17863
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.176723386077061e-05,
+      "loss": 0.6357,
+      "step": 17864
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.175706839717299e-05,
+      "loss": 0.4176,
+      "step": 17865
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.174690339644233e-05,
+      "loss": 0.4526,
+      "step": 17866
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.173673885870166e-05,
+      "loss": 0.4777,
+      "step": 17867
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.172657478407404e-05,
+      "loss": 0.4498,
+      "step": 17868
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.171641117268239e-05,
+      "loss": 0.4793,
+      "step": 17869
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.17062480246498e-05,
+      "loss": 0.4498,
+      "step": 17870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.169608534009926e-05,
+      "loss": 0.4724,
+      "step": 17871
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.168592311915375e-05,
+      "loss": 0.5403,
+      "step": 17872
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.167576136193631e-05,
+      "loss": 0.5261,
+      "step": 17873
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.16656000685699e-05,
+      "loss": 0.5064,
+      "step": 17874
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.165543923917752e-05,
+      "loss": 0.4925,
+      "step": 17875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.164527887388214e-05,
+      "loss": 0.4925,
+      "step": 17876
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.163511897280675e-05,
+      "loss": 0.4932,
+      "step": 17877
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.16249595360743e-05,
+      "loss": 0.5647,
+      "step": 17878
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.161480056380775e-05,
+      "loss": 0.516,
+      "step": 17879
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.160464205613009e-05,
+      "loss": 0.4698,
+      "step": 17880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.159448401316426e-05,
+      "loss": 0.4635,
+      "step": 17881
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.158432643503315e-05,
+      "loss": 0.5009,
+      "step": 17882
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.157416932185977e-05,
+      "loss": 0.4717,
+      "step": 17883
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.156401267376703e-05,
+      "loss": 0.5711,
+      "step": 17884
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.155385649087788e-05,
+      "loss": 0.5726,
+      "step": 17885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.154370077331522e-05,
+      "loss": 0.4751,
+      "step": 17886
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.153354552120194e-05,
+      "loss": 0.466,
+      "step": 17887
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.152339073466103e-05,
+      "loss": 0.4796,
+      "step": 17888
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.151323641381531e-05,
+      "loss": 0.5073,
+      "step": 17889
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.150308255878775e-05,
+      "loss": 0.4998,
+      "step": 17890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.149292916970122e-05,
+      "loss": 0.4648,
+      "step": 17891
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.14827762466786e-05,
+      "loss": 0.4015,
+      "step": 17892
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.147262378984277e-05,
+      "loss": 0.476,
+      "step": 17893
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.146247179931666e-05,
+      "loss": 0.4911,
+      "step": 17894
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.145232027522306e-05,
+      "loss": 0.4994,
+      "step": 17895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.144216921768491e-05,
+      "loss": 0.5698,
+      "step": 17896
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.143201862682502e-05,
+      "loss": 0.5123,
+      "step": 17897
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.14218685027663e-05,
+      "loss": 0.4768,
+      "step": 17898
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.141171884563155e-05,
+      "loss": 0.4965,
+      "step": 17899
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.140156965554362e-05,
+      "loss": 0.4388,
+      "step": 17900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.13914209326254e-05,
+      "loss": 0.471,
+      "step": 17901
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.138127267699965e-05,
+      "loss": 0.4969,
+      "step": 17902
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.137112488878924e-05,
+      "loss": 0.4498,
+      "step": 17903
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1360977568117e-05,
+      "loss": 0.4614,
+      "step": 17904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.135083071510573e-05,
+      "loss": 0.4631,
+      "step": 17905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.134068432987825e-05,
+      "loss": 0.4927,
+      "step": 17906
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.133053841255733e-05,
+      "loss": 0.4337,
+      "step": 17907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.132039296326583e-05,
+      "loss": 0.4,
+      "step": 17908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.13102479821265e-05,
+      "loss": 0.5255,
+      "step": 17909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.130010346926214e-05,
+      "loss": 0.3887,
+      "step": 17910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.128995942479553e-05,
+      "loss": 0.4628,
+      "step": 17911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.127981584884945e-05,
+      "loss": 0.5176,
+      "step": 17912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.126967274154667e-05,
+      "loss": 0.5205,
+      "step": 17913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.125953010300995e-05,
+      "loss": 0.5184,
+      "step": 17914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.124938793336205e-05,
+      "loss": 0.4638,
+      "step": 17915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.123924623272575e-05,
+      "loss": 0.5464,
+      "step": 17916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.122910500122379e-05,
+      "loss": 0.5017,
+      "step": 17917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.121896423897886e-05,
+      "loss": 0.4661,
+      "step": 17918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.120882394611376e-05,
+      "loss": 0.4709,
+      "step": 17919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.119868412275118e-05,
+      "loss": 0.4949,
+      "step": 17920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.118854476901389e-05,
+      "loss": 0.456,
+      "step": 17921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.117840588502458e-05,
+      "loss": 0.4639,
+      "step": 17922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.116826747090596e-05,
+      "loss": 0.5137,
+      "step": 17923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.115812952678075e-05,
+      "loss": 0.4125,
+      "step": 17924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.114799205277164e-05,
+      "loss": 0.5033,
+      "step": 17925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.113785504900135e-05,
+      "loss": 0.4106,
+      "step": 17926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.112771851559256e-05,
+      "loss": 0.4932,
+      "step": 17927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.111758245266794e-05,
+      "loss": 0.5221,
+      "step": 17928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.11074468603502e-05,
+      "loss": 0.3808,
+      "step": 17929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1097311738762e-05,
+      "loss": 0.5019,
+      "step": 17930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1087177088026e-05,
+      "loss": 0.5401,
+      "step": 17931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.107704290826488e-05,
+      "loss": 0.4136,
+      "step": 17932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.106690919960127e-05,
+      "loss": 0.5465,
+      "step": 17933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.105677596215787e-05,
+      "loss": 0.5727,
+      "step": 17934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.104664319605727e-05,
+      "loss": 0.458,
+      "step": 17935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.103651090142214e-05,
+      "loss": 0.4918,
+      "step": 17936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1026379078375117e-05,
+      "loss": 0.4318,
+      "step": 17937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.101624772703881e-05,
+      "loss": 0.5655,
+      "step": 17938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.100611684753588e-05,
+      "loss": 0.4984,
+      "step": 17939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.099598643998892e-05,
+      "loss": 0.4733,
+      "step": 17940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.098585650452051e-05,
+      "loss": 0.4572,
+      "step": 17941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.097572704125332e-05,
+      "loss": 0.4755,
+      "step": 17942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.09655980503099e-05,
+      "loss": 0.4128,
+      "step": 17943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.095546953181288e-05,
+      "loss": 0.4853,
+      "step": 17944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0945341485884824e-05,
+      "loss": 0.4913,
+      "step": 17945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0935213912648315e-05,
+      "loss": 0.437,
+      "step": 17946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.092508681222595e-05,
+      "loss": 0.4521,
+      "step": 17947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.091496018474028e-05,
+      "loss": 0.4154,
+      "step": 17948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.090483403031387e-05,
+      "loss": 0.5076,
+      "step": 17949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.089470834906932e-05,
+      "loss": 0.5059,
+      "step": 17950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.088458314112912e-05,
+      "loss": 0.5053,
+      "step": 17951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0874458406615876e-05,
+      "loss": 0.4631,
+      "step": 17952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.086433414565211e-05,
+      "loss": 0.4438,
+      "step": 17953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.085421035836034e-05,
+      "loss": 0.4761,
+      "step": 17954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.084408704486313e-05,
+      "loss": 0.5388,
+      "step": 17955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.083396420528298e-05,
+      "loss": 0.4536,
+      "step": 17956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.082384183974244e-05,
+      "loss": 0.5522,
+      "step": 17957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.081371994836399e-05,
+      "loss": 0.4913,
+      "step": 17958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.080359853127015e-05,
+      "loss": 0.4896,
+      "step": 17959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.079347758858344e-05,
+      "loss": 0.4893,
+      "step": 17960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.078335712042634e-05,
+      "loss": 0.4871,
+      "step": 17961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.077323712692133e-05,
+      "loss": 0.6495,
+      "step": 17962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0763117608190934e-05,
+      "loss": 0.5145,
+      "step": 17963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.075299856435759e-05,
+      "loss": 0.4682,
+      "step": 17964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.074287999554381e-05,
+      "loss": 0.5244,
+      "step": 17965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.073276190187205e-05,
+      "loss": 0.4247,
+      "step": 17966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0722644283464726e-05,
+      "loss": 0.5341,
+      "step": 17967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.071252714044438e-05,
+      "loss": 0.467,
+      "step": 17968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.070241047293338e-05,
+      "loss": 0.4574,
+      "step": 17969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.069229428105423e-05,
+      "loss": 0.4526,
+      "step": 17970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.068217856492937e-05,
+      "loss": 0.5081,
+      "step": 17971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.067206332468117e-05,
+      "loss": 0.5391,
+      "step": 17972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0661948560432134e-05,
+      "loss": 0.4977,
+      "step": 17973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0651834272304634e-05,
+      "loss": 0.4615,
+      "step": 17974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.064172046042112e-05,
+      "loss": 0.4574,
+      "step": 17975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0631607124903976e-05,
+      "loss": 0.5176,
+      "step": 17976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.062149426587561e-05,
+      "loss": 0.4321,
+      "step": 17977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.061138188345845e-05,
+      "loss": 0.4904,
+      "step": 17978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.060126997777487e-05,
+      "loss": 0.4085,
+      "step": 17979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0591158548947235e-05,
+      "loss": 0.5491,
+      "step": 17980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.058104759709797e-05,
+      "loss": 0.4075,
+      "step": 17981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0570937122349404e-05,
+      "loss": 0.4959,
+      "step": 17982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.056082712482396e-05,
+      "loss": 0.4958,
+      "step": 17983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.055071760464397e-05,
+      "loss": 0.4538,
+      "step": 17984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.054060856193179e-05,
+      "loss": 0.4391,
+      "step": 17985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0530499996809795e-05,
+      "loss": 0.4576,
+      "step": 17986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.05203919094003e-05,
+      "loss": 0.4365,
+      "step": 17987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.051028429982568e-05,
+      "loss": 0.5341,
+      "step": 17988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.050017716820827e-05,
+      "loss": 0.5464,
+      "step": 17989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.049007051467036e-05,
+      "loss": 0.4744,
+      "step": 17990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0479964339334326e-05,
+      "loss": 0.4697,
+      "step": 17991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.046985864232244e-05,
+      "loss": 0.5093,
+      "step": 17992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.045975342375705e-05,
+      "loss": 0.5089,
+      "step": 17993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.044964868376045e-05,
+      "loss": 0.5347,
+      "step": 17994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.043954442245492e-05,
+      "loss": 0.4113,
+      "step": 17995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.042944063996281e-05,
+      "loss": 0.4351,
+      "step": 17996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.041933733640636e-05,
+      "loss": 0.4429,
+      "step": 17997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0409234511907855e-05,
+      "loss": 0.388,
+      "step": 17998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.039913216658959e-05,
+      "loss": 0.3968,
+      "step": 17999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.038903030057383e-05,
+      "loss": 0.5066,
+      "step": 18000
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 0.47541379928588867,
+      "eval_runtime": 851.4684,
+      "eval_samples_per_second": 28.257,
+      "eval_steps_per_second": 14.129,
+      "step": 18000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.037892891398287e-05,
+      "loss": 0.4391,
+      "step": 18001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.036882800693893e-05,
+      "loss": 0.437,
+      "step": 18002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.03587275795643e-05,
+      "loss": 0.5586,
+      "step": 18003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.034862763198117e-05,
+      "loss": 0.4702,
+      "step": 18004
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0338528164311844e-05,
+      "loss": 0.579,
+      "step": 18005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0328429176678546e-05,
+      "loss": 0.4819,
+      "step": 18006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.031833066920347e-05,
+      "loss": 0.5436,
+      "step": 18007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.030823264200889e-05,
+      "loss": 0.5318,
+      "step": 18008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.029813509521698e-05,
+      "loss": 0.5216,
+      "step": 18009
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0288038028950014e-05,
+      "loss": 0.506,
+      "step": 18010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0277941443330146e-05,
+      "loss": 0.5115,
+      "step": 18011
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.02678453384796e-05,
+      "loss": 0.5196,
+      "step": 18012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.025774971452057e-05,
+      "loss": 0.5178,
+      "step": 18013
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.024765457157526e-05,
+      "loss": 0.4206,
+      "step": 18014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0237559909765815e-05,
+      "loss": 0.4811,
+      "step": 18015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.022746572921447e-05,
+      "loss": 0.4786,
+      "step": 18016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.021737203004334e-05,
+      "loss": 0.4677,
+      "step": 18017
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0207278812374645e-05,
+      "loss": 0.488,
+      "step": 18018
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0197186076330525e-05,
+      "loss": 0.5013,
+      "step": 18019
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0187093822033115e-05,
+      "loss": 0.4997,
+      "step": 18020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.01770020496046e-05,
+      "loss": 0.4948,
+      "step": 18021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.016691075916709e-05,
+      "loss": 0.424,
+      "step": 18022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.015681995084277e-05,
+      "loss": 0.494,
+      "step": 18023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0146729624753725e-05,
+      "loss": 0.4491,
+      "step": 18024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.01366397810221e-05,
+      "loss": 0.4541,
+      "step": 18025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0126550419770024e-05,
+      "loss": 0.4485,
+      "step": 18026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0116461541119585e-05,
+      "loss": 0.4061,
+      "step": 18027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.010637314519294e-05,
+      "loss": 0.449,
+      "step": 18028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.009628523211215e-05,
+      "loss": 0.5396,
+      "step": 18029
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.008619780199931e-05,
+      "loss": 0.5348,
+      "step": 18030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.007611085497655e-05,
+      "loss": 0.5547,
+      "step": 18031
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0066024391165944e-05,
+      "loss": 0.4793,
+      "step": 18032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.005593841068953e-05,
+      "loss": 0.4843,
+      "step": 18033
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.004585291366942e-05,
+      "loss": 0.526,
+      "step": 18034
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.003576790022767e-05,
+      "loss": 0.5227,
+      "step": 18035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0025683370486374e-05,
+      "loss": 0.6062,
+      "step": 18036
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.001559932456755e-05,
+      "loss": 0.4914,
+      "step": 18037
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.000551576259324e-05,
+      "loss": 0.4759,
+      "step": 18038
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.999543268468553e-05,
+      "loss": 0.4652,
+      "step": 18039
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.998535009096642e-05,
+      "loss": 0.5482,
+      "step": 18040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.997526798155798e-05,
+      "loss": 0.504,
+      "step": 18041
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.996518635658222e-05,
+      "loss": 0.4844,
+      "step": 18042
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.995510521616113e-05,
+      "loss": 0.4624,
+      "step": 18043
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9945024560416786e-05,
+      "loss": 0.5246,
+      "step": 18044
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.993494438947116e-05,
+      "loss": 0.4487,
+      "step": 18045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.992486470344625e-05,
+      "loss": 0.5926,
+      "step": 18046
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9914785502464074e-05,
+      "loss": 0.5133,
+      "step": 18047
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.990470678664659e-05,
+      "loss": 0.4746,
+      "step": 18048
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9894628556115854e-05,
+      "loss": 0.5425,
+      "step": 18049
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.988455081099378e-05,
+      "loss": 0.4563,
+      "step": 18050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.987447355140235e-05,
+      "loss": 0.5235,
+      "step": 18051
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9864396777463575e-05,
+      "loss": 0.5329,
+      "step": 18052
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.985432048929935e-05,
+      "loss": 0.5332,
+      "step": 18053
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9844244687031716e-05,
+      "loss": 0.465,
+      "step": 18054
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9834169370782566e-05,
+      "loss": 0.4961,
+      "step": 18055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9824094540673835e-05,
+      "loss": 0.5026,
+      "step": 18056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.98140201968275e-05,
+      "loss": 0.4262,
+      "step": 18057
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9803946339365456e-05,
+      "loss": 0.5293,
+      "step": 18058
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.979387296840969e-05,
+      "loss": 0.5842,
+      "step": 18059
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.978380008408209e-05,
+      "loss": 0.421,
+      "step": 18060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9773727686504524e-05,
+      "loss": 0.4331,
+      "step": 18061
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9763655775798976e-05,
+      "loss": 0.5045,
+      "step": 18062
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.975358435208732e-05,
+      "loss": 0.5043,
+      "step": 18063
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.974351341549142e-05,
+      "loss": 0.4486,
+      "step": 18064
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.973344296613325e-05,
+      "loss": 0.5165,
+      "step": 18065
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9723373004134596e-05,
+      "loss": 0.5176,
+      "step": 18066
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.971330352961744e-05,
+      "loss": 0.4785,
+      "step": 18067
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.970323454270358e-05,
+      "loss": 0.5187,
+      "step": 18068
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9693166043514895e-05,
+      "loss": 0.4314,
+      "step": 18069
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.968309803217328e-05,
+      "loss": 0.5444,
+      "step": 18070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.967303050880055e-05,
+      "loss": 0.3882,
+      "step": 18071
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9662963473518604e-05,
+      "loss": 0.4816,
+      "step": 18072
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.965289692644927e-05,
+      "loss": 0.4698,
+      "step": 18073
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.964283086771435e-05,
+      "loss": 0.5783,
+      "step": 18074
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9632765297435735e-05,
+      "loss": 0.5031,
+      "step": 18075
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9622700215735186e-05,
+      "loss": 0.4362,
+      "step": 18076
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.961263562273458e-05,
+      "loss": 0.4566,
+      "step": 18077
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.960257151855573e-05,
+      "loss": 0.5397,
+      "step": 18078
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9592507903320396e-05,
+      "loss": 0.4608,
+      "step": 18079
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9582444777150426e-05,
+      "loss": 0.4977,
+      "step": 18080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.957238214016762e-05,
+      "loss": 0.3787,
+      "step": 18081
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.956231999249372e-05,
+      "loss": 0.4518,
+      "step": 18082
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.955225833425058e-05,
+      "loss": 0.491,
+      "step": 18083
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.954219716555991e-05,
+      "loss": 0.4136,
+      "step": 18084
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.953213648654354e-05,
+      "loss": 0.4357,
+      "step": 18085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9522076297323227e-05,
+      "loss": 0.5391,
+      "step": 18086
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.951201659802068e-05,
+      "loss": 0.4413,
+      "step": 18087
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9501957388757734e-05,
+      "loss": 0.5076,
+      "step": 18088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9491898669656076e-05,
+      "loss": 0.4672,
+      "step": 18089
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.948184044083751e-05,
+      "loss": 0.4501,
+      "step": 18090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.947178270242373e-05,
+      "loss": 0.5396,
+      "step": 18091
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9461725454536455e-05,
+      "loss": 0.4523,
+      "step": 18092
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9451668697297456e-05,
+      "loss": 0.483,
+      "step": 18093
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.944161243082841e-05,
+      "loss": 0.4991,
+      "step": 18094
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.943155665525106e-05,
+      "loss": 0.5614,
+      "step": 18095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.942150137068714e-05,
+      "loss": 0.4578,
+      "step": 18096
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.941144657725828e-05,
+      "loss": 0.469,
+      "step": 18097
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.940139227508623e-05,
+      "loss": 0.4911,
+      "step": 18098
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.939133846429268e-05,
+      "loss": 0.4585,
+      "step": 18099
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.938128514499928e-05,
+      "loss": 0.4274,
+      "step": 18100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.937123231732774e-05,
+      "loss": 0.3977,
+      "step": 18101
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9361179981399714e-05,
+      "loss": 0.5496,
+      "step": 18102
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9351128137336884e-05,
+      "loss": 0.4985,
+      "step": 18103
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9341076785260916e-05,
+      "loss": 0.4949,
+      "step": 18104
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9331025925293425e-05,
+      "loss": 0.4475,
+      "step": 18105
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9320975557556126e-05,
+      "loss": 0.4305,
+      "step": 18106
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9310925682170585e-05,
+      "loss": 0.5059,
+      "step": 18107
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.930087629925851e-05,
+      "loss": 0.4847,
+      "step": 18108
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9290827408941495e-05,
+      "loss": 0.4252,
+      "step": 18109
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.928077901134115e-05,
+      "loss": 0.4569,
+      "step": 18110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.927073110657915e-05,
+      "loss": 0.5276,
+      "step": 18111
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.926068369477704e-05,
+      "loss": 0.4797,
+      "step": 18112
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.925063677605647e-05,
+      "loss": 0.4542,
+      "step": 18113
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.924059035053904e-05,
+      "loss": 0.536,
+      "step": 18114
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9230544418346315e-05,
+      "loss": 0.4261,
+      "step": 18115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.922049897959991e-05,
+      "loss": 0.5173,
+      "step": 18116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9210454034421426e-05,
+      "loss": 0.574,
+      "step": 18117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9200409582932406e-05,
+      "loss": 0.5,
+      "step": 18118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.919036562525443e-05,
+      "loss": 0.5089,
+      "step": 18119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.918032216150906e-05,
+      "loss": 0.4824,
+      "step": 18120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9170279191817876e-05,
+      "loss": 0.4849,
+      "step": 18121
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.916023671630242e-05,
+      "loss": 0.4185,
+      "step": 18122
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9150194735084206e-05,
+      "loss": 0.4865,
+      "step": 18123
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9140153248284843e-05,
+      "loss": 0.4531,
+      "step": 18124
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.91301122560258e-05,
+      "loss": 0.4419,
+      "step": 18125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.912007175842866e-05,
+      "loss": 0.4617,
+      "step": 18126
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.911003175561494e-05,
+      "loss": 0.4626,
+      "step": 18127
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.90999922477061e-05,
+      "loss": 0.5072,
+      "step": 18128
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.908995323482371e-05,
+      "loss": 0.479,
+      "step": 18129
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9079914717089265e-05,
+      "loss": 0.4473,
+      "step": 18130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.906987669462424e-05,
+      "loss": 0.4713,
+      "step": 18131
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.905983916755016e-05,
+      "loss": 0.4868,
+      "step": 18132
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9049802135988474e-05,
+      "loss": 0.4732,
+      "step": 18133
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.903976560006071e-05,
+      "loss": 0.4657,
+      "step": 18134
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9029729559888325e-05,
+      "loss": 0.4429,
+      "step": 18135
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9019694015592776e-05,
+      "loss": 0.5333,
+      "step": 18136
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9009658967295556e-05,
+      "loss": 0.4874,
+      "step": 18137
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.899962441511808e-05,
+      "loss": 0.4693,
+      "step": 18138
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8989590359181836e-05,
+      "loss": 0.5162,
+      "step": 18139
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.897955679960823e-05,
+      "loss": 0.406,
+      "step": 18140
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.896952373651874e-05,
+      "loss": 0.4443,
+      "step": 18141
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8959491170034775e-05,
+      "loss": 0.4078,
+      "step": 18142
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.894945910027778e-05,
+      "loss": 0.4597,
+      "step": 18143
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8939427527369185e-05,
+      "loss": 0.5231,
+      "step": 18144
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8929396451430364e-05,
+      "loss": 0.5077,
+      "step": 18145
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8919365872582776e-05,
+      "loss": 0.5236,
+      "step": 18146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.89093357909478e-05,
+      "loss": 0.4511,
+      "step": 18147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.889930620664681e-05,
+      "loss": 0.4937,
+      "step": 18148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8889277119801246e-05,
+      "loss": 0.5336,
+      "step": 18149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.887924853053245e-05,
+      "loss": 0.4709,
+      "step": 18150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.886922043896184e-05,
+      "loss": 0.5062,
+      "step": 18151
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.885919284521076e-05,
+      "loss": 0.54,
+      "step": 18152
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.884916574940058e-05,
+      "loss": 0.5067,
+      "step": 18153
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.883913915165269e-05,
+      "loss": 0.3933,
+      "step": 18154
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.882911305208839e-05,
+      "loss": 0.5044,
+      "step": 18155
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.881908745082909e-05,
+      "loss": 0.4378,
+      "step": 18156
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8809062347996116e-05,
+      "loss": 0.4311,
+      "step": 18157
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.879903774371076e-05,
+      "loss": 0.4676,
+      "step": 18158
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.878901363809441e-05,
+      "loss": 0.4064,
+      "step": 18159
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8778990031268364e-05,
+      "loss": 0.4497,
+      "step": 18160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.876896692335396e-05,
+      "loss": 0.5295,
+      "step": 18161
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8758944314472506e-05,
+      "loss": 0.4108,
+      "step": 18162
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.874892220474526e-05,
+      "loss": 0.5133,
+      "step": 18163
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.87389005942936e-05,
+      "loss": 0.4355,
+      "step": 18164
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8728879483238777e-05,
+      "loss": 0.5448,
+      "step": 18165
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.871885887170207e-05,
+      "loss": 0.5513,
+      "step": 18166
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8708838759804796e-05,
+      "loss": 0.5751,
+      "step": 18167
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.869881914766821e-05,
+      "loss": 0.5226,
+      "step": 18168
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.868880003541359e-05,
+      "loss": 0.4446,
+      "step": 18169
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.867878142316221e-05,
+      "loss": 0.4844,
+      "step": 18170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.866876331103529e-05,
+      "loss": 0.5321,
+      "step": 18171
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8658745699154125e-05,
+      "loss": 0.411,
+      "step": 18172
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8648728587639945e-05,
+      "loss": 0.5851,
+      "step": 18173
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8638711976613994e-05,
+      "loss": 0.4678,
+      "step": 18174
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8628695866197504e-05,
+      "loss": 0.4638,
+      "step": 18175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.861868025651169e-05,
+      "loss": 0.3879,
+      "step": 18176
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.86086651476778e-05,
+      "loss": 0.4049,
+      "step": 18177
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.859865053981703e-05,
+      "loss": 0.5082,
+      "step": 18178
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.858863643305062e-05,
+      "loss": 0.5054,
+      "step": 18179
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.857862282749973e-05,
+      "loss": 0.5161,
+      "step": 18180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.856860972328557e-05,
+      "loss": 0.3856,
+      "step": 18181
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.855859712052937e-05,
+      "loss": 0.3836,
+      "step": 18182
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.854858501935227e-05,
+      "loss": 0.5987,
+      "step": 18183
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8538573419875474e-05,
+      "loss": 0.5183,
+      "step": 18184
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8528562322220146e-05,
+      "loss": 0.4367,
+      "step": 18185
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8518551726507455e-05,
+      "loss": 0.449,
+      "step": 18186
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.850854163285857e-05,
+      "loss": 0.4519,
+      "step": 18187
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.849853204139465e-05,
+      "loss": 0.5301,
+      "step": 18188
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.848852295223681e-05,
+      "loss": 0.4739,
+      "step": 18189
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.847851436550624e-05,
+      "loss": 0.5493,
+      "step": 18190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.846850628132403e-05,
+      "loss": 0.4345,
+      "step": 18191
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.845849869981137e-05,
+      "loss": 0.4846,
+      "step": 18192
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.844849162108934e-05,
+      "loss": 0.4668,
+      "step": 18193
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8438485045279066e-05,
+      "loss": 0.4643,
+      "step": 18194
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.842847897250168e-05,
+      "loss": 0.4394,
+      "step": 18195
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.841847340287827e-05,
+      "loss": 0.5364,
+      "step": 18196
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.840846833652992e-05,
+      "loss": 0.4393,
+      "step": 18197
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.839846377357776e-05,
+      "loss": 0.4342,
+      "step": 18198
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.838845971414286e-05,
+      "loss": 0.4702,
+      "step": 18199
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8378456158346314e-05,
+      "loss": 0.4388,
+      "step": 18200
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 0.4717486500740051,
+      "eval_runtime": 851.7916,
+      "eval_samples_per_second": 28.246,
+      "eval_steps_per_second": 14.123,
+      "step": 18200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.836845310630921e-05,
+      "loss": 0.4437,
+      "step": 18201
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.835845055815259e-05,
+      "loss": 0.5751,
+      "step": 18202
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.834844851399749e-05,
+      "loss": 0.4595,
+      "step": 18203
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8338446973965045e-05,
+      "loss": 0.4863,
+      "step": 18204
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8328445938176266e-05,
+      "loss": 0.4231,
+      "step": 18205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.831844540675221e-05,
+      "loss": 0.4969,
+      "step": 18206
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8308445379813906e-05,
+      "loss": 0.4821,
+      "step": 18207
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.829844585748234e-05,
+      "loss": 0.4745,
+      "step": 18208
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.828844683987863e-05,
+      "loss": 0.6099,
+      "step": 18209
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.827844832712377e-05,
+      "loss": 0.408,
+      "step": 18210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.826845031933874e-05,
+      "loss": 0.5273,
+      "step": 18211
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.825845281664459e-05,
+      "loss": 0.5084,
+      "step": 18212
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.824845581916225e-05,
+      "loss": 0.5029,
+      "step": 18213
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.823845932701283e-05,
+      "loss": 0.4776,
+      "step": 18214
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8228463340317244e-05,
+      "loss": 0.5019,
+      "step": 18215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.821846785919651e-05,
+      "loss": 0.4898,
+      "step": 18216
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.820847288377158e-05,
+      "loss": 0.472,
+      "step": 18217
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.81984784141634e-05,
+      "loss": 0.4492,
+      "step": 18218
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.818848445049302e-05,
+      "loss": 0.5332,
+      "step": 18219
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.817849099288135e-05,
+      "loss": 0.4879,
+      "step": 18220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8168498041449346e-05,
+      "loss": 0.4498,
+      "step": 18221
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.815850559631797e-05,
+      "loss": 0.4999,
+      "step": 18222
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.814851365760814e-05,
+      "loss": 0.5032,
+      "step": 18223
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8138522225440786e-05,
+      "loss": 0.4477,
+      "step": 18224
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.812853129993687e-05,
+      "loss": 0.5137,
+      "step": 18225
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.811854088121731e-05,
+      "loss": 0.4121,
+      "step": 18226
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.810855096940302e-05,
+      "loss": 0.4933,
+      "step": 18227
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.80985615646149e-05,
+      "loss": 0.5244,
+      "step": 18228
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.808857266697383e-05,
+      "loss": 0.4637,
+      "step": 18229
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.807858427660078e-05,
+      "loss": 0.5582,
+      "step": 18230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8068596393616594e-05,
+      "loss": 0.4114,
+      "step": 18231
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8058609018142177e-05,
+      "loss": 0.5266,
+      "step": 18232
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.804862215029839e-05,
+      "loss": 0.5056,
+      "step": 18233
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.803863579020609e-05,
+      "loss": 0.5054,
+      "step": 18234
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.80286499379862e-05,
+      "loss": 0.4783,
+      "step": 18235
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.801866459375956e-05,
+      "loss": 0.4754,
+      "step": 18236
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.800867975764703e-05,
+      "loss": 0.5828,
+      "step": 18237
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7998695429769444e-05,
+      "loss": 0.4827,
+      "step": 18238
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.798871161024762e-05,
+      "loss": 0.4431,
+      "step": 18239
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.797872829920245e-05,
+      "loss": 0.5773,
+      "step": 18240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.796874549675475e-05,
+      "loss": 0.4403,
+      "step": 18241
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.795876320302536e-05,
+      "loss": 0.4241,
+      "step": 18242
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.794878141813505e-05,
+      "loss": 0.459,
+      "step": 18243
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.793880014220463e-05,
+      "loss": 0.4222,
+      "step": 18244
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.792881937535498e-05,
+      "loss": 0.5468,
+      "step": 18245
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7918839117706855e-05,
+      "loss": 0.4668,
+      "step": 18246
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.790885936938105e-05,
+      "loss": 0.5031,
+      "step": 18247
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7898880130498356e-05,
+      "loss": 0.4826,
+      "step": 18248
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.788890140117953e-05,
+      "loss": 0.4283,
+      "step": 18249
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.78789231815454e-05,
+      "loss": 0.5023,
+      "step": 18250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7868945471716694e-05,
+      "loss": 0.465,
+      "step": 18251
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.785896827181421e-05,
+      "loss": 0.4278,
+      "step": 18252
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7848991581958665e-05,
+      "loss": 0.5212,
+      "step": 18253
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7839015402270855e-05,
+      "loss": 0.4739,
+      "step": 18254
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.782903973287144e-05,
+      "loss": 0.4657,
+      "step": 18255
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7819064573881266e-05,
+      "loss": 0.3904,
+      "step": 18256
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.780908992542102e-05,
+      "loss": 0.4697,
+      "step": 18257
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.779911578761143e-05,
+      "loss": 0.4448,
+      "step": 18258
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.778914216057321e-05,
+      "loss": 0.5103,
+      "step": 18259
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.777916904442704e-05,
+      "loss": 0.433,
+      "step": 18260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.776919643929371e-05,
+      "loss": 0.5237,
+      "step": 18261
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.775922434529386e-05,
+      "loss": 0.5128,
+      "step": 18262
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.774925276254821e-05,
+      "loss": 0.5723,
+      "step": 18263
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.773928169117745e-05,
+      "loss": 0.4508,
+      "step": 18264
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.772931113130221e-05,
+      "loss": 0.5025,
+      "step": 18265
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.771934108304325e-05,
+      "loss": 0.5804,
+      "step": 18266
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.770937154652121e-05,
+      "loss": 0.5304,
+      "step": 18267
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.769940252185675e-05,
+      "loss": 0.497,
+      "step": 18268
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.768943400917053e-05,
+      "loss": 0.4277,
+      "step": 18269
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7679466008583137e-05,
+      "loss": 0.4616,
+      "step": 18270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7669498520215336e-05,
+      "loss": 0.5072,
+      "step": 18271
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.765953154418775e-05,
+      "loss": 0.4728,
+      "step": 18272
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.764956508062092e-05,
+      "loss": 0.6094,
+      "step": 18273
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.76395991296355e-05,
+      "loss": 0.4388,
+      "step": 18274
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.762963369135217e-05,
+      "loss": 0.5028,
+      "step": 18275
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.761966876589151e-05,
+      "loss": 0.5039,
+      "step": 18276
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.760970435337414e-05,
+      "loss": 0.5069,
+      "step": 18277
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.759974045392064e-05,
+      "loss": 0.4439,
+      "step": 18278
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7589777067651586e-05,
+      "loss": 0.4767,
+      "step": 18279
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.757981419468763e-05,
+      "loss": 0.4771,
+      "step": 18280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7569851835149344e-05,
+      "loss": 0.4622,
+      "step": 18281
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.755988998915729e-05,
+      "loss": 0.3775,
+      "step": 18282
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.754992865683203e-05,
+      "loss": 0.527,
+      "step": 18283
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.753996783829409e-05,
+      "loss": 0.5422,
+      "step": 18284
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.753000753366414e-05,
+      "loss": 0.4589,
+      "step": 18285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.752004774306267e-05,
+      "loss": 0.5504,
+      "step": 18286
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7510088466610214e-05,
+      "loss": 0.4782,
+      "step": 18287
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.750012970442733e-05,
+      "loss": 0.4361,
+      "step": 18288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.74901714566345e-05,
+      "loss": 0.4716,
+      "step": 18289
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.748021372335235e-05,
+      "loss": 0.5915,
+      "step": 18290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7470256504701347e-05,
+      "loss": 0.5028,
+      "step": 18291
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.746029980080201e-05,
+      "loss": 0.4278,
+      "step": 18292
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.745034361177486e-05,
+      "loss": 0.4672,
+      "step": 18293
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.744038793774038e-05,
+      "loss": 0.5183,
+      "step": 18294
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.743043277881903e-05,
+      "loss": 0.4745,
+      "step": 18295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.74204781351314e-05,
+      "loss": 0.5001,
+      "step": 18296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7410524006797915e-05,
+      "loss": 0.4438,
+      "step": 18297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.740057039393906e-05,
+      "loss": 0.5024,
+      "step": 18298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7390617296675294e-05,
+      "loss": 0.5581,
+      "step": 18299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7380664715127074e-05,
+      "loss": 0.4608,
+      "step": 18300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.737071264941492e-05,
+      "loss": 0.5087,
+      "step": 18301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7360761099659235e-05,
+      "loss": 0.5197,
+      "step": 18302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.735081006598048e-05,
+      "loss": 0.4678,
+      "step": 18303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7340859548499085e-05,
+      "loss": 0.4779,
+      "step": 18304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.733090954733547e-05,
+      "loss": 0.5192,
+      "step": 18305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.732096006261011e-05,
+      "loss": 0.4763,
+      "step": 18306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.73110110944434e-05,
+      "loss": 0.453,
+      "step": 18307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.730106264295576e-05,
+      "loss": 0.5641,
+      "step": 18308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.72911147082676e-05,
+      "loss": 0.4761,
+      "step": 18309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.728116729049928e-05,
+      "loss": 0.4325,
+      "step": 18310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.727122038977129e-05,
+      "loss": 0.502,
+      "step": 18311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.726127400620395e-05,
+      "loss": 0.5681,
+      "step": 18312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.725132813991766e-05,
+      "loss": 0.4443,
+      "step": 18313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.72413827910328e-05,
+      "loss": 0.4513,
+      "step": 18314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.723143795966972e-05,
+      "loss": 0.5056,
+      "step": 18315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.722149364594883e-05,
+      "loss": 0.4835,
+      "step": 18316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.721154984999047e-05,
+      "loss": 0.4239,
+      "step": 18317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7201606571915e-05,
+      "loss": 0.4577,
+      "step": 18318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.719166381184274e-05,
+      "loss": 0.4303,
+      "step": 18319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.718172156989402e-05,
+      "loss": 0.589,
+      "step": 18320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7171779846189244e-05,
+      "loss": 0.4503,
+      "step": 18321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7161838640848685e-05,
+      "loss": 0.402,
+      "step": 18322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7151897953992686e-05,
+      "loss": 0.4546,
+      "step": 18323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7141957785741554e-05,
+      "loss": 0.5071,
+      "step": 18324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7132018136215594e-05,
+      "loss": 0.5025,
+      "step": 18325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.712207900553508e-05,
+      "loss": 0.4901,
+      "step": 18326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.711214039382037e-05,
+      "loss": 0.5166,
+      "step": 18327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.710220230119173e-05,
+      "loss": 0.4897,
+      "step": 18328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.709226472776944e-05,
+      "loss": 0.5862,
+      "step": 18329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7082327673673785e-05,
+      "loss": 0.5611,
+      "step": 18330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.707239113902498e-05,
+      "loss": 0.3681,
+      "step": 18331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.706245512394338e-05,
+      "loss": 0.4552,
+      "step": 18332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.70525196285492e-05,
+      "loss": 0.4317,
+      "step": 18333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.704258465296271e-05,
+      "loss": 0.4457,
+      "step": 18334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7032650197304125e-05,
+      "loss": 0.5172,
+      "step": 18335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.702271626169368e-05,
+      "loss": 0.5242,
+      "step": 18336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.701278284625165e-05,
+      "loss": 0.4417,
+      "step": 18337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.700284995109826e-05,
+      "loss": 0.4328,
+      "step": 18338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.699291757635371e-05,
+      "loss": 0.481,
+      "step": 18339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.698298572213823e-05,
+      "loss": 0.4722,
+      "step": 18340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.697305438857196e-05,
+      "loss": 0.4543,
+      "step": 18341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.69631235757752e-05,
+      "loss": 0.5192,
+      "step": 18342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.695319328386811e-05,
+      "loss": 0.4695,
+      "step": 18343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.694326351297088e-05,
+      "loss": 0.5955,
+      "step": 18344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.693333426320368e-05,
+      "loss": 0.4603,
+      "step": 18345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6923405534686666e-05,
+      "loss": 0.5996,
+      "step": 18346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6913477327540085e-05,
+      "loss": 0.4034,
+      "step": 18347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.690354964188404e-05,
+      "loss": 0.4345,
+      "step": 18348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6893622477838716e-05,
+      "loss": 0.5634,
+      "step": 18349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.688369583552424e-05,
+      "loss": 0.5874,
+      "step": 18350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6873769715060734e-05,
+      "loss": 0.4857,
+      "step": 18351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.686384411656842e-05,
+      "loss": 0.5818,
+      "step": 18352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.685391904016738e-05,
+      "loss": 0.4924,
+      "step": 18353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.684399448597775e-05,
+      "loss": 0.4889,
+      "step": 18354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6834070454119635e-05,
+      "loss": 0.4711,
+      "step": 18355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.682414694471316e-05,
+      "loss": 0.3941,
+      "step": 18356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.681422395787841e-05,
+      "loss": 0.3636,
+      "step": 18357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.680430149373552e-05,
+      "loss": 0.4776,
+      "step": 18358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.67943795524046e-05,
+      "loss": 0.5227,
+      "step": 18359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6784458134005705e-05,
+      "loss": 0.5284,
+      "step": 18360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.677453723865891e-05,
+      "loss": 0.3953,
+      "step": 18361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6764616866484265e-05,
+      "loss": 0.4649,
+      "step": 18362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.675469701760192e-05,
+      "loss": 0.5039,
+      "step": 18363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6744777692131915e-05,
+      "loss": 0.4426,
+      "step": 18364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.673485889019427e-05,
+      "loss": 0.4464,
+      "step": 18365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.672494061190906e-05,
+      "loss": 0.5064,
+      "step": 18366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6715022857396296e-05,
+      "loss": 0.4321,
+      "step": 18367
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.670510562677607e-05,
+      "loss": 0.542,
+      "step": 18368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.669518892016839e-05,
+      "loss": 0.4331,
+      "step": 18369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.668527273769327e-05,
+      "loss": 0.3965,
+      "step": 18370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.667535707947075e-05,
+      "loss": 0.4903,
+      "step": 18371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.666544194562078e-05,
+      "loss": 0.5137,
+      "step": 18372
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.665552733626347e-05,
+      "loss": 0.4733,
+      "step": 18373
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.664561325151875e-05,
+      "loss": 0.536,
+      "step": 18374
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.663569969150664e-05,
+      "loss": 0.502,
+      "step": 18375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.66257866563471e-05,
+      "loss": 0.5347,
+      "step": 18376
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6615874146160106e-05,
+      "loss": 0.4944,
+      "step": 18377
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.660596216106568e-05,
+      "loss": 0.511,
+      "step": 18378
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.659605070118379e-05,
+      "loss": 0.5121,
+      "step": 18379
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.658613976663434e-05,
+      "loss": 0.4509,
+      "step": 18380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.657622935753734e-05,
+      "loss": 0.4331,
+      "step": 18381
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.656631947401266e-05,
+      "loss": 0.524,
+      "step": 18382
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.655641011618035e-05,
+      "loss": 0.4973,
+      "step": 18383
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.654650128416028e-05,
+      "loss": 0.4544,
+      "step": 18384
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.653659297807241e-05,
+      "loss": 0.483,
+      "step": 18385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6526685198036636e-05,
+      "loss": 0.4979,
+      "step": 18386
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.651677794417285e-05,
+      "loss": 0.4445,
+      "step": 18387
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.650687121660104e-05,
+      "loss": 0.4959,
+      "step": 18388
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.649696501544108e-05,
+      "loss": 0.4377,
+      "step": 18389
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6487059340812845e-05,
+      "loss": 0.4745,
+      "step": 18390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6477154192836255e-05,
+      "loss": 0.4635,
+      "step": 18391
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.646724957163116e-05,
+      "loss": 0.5272,
+      "step": 18392
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6457345477317434e-05,
+      "loss": 0.5071,
+      "step": 18393
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.644744191001501e-05,
+      "loss": 0.5249,
+      "step": 18394
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6437538869843735e-05,
+      "loss": 0.4612,
+      "step": 18395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.642763635692343e-05,
+      "loss": 0.458,
+      "step": 18396
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6417734371373986e-05,
+      "loss": 0.4948,
+      "step": 18397
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.640783291331519e-05,
+      "loss": 0.4674,
+      "step": 18398
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.639793198286696e-05,
+      "loss": 0.426,
+      "step": 18399
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6388031580149116e-05,
+      "loss": 0.5436,
+      "step": 18400
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 0.4682307839393616,
+      "eval_runtime": 845.2935,
+      "eval_samples_per_second": 28.463,
+      "eval_steps_per_second": 14.232,
+      "step": 18400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6378131705281455e-05,
+      "loss": 0.4314,
+      "step": 18401
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.636823235838382e-05,
+      "loss": 0.3964,
+      "step": 18402
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.635833353957597e-05,
+      "loss": 0.6029,
+      "step": 18403
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6348435248977806e-05,
+      "loss": 0.4779,
+      "step": 18404
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.633853748670907e-05,
+      "loss": 0.4449,
+      "step": 18405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.63286402528896e-05,
+      "loss": 0.4979,
+      "step": 18406
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6318743547639155e-05,
+      "loss": 0.5535,
+      "step": 18407
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.630884737107751e-05,
+      "loss": 0.4571,
+      "step": 18408
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.629895172332445e-05,
+      "loss": 0.4641,
+      "step": 18409
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.628905660449977e-05,
+      "loss": 0.5349,
+      "step": 18410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.62791620147232e-05,
+      "loss": 0.4453,
+      "step": 18411
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.626926795411447e-05,
+      "loss": 0.4653,
+      "step": 18412
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.625937442279341e-05,
+      "loss": 0.3922,
+      "step": 18413
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.624948142087973e-05,
+      "loss": 0.4761,
+      "step": 18414
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.623958894849317e-05,
+      "loss": 0.4219,
+      "step": 18415
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.622969700575345e-05,
+      "loss": 0.5343,
+      "step": 18416
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.621980559278026e-05,
+      "loss": 0.499,
+      "step": 18417
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.620991470969341e-05,
+      "loss": 0.3953,
+      "step": 18418
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6200024356612556e-05,
+      "loss": 0.5649,
+      "step": 18419
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6190134533657424e-05,
+      "loss": 0.4935,
+      "step": 18420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6180245240947695e-05,
+      "loss": 0.5571,
+      "step": 18421
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6170356478603024e-05,
+      "loss": 0.4186,
+      "step": 18422
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.616046824674319e-05,
+      "loss": 0.5311,
+      "step": 18423
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.615058054548783e-05,
+      "loss": 0.5292,
+      "step": 18424
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.614069337495662e-05,
+      "loss": 0.4971,
+      "step": 18425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6130806735269204e-05,
+      "loss": 0.5265,
+      "step": 18426
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.612092062654528e-05,
+      "loss": 0.4289,
+      "step": 18427
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.611103504890444e-05,
+      "loss": 0.4626,
+      "step": 18428
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6101150002466405e-05,
+      "loss": 0.5274,
+      "step": 18429
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.609126548735081e-05,
+      "loss": 0.4982,
+      "step": 18430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6081381503677255e-05,
+      "loss": 0.405,
+      "step": 18431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6071498051565395e-05,
+      "loss": 0.4237,
+      "step": 18432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6061615131134795e-05,
+      "loss": 0.422,
+      "step": 18433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6051732742505156e-05,
+      "loss": 0.5191,
+      "step": 18434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6041850885796054e-05,
+      "loss": 0.4727,
+      "step": 18435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.603196956112708e-05,
+      "loss": 0.502,
+      "step": 18436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6022088768617856e-05,
+      "loss": 0.4259,
+      "step": 18437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.601220850838791e-05,
+      "loss": 0.4898,
+      "step": 18438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6002328780556915e-05,
+      "loss": 0.4779,
+      "step": 18439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.59924495852444e-05,
+      "loss": 0.4276,
+      "step": 18440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.598257092256994e-05,
+      "loss": 0.5349,
+      "step": 18441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5972692792653105e-05,
+      "loss": 0.489,
+      "step": 18442
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5962815195613416e-05,
+      "loss": 0.4781,
+      "step": 18443
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.595293813157049e-05,
+      "loss": 0.6039,
+      "step": 18444
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.594306160064384e-05,
+      "loss": 0.4716,
+      "step": 18445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.593318560295302e-05,
+      "loss": 0.5007,
+      "step": 18446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.592331013861753e-05,
+      "loss": 0.4799,
+      "step": 18447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5913435207756895e-05,
+      "loss": 0.5464,
+      "step": 18448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5903560810490674e-05,
+      "loss": 0.415,
+      "step": 18449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5893686946938374e-05,
+      "loss": 0.4311,
+      "step": 18450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.588381361721948e-05,
+      "loss": 0.5248,
+      "step": 18451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.58739408214535e-05,
+      "loss": 0.4909,
+      "step": 18452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.58640685597599e-05,
+      "loss": 0.5233,
+      "step": 18453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.585419683225822e-05,
+      "loss": 0.4519,
+      "step": 18454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.584432563906792e-05,
+      "loss": 0.436,
+      "step": 18455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.583445498030848e-05,
+      "loss": 0.4537,
+      "step": 18456
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.582458485609936e-05,
+      "loss": 0.4954,
+      "step": 18457
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.581471526656e-05,
+      "loss": 0.4809,
+      "step": 18458
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5804846211809856e-05,
+      "loss": 0.511,
+      "step": 18459
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.579497769196842e-05,
+      "loss": 0.4933,
+      "step": 18460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.578510970715511e-05,
+      "loss": 0.5361,
+      "step": 18461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.577524225748938e-05,
+      "loss": 0.4197,
+      "step": 18462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.576537534309062e-05,
+      "loss": 0.5334,
+      "step": 18463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.575550896407824e-05,
+      "loss": 0.4641,
+      "step": 18464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.574564312057172e-05,
+      "loss": 0.4254,
+      "step": 18465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.573577781269045e-05,
+      "loss": 0.4853,
+      "step": 18466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.572591304055381e-05,
+      "loss": 0.4828,
+      "step": 18467
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.571604880428122e-05,
+      "loss": 0.4844,
+      "step": 18468
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5706185103992015e-05,
+      "loss": 0.5087,
+      "step": 18469
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.569632193980565e-05,
+      "loss": 0.517,
+      "step": 18470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5686459311841486e-05,
+      "loss": 0.4962,
+      "step": 18471
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5676597220218876e-05,
+      "loss": 0.4906,
+      "step": 18472
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.566673566505718e-05,
+      "loss": 0.4848,
+      "step": 18473
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.565687464647573e-05,
+      "loss": 0.5076,
+      "step": 18474
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.564701416459397e-05,
+      "loss": 0.5031,
+      "step": 18475
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.563715421953117e-05,
+      "loss": 0.4873,
+      "step": 18476
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.562729481140669e-05,
+      "loss": 0.611,
+      "step": 18477
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.561743594033986e-05,
+      "loss": 0.5318,
+      "step": 18478
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.560757760644996e-05,
+      "loss": 0.4689,
+      "step": 18479
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.559771980985638e-05,
+      "loss": 0.4858,
+      "step": 18480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.558786255067841e-05,
+      "loss": 0.5506,
+      "step": 18481
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.557800582903536e-05,
+      "loss": 0.422,
+      "step": 18482
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.556814964504651e-05,
+      "loss": 0.4164,
+      "step": 18483
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5558293998831125e-05,
+      "loss": 0.4646,
+      "step": 18484
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5548438890508555e-05,
+      "loss": 0.4887,
+      "step": 18485
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.553858432019805e-05,
+      "loss": 0.4259,
+      "step": 18486
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.55287302880189e-05,
+      "loss": 0.4801,
+      "step": 18487
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.551887679409035e-05,
+      "loss": 0.4859,
+      "step": 18488
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5509023838531626e-05,
+      "loss": 0.442,
+      "step": 18489
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.549917142146206e-05,
+      "loss": 0.4219,
+      "step": 18490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.548931954300086e-05,
+      "loss": 0.4538,
+      "step": 18491
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.547946820326727e-05,
+      "loss": 0.5426,
+      "step": 18492
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.546961740238052e-05,
+      "loss": 0.5345,
+      "step": 18493
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.545976714045984e-05,
+      "loss": 0.5596,
+      "step": 18494
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5449917417624417e-05,
+      "loss": 0.4403,
+      "step": 18495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.544006823399353e-05,
+      "loss": 0.4851,
+      "step": 18496
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.543021958968636e-05,
+      "loss": 0.4831,
+      "step": 18497
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5420371484822106e-05,
+      "loss": 0.5788,
+      "step": 18498
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.541052391951995e-05,
+      "loss": 0.491,
+      "step": 18499
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.540067689389907e-05,
+      "loss": 0.4726,
+      "step": 18500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5390830408078695e-05,
+      "loss": 0.5197,
+      "step": 18501
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.538098446217798e-05,
+      "loss": 0.5123,
+      "step": 18502
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.537113905631608e-05,
+      "loss": 0.4803,
+      "step": 18503
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.536129419061218e-05,
+      "loss": 0.5711,
+      "step": 18504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5351449865185365e-05,
+      "loss": 0.4408,
+      "step": 18505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5341606080154886e-05,
+      "loss": 0.3819,
+      "step": 18506
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.533176283563985e-05,
+      "loss": 0.4804,
+      "step": 18507
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5321920131759374e-05,
+      "loss": 0.5224,
+      "step": 18508
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.53120779686326e-05,
+      "loss": 0.5958,
+      "step": 18509
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.53022363463786e-05,
+      "loss": 0.453,
+      "step": 18510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.529239526511658e-05,
+      "loss": 0.4374,
+      "step": 18511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.528255472496562e-05,
+      "loss": 0.4299,
+      "step": 18512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.52727147260448e-05,
+      "loss": 0.4622,
+      "step": 18513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5262875268473226e-05,
+      "loss": 0.4993,
+      "step": 18514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.525303635236996e-05,
+      "loss": 0.5304,
+      "step": 18515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5243197977854155e-05,
+      "loss": 0.5248,
+      "step": 18516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5233360145044834e-05,
+      "loss": 0.4457,
+      "step": 18517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5223522854061096e-05,
+      "loss": 0.448,
+      "step": 18518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.521368610502199e-05,
+      "loss": 0.4369,
+      "step": 18519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.520384989804653e-05,
+      "loss": 0.4677,
+      "step": 18520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.519401423325385e-05,
+      "loss": 0.5074,
+      "step": 18521
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.518417911076297e-05,
+      "loss": 0.49,
+      "step": 18522
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.51743445306929e-05,
+      "loss": 0.5261,
+      "step": 18523
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.516451049316269e-05,
+      "loss": 0.5055,
+      "step": 18524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.515467699829138e-05,
+      "loss": 0.4154,
+      "step": 18525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5144844046197897e-05,
+      "loss": 0.4748,
+      "step": 18526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.513501163700138e-05,
+      "loss": 0.4522,
+      "step": 18527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5125179770820766e-05,
+      "loss": 0.4393,
+      "step": 18528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.511534844777507e-05,
+      "loss": 0.4616,
+      "step": 18529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5105517667983283e-05,
+      "loss": 0.5254,
+      "step": 18530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.509568743156435e-05,
+      "loss": 0.5231,
+      "step": 18531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.508585773863732e-05,
+      "loss": 0.4744,
+      "step": 18532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.507602858932113e-05,
+      "loss": 0.5009,
+      "step": 18533
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5066199983734736e-05,
+      "loss": 0.5255,
+      "step": 18534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5056371921997116e-05,
+      "loss": 0.4906,
+      "step": 18535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.504654440422717e-05,
+      "loss": 0.4133,
+      "step": 18536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.503671743054393e-05,
+      "loss": 0.4449,
+      "step": 18537
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5026891001066295e-05,
+      "loss": 0.4683,
+      "step": 18538
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.501706511591319e-05,
+      "loss": 0.437,
+      "step": 18539
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.500723977520355e-05,
+      "loss": 0.481,
+      "step": 18540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.499741497905628e-05,
+      "loss": 0.6202,
+      "step": 18541
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.498759072759031e-05,
+      "loss": 0.5221,
+      "step": 18542
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.497776702092454e-05,
+      "loss": 0.5303,
+      "step": 18543
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.496794385917786e-05,
+      "loss": 0.4279,
+      "step": 18544
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.495812124246914e-05,
+      "loss": 0.4625,
+      "step": 18545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4948299170917325e-05,
+      "loss": 0.5237,
+      "step": 18546
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.493847764464128e-05,
+      "loss": 0.4924,
+      "step": 18547
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.492865666375987e-05,
+      "loss": 0.4356,
+      "step": 18548
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.491883622839196e-05,
+      "loss": 0.5709,
+      "step": 18549
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.490901633865635e-05,
+      "loss": 0.4498,
+      "step": 18550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4899196994671996e-05,
+      "loss": 0.4322,
+      "step": 18551
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4889378196557705e-05,
+      "loss": 0.458,
+      "step": 18552
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4879559944432314e-05,
+      "loss": 0.4764,
+      "step": 18553
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4869742238414644e-05,
+      "loss": 0.5118,
+      "step": 18554
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.48599250786235e-05,
+      "loss": 0.4313,
+      "step": 18555
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4850108465177776e-05,
+      "loss": 0.5156,
+      "step": 18556
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4840292398196235e-05,
+      "loss": 0.572,
+      "step": 18557
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4830476877797696e-05,
+      "loss": 0.4632,
+      "step": 18558
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.482066190410097e-05,
+      "loss": 0.4888,
+      "step": 18559
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.481084747722482e-05,
+      "loss": 0.4536,
+      "step": 18560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.480103359728803e-05,
+      "loss": 0.5221,
+      "step": 18561
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.479122026440944e-05,
+      "loss": 0.4616,
+      "step": 18562
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.478140747870777e-05,
+      "loss": 0.5092,
+      "step": 18563
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.477159524030182e-05,
+      "loss": 0.4674,
+      "step": 18564
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.476178354931032e-05,
+      "loss": 0.4575,
+      "step": 18565
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4751972405852005e-05,
+      "loss": 0.3895,
+      "step": 18566
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.47421618100457e-05,
+      "loss": 0.474,
+      "step": 18567
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4732351762010095e-05,
+      "loss": 0.4617,
+      "step": 18568
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.472254226186393e-05,
+      "loss": 0.4014,
+      "step": 18569
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.471273330972594e-05,
+      "loss": 0.5317,
+      "step": 18570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.47029249057148e-05,
+      "loss": 0.4761,
+      "step": 18571
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4693117049949295e-05,
+      "loss": 0.4738,
+      "step": 18572
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4683309742548116e-05,
+      "loss": 0.4822,
+      "step": 18573
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.467350298362994e-05,
+      "loss": 0.3782,
+      "step": 18574
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4663696773313466e-05,
+      "loss": 0.4379,
+      "step": 18575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.465389111171736e-05,
+      "loss": 0.4408,
+      "step": 18576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.464408599896036e-05,
+      "loss": 0.4951,
+      "step": 18577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.463428143516112e-05,
+      "loss": 0.5566,
+      "step": 18578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4624477420438304e-05,
+      "loss": 0.4192,
+      "step": 18579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.461467395491056e-05,
+      "loss": 0.51,
+      "step": 18580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.46048710386965e-05,
+      "loss": 0.4472,
+      "step": 18581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.459506867191488e-05,
+      "loss": 0.4223,
+      "step": 18582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.458526685468428e-05,
+      "loss": 0.458,
+      "step": 18583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.457546558712333e-05,
+      "loss": 0.4073,
+      "step": 18584
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.456566486935067e-05,
+      "loss": 0.552,
+      "step": 18585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.455586470148488e-05,
+      "loss": 0.3834,
+      "step": 18586
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.454606508364466e-05,
+      "loss": 0.5156,
+      "step": 18587
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.453626601594857e-05,
+      "loss": 0.4149,
+      "step": 18588
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.45264674985152e-05,
+      "loss": 0.5041,
+      "step": 18589
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4516669531463174e-05,
+      "loss": 0.5592,
+      "step": 18590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.450687211491101e-05,
+      "loss": 0.5049,
+      "step": 18591
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.449707524897738e-05,
+      "loss": 0.4741,
+      "step": 18592
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4487278933780826e-05,
+      "loss": 0.5278,
+      "step": 18593
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4477483169439924e-05,
+      "loss": 0.5455,
+      "step": 18594
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.44676879560732e-05,
+      "loss": 0.5205,
+      "step": 18595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4457893293799246e-05,
+      "loss": 0.5498,
+      "step": 18596
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4448099182736545e-05,
+      "loss": 0.4406,
+      "step": 18597
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.443830562300374e-05,
+      "loss": 0.482,
+      "step": 18598
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.442851261471931e-05,
+      "loss": 0.4605,
+      "step": 18599
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.441872015800179e-05,
+      "loss": 0.3881,
+      "step": 18600
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 0.46492907404899597,
+      "eval_runtime": 843.6337,
+      "eval_samples_per_second": 28.519,
+      "eval_steps_per_second": 14.26,
+      "step": 18600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.440892825296971e-05,
+      "loss": 0.4523,
+      "step": 18601
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.439913689974151e-05,
+      "loss": 0.5602,
+      "step": 18602
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.438934609843581e-05,
+      "loss": 0.4681,
+      "step": 18603
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.437955584917107e-05,
+      "loss": 0.5071,
+      "step": 18604
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.436976615206577e-05,
+      "loss": 0.4466,
+      "step": 18605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.43599770072384e-05,
+      "loss": 0.5027,
+      "step": 18606
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.435018841480744e-05,
+      "loss": 0.5114,
+      "step": 18607
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4340400374891374e-05,
+      "loss": 0.4118,
+      "step": 18608
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4330612887608676e-05,
+      "loss": 0.4409,
+      "step": 18609
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.432082595307781e-05,
+      "loss": 0.4979,
+      "step": 18610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.431103957141721e-05,
+      "loss": 0.4824,
+      "step": 18611
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.430125374274531e-05,
+      "loss": 0.5986,
+      "step": 18612
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.429146846718058e-05,
+      "loss": 0.3514,
+      "step": 18613
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4281683744841484e-05,
+      "loss": 0.4728,
+      "step": 18614
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.42718995758464e-05,
+      "loss": 0.5327,
+      "step": 18615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4262115960313756e-05,
+      "loss": 0.4827,
+      "step": 18616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4252332898361934e-05,
+      "loss": 0.5135,
+      "step": 18617
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.424255039010944e-05,
+      "loss": 0.4353,
+      "step": 18618
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4232768435674594e-05,
+      "loss": 0.4302,
+      "step": 18619
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.422298703517583e-05,
+      "loss": 0.4919,
+      "step": 18620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.421320618873151e-05,
+      "loss": 0.4866,
+      "step": 18621
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.420342589645998e-05,
+      "loss": 0.5141,
+      "step": 18622
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.41936461584797e-05,
+      "loss": 0.4958,
+      "step": 18623
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4183866974909005e-05,
+      "loss": 0.439,
+      "step": 18624
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.417408834586623e-05,
+      "loss": 0.4594,
+      "step": 18625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4164310271469756e-05,
+      "loss": 0.448,
+      "step": 18626
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.415453275183791e-05,
+      "loss": 0.394,
+      "step": 18627
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.414475578708901e-05,
+      "loss": 0.4225,
+      "step": 18628
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4134979377341454e-05,
+      "loss": 0.4702,
+      "step": 18629
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.412520352271353e-05,
+      "loss": 0.4778,
+      "step": 18630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.411542822332358e-05,
+      "loss": 0.4646,
+      "step": 18631
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.41056534792899e-05,
+      "loss": 0.4271,
+      "step": 18632
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4095879290730745e-05,
+      "loss": 0.4322,
+      "step": 18633
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.408610565776453e-05,
+      "loss": 0.5169,
+      "step": 18634
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.407633258050948e-05,
+      "loss": 0.5374,
+      "step": 18635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.406656005908388e-05,
+      "loss": 0.5015,
+      "step": 18636
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.405678809360604e-05,
+      "loss": 0.5482,
+      "step": 18637
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.404701668419415e-05,
+      "loss": 0.5647,
+      "step": 18638
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.403724583096659e-05,
+      "loss": 0.4294,
+      "step": 18639
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.402747553404157e-05,
+      "loss": 0.4827,
+      "step": 18640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.401770579353735e-05,
+      "loss": 0.5102,
+      "step": 18641
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.400793660957216e-05,
+      "loss": 0.4962,
+      "step": 18642
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.399816798226423e-05,
+      "loss": 0.5336,
+      "step": 18643
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.398839991173182e-05,
+      "loss": 0.4888,
+      "step": 18644
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3978632398093165e-05,
+      "loss": 0.5008,
+      "step": 18645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3968865441466466e-05,
+      "loss": 0.4691,
+      "step": 18646
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.395909904196994e-05,
+      "loss": 0.5045,
+      "step": 18647
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3949333199721755e-05,
+      "loss": 0.5364,
+      "step": 18648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.393956791484018e-05,
+      "loss": 0.4316,
+      "step": 18649
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.392980318744337e-05,
+      "loss": 0.4492,
+      "step": 18650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.392003901764953e-05,
+      "loss": 0.4104,
+      "step": 18651
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3910275405576807e-05,
+      "loss": 0.5254,
+      "step": 18652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.390051235134335e-05,
+      "loss": 0.5293,
+      "step": 18653
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.38907498550674e-05,
+      "loss": 0.505,
+      "step": 18654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3880987916867086e-05,
+      "loss": 0.4904,
+      "step": 18655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3871226536860566e-05,
+      "loss": 0.4806,
+      "step": 18656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.386146571516595e-05,
+      "loss": 0.5178,
+      "step": 18657
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3851705451901413e-05,
+      "loss": 0.4296,
+      "step": 18658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.384194574718504e-05,
+      "loss": 0.5258,
+      "step": 18659
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.383218660113502e-05,
+      "loss": 0.3909,
+      "step": 18660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.382242801386943e-05,
+      "loss": 0.4793,
+      "step": 18661
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3812669985506404e-05,
+      "loss": 0.3983,
+      "step": 18662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.380291251616403e-05,
+      "loss": 0.4148,
+      "step": 18663
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.379315560596038e-05,
+      "loss": 0.4677,
+      "step": 18664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.37833992550136e-05,
+      "loss": 0.4097,
+      "step": 18665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.377364346344176e-05,
+      "loss": 0.4222,
+      "step": 18666
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3763888231362925e-05,
+      "loss": 0.4961,
+      "step": 18667
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.375413355889516e-05,
+      "loss": 0.4338,
+      "step": 18668
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3744379446156515e-05,
+      "loss": 0.402,
+      "step": 18669
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.37346258932651e-05,
+      "loss": 0.4567,
+      "step": 18670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.372487290033893e-05,
+      "loss": 0.4848,
+      "step": 18671
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.371512046749606e-05,
+      "loss": 0.4708,
+      "step": 18672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.370536859485451e-05,
+      "loss": 0.453,
+      "step": 18673
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.369561728253229e-05,
+      "loss": 0.3958,
+      "step": 18674
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.36858665306475e-05,
+      "loss": 0.3993,
+      "step": 18675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.367611633931815e-05,
+      "loss": 0.4877,
+      "step": 18676
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3666366708662165e-05,
+      "loss": 0.426,
+      "step": 18677
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3656617638797545e-05,
+      "loss": 0.5145,
+      "step": 18678
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.364686912984238e-05,
+      "loss": 0.5397,
+      "step": 18679
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.363712118191463e-05,
+      "loss": 0.5139,
+      "step": 18680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.362737379513225e-05,
+      "loss": 0.3677,
+      "step": 18681
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3617626969613235e-05,
+      "loss": 0.5167,
+      "step": 18682
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.360788070547549e-05,
+      "loss": 0.4379,
+      "step": 18683
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.359813500283708e-05,
+      "loss": 0.4621,
+      "step": 18684
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.358838986181592e-05,
+      "loss": 0.5616,
+      "step": 18685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.357864528252996e-05,
+      "loss": 0.4578,
+      "step": 18686
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3568901265097125e-05,
+      "loss": 0.4358,
+      "step": 18687
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.355915780963532e-05,
+      "loss": 0.4616,
+      "step": 18688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.354941491626256e-05,
+      "loss": 0.4221,
+      "step": 18689
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.353967258509671e-05,
+      "loss": 0.4484,
+      "step": 18690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.352993081625571e-05,
+      "loss": 0.4569,
+      "step": 18691
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3520189609857454e-05,
+      "loss": 0.5649,
+      "step": 18692
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3510448966019785e-05,
+      "loss": 0.4313,
+      "step": 18693
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3500708884860696e-05,
+      "loss": 0.5114,
+      "step": 18694
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3490969366498046e-05,
+      "loss": 0.4223,
+      "step": 18695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.348123041104969e-05,
+      "loss": 0.4235,
+      "step": 18696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.347149201863353e-05,
+      "loss": 0.5091,
+      "step": 18697
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.34617541893674e-05,
+      "loss": 0.487,
+      "step": 18698
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.345201692336914e-05,
+      "loss": 0.3961,
+      "step": 18699
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.344228022075668e-05,
+      "loss": 0.4696,
+      "step": 18700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.343254408164784e-05,
+      "loss": 0.5813,
+      "step": 18701
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3422808506160436e-05,
+      "loss": 0.5289,
+      "step": 18702
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3413073494412316e-05,
+      "loss": 0.5708,
+      "step": 18703
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.340333904652126e-05,
+      "loss": 0.4627,
+      "step": 18704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.339360516260516e-05,
+      "loss": 0.5509,
+      "step": 18705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.338387184278182e-05,
+      "loss": 0.5105,
+      "step": 18706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3374139087168995e-05,
+      "loss": 0.4467,
+      "step": 18707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.336440689588453e-05,
+      "loss": 0.528,
+      "step": 18708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.335467526904616e-05,
+      "loss": 0.4614,
+      "step": 18709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.334494420677174e-05,
+      "loss": 0.4849,
+      "step": 18710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.333521370917902e-05,
+      "loss": 0.4397,
+      "step": 18711
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3325483776385774e-05,
+      "loss": 0.4777,
+      "step": 18712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3315754408509756e-05,
+      "loss": 0.4498,
+      "step": 18713
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3306025605668696e-05,
+      "loss": 0.4754,
+      "step": 18714
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3296297367980404e-05,
+      "loss": 0.4739,
+      "step": 18715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3286569695562613e-05,
+      "loss": 0.5209,
+      "step": 18716
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3276842588533047e-05,
+      "loss": 0.5042,
+      "step": 18717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.326711604700942e-05,
+      "loss": 0.4583,
+      "step": 18718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.325739007110945e-05,
+      "loss": 0.4396,
+      "step": 18719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3247664660950905e-05,
+      "loss": 0.3532,
+      "step": 18720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.323793981665146e-05,
+      "loss": 0.4865,
+      "step": 18721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3228215538328816e-05,
+      "loss": 0.3998,
+      "step": 18722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.321849182610069e-05,
+      "loss": 0.6418,
+      "step": 18723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.320876868008471e-05,
+      "loss": 0.4359,
+      "step": 18724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3199046100398655e-05,
+      "loss": 0.5013,
+      "step": 18725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.318932408716013e-05,
+      "loss": 0.5091,
+      "step": 18726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.317960264048685e-05,
+      "loss": 0.4945,
+      "step": 18727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.316988176049642e-05,
+      "loss": 0.4141,
+      "step": 18728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.316016144730655e-05,
+      "loss": 0.4214,
+      "step": 18729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.315044170103482e-05,
+      "loss": 0.4452,
+      "step": 18730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.314072252179896e-05,
+      "loss": 0.4648,
+      "step": 18731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.313100390971655e-05,
+      "loss": 0.4942,
+      "step": 18732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3121285864905235e-05,
+      "loss": 0.3682,
+      "step": 18733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3111568387482636e-05,
+      "loss": 0.5021,
+      "step": 18734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.31018514775663e-05,
+      "loss": 0.4705,
+      "step": 18735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3092135135273955e-05,
+      "loss": 0.5024,
+      "step": 18736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.308241936072313e-05,
+      "loss": 0.4662,
+      "step": 18737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3072704154031425e-05,
+      "loss": 0.4794,
+      "step": 18738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.306298951531643e-05,
+      "loss": 0.4832,
+      "step": 18739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.305327544469569e-05,
+      "loss": 0.5384,
+      "step": 18740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.304356194228686e-05,
+      "loss": 0.4722,
+      "step": 18741
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.303384900820744e-05,
+      "loss": 0.4614,
+      "step": 18742
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.302413664257501e-05,
+      "loss": 0.5075,
+      "step": 18743
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.301442484550711e-05,
+      "loss": 0.4785,
+      "step": 18744
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.300471361712127e-05,
+      "loss": 0.4871,
+      "step": 18745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.299500295753509e-05,
+      "loss": 0.6529,
+      "step": 18746
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.298529286686607e-05,
+      "loss": 0.484,
+      "step": 18747
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2975583345231705e-05,
+      "loss": 0.4696,
+      "step": 18748
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.296587439274955e-05,
+      "loss": 0.4922,
+      "step": 18749
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2956166009537056e-05,
+      "loss": 0.5249,
+      "step": 18750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.294645819571181e-05,
+      "loss": 0.5447,
+      "step": 18751
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2936750951391277e-05,
+      "loss": 0.4565,
+      "step": 18752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2927044276692925e-05,
+      "loss": 0.4207,
+      "step": 18753
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2917338171734276e-05,
+      "loss": 0.56,
+      "step": 18754
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2907632636632723e-05,
+      "loss": 0.4553,
+      "step": 18755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2897927671505855e-05,
+      "loss": 0.4979,
+      "step": 18756
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.288822327647105e-05,
+      "loss": 0.5267,
+      "step": 18757
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.287851945164581e-05,
+      "loss": 0.5208,
+      "step": 18758
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2868816197147543e-05,
+      "loss": 0.5182,
+      "step": 18759
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.285911351309373e-05,
+      "loss": 0.4899,
+      "step": 18760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.284941139960173e-05,
+      "loss": 0.4194,
+      "step": 18761
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.283970985678908e-05,
+      "loss": 0.4238,
+      "step": 18762
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.283000888477313e-05,
+      "loss": 0.4505,
+      "step": 18763
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.282030848367132e-05,
+      "loss": 0.4575,
+      "step": 18764
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.281060865360105e-05,
+      "loss": 0.4985,
+      "step": 18765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.280090939467967e-05,
+      "loss": 0.4005,
+      "step": 18766
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.279121070702466e-05,
+      "loss": 0.5135,
+      "step": 18767
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.278151259075339e-05,
+      "loss": 0.4923,
+      "step": 18768
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2771815045983206e-05,
+      "loss": 0.5085,
+      "step": 18769
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.276211807283149e-05,
+      "loss": 0.4391,
+      "step": 18770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2752421671415564e-05,
+      "loss": 0.4666,
+      "step": 18771
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2742725841852894e-05,
+      "loss": 0.5361,
+      "step": 18772
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2733030584260755e-05,
+      "loss": 0.5085,
+      "step": 18773
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.272333589875651e-05,
+      "loss": 0.4531,
+      "step": 18774
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.27136417854575e-05,
+      "loss": 0.4334,
+      "step": 18775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2703948244481e-05,
+      "loss": 0.4662,
+      "step": 18776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.269425527594444e-05,
+      "loss": 0.45,
+      "step": 18777
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.268456287996507e-05,
+      "loss": 0.4116,
+      "step": 18778
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.267487105666022e-05,
+      "loss": 0.4715,
+      "step": 18779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.266517980614718e-05,
+      "loss": 0.5913,
+      "step": 18780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.265548912854322e-05,
+      "loss": 0.4851,
+      "step": 18781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.26457990239657e-05,
+      "loss": 0.5532,
+      "step": 18782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.263610949253186e-05,
+      "loss": 0.4083,
+      "step": 18783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.262642053435899e-05,
+      "loss": 0.4783,
+      "step": 18784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.261673214956434e-05,
+      "loss": 0.4524,
+      "step": 18785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2607044338265155e-05,
+      "loss": 0.454,
+      "step": 18786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.259735710057874e-05,
+      "loss": 0.5379,
+      "step": 18787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.258767043662232e-05,
+      "loss": 0.5218,
+      "step": 18788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.257798434651313e-05,
+      "loss": 0.5368,
+      "step": 18789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2568298830368425e-05,
+      "loss": 0.3522,
+      "step": 18790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2558613888305364e-05,
+      "loss": 0.4842,
+      "step": 18791
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.254892952044126e-05,
+      "loss": 0.4927,
+      "step": 18792
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.253924572689328e-05,
+      "loss": 0.5942,
+      "step": 18793
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.252956250777865e-05,
+      "loss": 0.5034,
+      "step": 18794
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.251987986321454e-05,
+      "loss": 0.4626,
+      "step": 18795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.251019779331816e-05,
+      "loss": 0.4614,
+      "step": 18796
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.250051629820665e-05,
+      "loss": 0.4215,
+      "step": 18797
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2490835377997273e-05,
+      "loss": 0.5587,
+      "step": 18798
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.248115503280715e-05,
+      "loss": 0.4932,
+      "step": 18799
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.247147526275347e-05,
+      "loss": 0.6051,
+      "step": 18800
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 0.4622979462146759,
+      "eval_runtime": 845.2335,
+      "eval_samples_per_second": 28.466,
+      "eval_steps_per_second": 14.233,
+      "step": 18800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.246179606795335e-05,
+      "loss": 0.4468,
+      "step": 18801
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.245211744852394e-05,
+      "loss": 0.5365,
+      "step": 18802
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2442439404582446e-05,
+      "loss": 0.4236,
+      "step": 18803
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.243276193624596e-05,
+      "loss": 0.4873,
+      "step": 18804
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2423085043631624e-05,
+      "loss": 0.4901,
+      "step": 18805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.241340872685654e-05,
+      "loss": 0.4529,
+      "step": 18806
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.240373298603781e-05,
+      "loss": 0.4357,
+      "step": 18807
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.239405782129261e-05,
+      "loss": 0.4186,
+      "step": 18808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2384383232738e-05,
+      "loss": 0.4462,
+      "step": 18809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.237470922049108e-05,
+      "loss": 0.4781,
+      "step": 18810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2365035784668934e-05,
+      "loss": 0.494,
+      "step": 18811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.235536292538865e-05,
+      "loss": 0.5173,
+      "step": 18812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.234569064276729e-05,
+      "loss": 0.467,
+      "step": 18813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2336018936921924e-05,
+      "loss": 0.4417,
+      "step": 18814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.232634780796961e-05,
+      "loss": 0.4099,
+      "step": 18815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.231667725602738e-05,
+      "loss": 0.4815,
+      "step": 18816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.230700728121234e-05,
+      "loss": 0.3954,
+      "step": 18817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.229733788364149e-05,
+      "loss": 0.568,
+      "step": 18818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.228766906343186e-05,
+      "loss": 0.4751,
+      "step": 18819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.227800082070049e-05,
+      "loss": 0.4589,
+      "step": 18820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.226833315556434e-05,
+      "loss": 0.5139,
+      "step": 18821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2258666068140516e-05,
+      "loss": 0.5361,
+      "step": 18822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2248999558545985e-05,
+      "loss": 0.5387,
+      "step": 18823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.223933362689773e-05,
+      "loss": 0.5334,
+      "step": 18824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.222966827331273e-05,
+      "loss": 0.4614,
+      "step": 18825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.222000349790797e-05,
+      "loss": 0.5002,
+      "step": 18826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.221033930080046e-05,
+      "loss": 0.5752,
+      "step": 18827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.220067568210715e-05,
+      "loss": 0.4637,
+      "step": 18828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2191012641945e-05,
+      "loss": 0.52,
+      "step": 18829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.218135018043098e-05,
+      "loss": 0.4415,
+      "step": 18830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.217168829768202e-05,
+      "loss": 0.427,
+      "step": 18831
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.216202699381501e-05,
+      "loss": 0.4855,
+      "step": 18832
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2152366268946994e-05,
+      "loss": 0.5433,
+      "step": 18833
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.214270612319484e-05,
+      "loss": 0.4805,
+      "step": 18834
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.213304655667547e-05,
+      "loss": 0.5405,
+      "step": 18835
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.21233875695058e-05,
+      "loss": 0.547,
+      "step": 18836
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2113729161802695e-05,
+      "loss": 0.3931,
+      "step": 18837
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2104071333683136e-05,
+      "loss": 0.4975,
+      "step": 18838
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.209441408526397e-05,
+      "loss": 0.4419,
+      "step": 18839
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2084757416662075e-05,
+      "loss": 0.4803,
+      "step": 18840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.207510132799436e-05,
+      "loss": 0.4495,
+      "step": 18841
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2065445819377614e-05,
+      "loss": 0.4775,
+      "step": 18842
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.20557908909288e-05,
+      "loss": 0.5349,
+      "step": 18843
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2046136542764755e-05,
+      "loss": 0.5451,
+      "step": 18844
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.20364827750023e-05,
+      "loss": 0.4664,
+      "step": 18845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2026829587758284e-05,
+      "loss": 0.4835,
+      "step": 18846
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.201717698114951e-05,
+      "loss": 0.4317,
+      "step": 18847
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2007524955292884e-05,
+      "loss": 0.468,
+      "step": 18848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.199787351030518e-05,
+      "loss": 0.5033,
+      "step": 18849
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.198822264630322e-05,
+      "loss": 0.3779,
+      "step": 18850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.197857236340382e-05,
+      "loss": 0.5043,
+      "step": 18851
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.196892266172373e-05,
+      "loss": 0.4738,
+      "step": 18852
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1959273541379816e-05,
+      "loss": 0.4945,
+      "step": 18853
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.194962500248883e-05,
+      "loss": 0.4817,
+      "step": 18854
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.193997704516756e-05,
+      "loss": 0.4521,
+      "step": 18855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.193032966953279e-05,
+      "loss": 0.4284,
+      "step": 18856
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.192068287570121e-05,
+      "loss": 0.4544,
+      "step": 18857
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.191103666378968e-05,
+      "loss": 0.4351,
+      "step": 18858
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1901391033914914e-05,
+      "loss": 0.4841,
+      "step": 18859
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.189174598619364e-05,
+      "loss": 0.5103,
+      "step": 18860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.188210152074262e-05,
+      "loss": 0.4754,
+      "step": 18861
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.187245763767856e-05,
+      "loss": 0.4271,
+      "step": 18862
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.186281433711816e-05,
+      "loss": 0.4871,
+      "step": 18863
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.185317161917821e-05,
+      "loss": 0.4489,
+      "step": 18864
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.184352948397537e-05,
+      "loss": 0.5201,
+      "step": 18865
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1833887931626354e-05,
+      "loss": 0.3883,
+      "step": 18866
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1824246962247856e-05,
+      "loss": 0.433,
+      "step": 18867
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1814606575956526e-05,
+      "loss": 0.4701,
+      "step": 18868
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.180496677286911e-05,
+      "loss": 0.4552,
+      "step": 18869
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.179532755310226e-05,
+      "loss": 0.4913,
+      "step": 18870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.178568891677262e-05,
+      "loss": 0.4916,
+      "step": 18871
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1776050863996886e-05,
+      "loss": 0.4047,
+      "step": 18872
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1766413394891634e-05,
+      "loss": 0.417,
+      "step": 18873
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.17567765095736e-05,
+      "loss": 0.3883,
+      "step": 18874
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1747140208159405e-05,
+      "loss": 0.4925,
+      "step": 18875
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.173750449076564e-05,
+      "loss": 0.4388,
+      "step": 18876
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1727869357508977e-05,
+      "loss": 0.579,
+      "step": 18877
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.171823480850595e-05,
+      "loss": 0.4546,
+      "step": 18878
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1708600843873265e-05,
+      "loss": 0.4128,
+      "step": 18879
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.169896746372749e-05,
+      "loss": 0.5013,
+      "step": 18880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.168933466818522e-05,
+      "loss": 0.4508,
+      "step": 18881
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.167970245736304e-05,
+      "loss": 0.4638,
+      "step": 18882
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.167007083137749e-05,
+      "loss": 0.4355,
+      "step": 18883
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.166043979034524e-05,
+      "loss": 0.4805,
+      "step": 18884
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.16508093343828e-05,
+      "loss": 0.4163,
+      "step": 18885
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.164117946360674e-05,
+      "loss": 0.4848,
+      "step": 18886
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.163155017813359e-05,
+      "loss": 0.4637,
+      "step": 18887
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1621921478079895e-05,
+      "loss": 0.4999,
+      "step": 18888
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1612293363562256e-05,
+      "loss": 0.4697,
+      "step": 18889
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.160266583469717e-05,
+      "loss": 0.537,
+      "step": 18890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.159303889160114e-05,
+      "loss": 0.4455,
+      "step": 18891
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.158341253439072e-05,
+      "loss": 0.4243,
+      "step": 18892
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.157378676318235e-05,
+      "loss": 0.5005,
+      "step": 18893
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.156416157809263e-05,
+      "loss": 0.4875,
+      "step": 18894
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.155453697923802e-05,
+      "loss": 0.4615,
+      "step": 18895
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1544912966734994e-05,
+      "loss": 0.5232,
+      "step": 18896
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.153528954070004e-05,
+      "loss": 0.5035,
+      "step": 18897
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.152566670124965e-05,
+      "loss": 0.4592,
+      "step": 18898
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.151604444850023e-05,
+      "loss": 0.3937,
+      "step": 18899
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1506422782568345e-05,
+      "loss": 0.4295,
+      "step": 18900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.149680170357039e-05,
+      "loss": 0.4325,
+      "step": 18901
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.148718121162279e-05,
+      "loss": 0.5199,
+      "step": 18902
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1477561306842046e-05,
+      "loss": 0.4955,
+      "step": 18903
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1467941989344506e-05,
+      "loss": 0.4161,
+      "step": 18904
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.145832325924669e-05,
+      "loss": 0.4421,
+      "step": 18905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.144870511666497e-05,
+      "loss": 0.5119,
+      "step": 18906
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.143908756171576e-05,
+      "loss": 0.4485,
+      "step": 18907
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.142947059451546e-05,
+      "loss": 0.4399,
+      "step": 18908
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.141985421518045e-05,
+      "loss": 0.4607,
+      "step": 18909
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1410238423827174e-05,
+      "loss": 0.4691,
+      "step": 18910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.140062322057199e-05,
+      "loss": 0.4865,
+      "step": 18911
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1391008605531255e-05,
+      "loss": 0.5444,
+      "step": 18912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.138139457882136e-05,
+      "loss": 0.3824,
+      "step": 18913
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.137178114055863e-05,
+      "loss": 0.4615,
+      "step": 18914
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.136216829085947e-05,
+      "loss": 0.4986,
+      "step": 18915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.135255602984022e-05,
+      "loss": 0.5164,
+      "step": 18916
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1342944357617195e-05,
+      "loss": 0.4796,
+      "step": 18917
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1333333274306737e-05,
+      "loss": 0.4477,
+      "step": 18918
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.132372278002514e-05,
+      "loss": 0.4696,
+      "step": 18919
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1314112874888795e-05,
+      "loss": 0.5655,
+      "step": 18920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.130450355901398e-05,
+      "loss": 0.5107,
+      "step": 18921
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.129489483251699e-05,
+      "loss": 0.4948,
+      "step": 18922
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.128528669551413e-05,
+      "loss": 0.4782,
+      "step": 18923
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.127567914812166e-05,
+      "loss": 0.4517,
+      "step": 18924
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.126607219045592e-05,
+      "loss": 0.5655,
+      "step": 18925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.125646582263317e-05,
+      "loss": 0.4239,
+      "step": 18926
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1246860044769665e-05,
+      "loss": 0.4268,
+      "step": 18927
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.123725485698165e-05,
+      "loss": 0.4793,
+      "step": 18928
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1227650259385416e-05,
+      "loss": 0.4601,
+      "step": 18929
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.121804625209716e-05,
+      "loss": 0.4953,
+      "step": 18930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1208442835233186e-05,
+      "loss": 0.506,
+      "step": 18931
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.11988400089097e-05,
+      "loss": 0.4741,
+      "step": 18932
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.118923777324293e-05,
+      "loss": 0.3795,
+      "step": 18933
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1179636128349085e-05,
+      "loss": 0.5322,
+      "step": 18934
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.117003507434435e-05,
+      "loss": 0.4041,
+      "step": 18935
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1160434611345e-05,
+      "loss": 0.5073,
+      "step": 18936
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.115083473946719e-05,
+      "loss": 0.4357,
+      "step": 18937
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.114123545882712e-05,
+      "loss": 0.412,
+      "step": 18938
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.113163676954097e-05,
+      "loss": 0.4692,
+      "step": 18939
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.112203867172487e-05,
+      "loss": 0.4699,
+      "step": 18940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.111244116549509e-05,
+      "loss": 0.4627,
+      "step": 18941
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.110284425096772e-05,
+      "loss": 0.4547,
+      "step": 18942
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.109324792825894e-05,
+      "loss": 0.4116,
+      "step": 18943
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.108365219748488e-05,
+      "loss": 0.386,
+      "step": 18944
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1074057058761645e-05,
+      "loss": 0.4069,
+      "step": 18945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1064462512205514e-05,
+      "loss": 0.5175,
+      "step": 18946
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.105486855793245e-05,
+      "loss": 0.4564,
+      "step": 18947
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.104527519605864e-05,
+      "loss": 0.4472,
+      "step": 18948
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1035682426700146e-05,
+      "loss": 0.4597,
+      "step": 18949
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.102609024997316e-05,
+      "loss": 0.4319,
+      "step": 18950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1016498665993725e-05,
+      "loss": 0.4917,
+      "step": 18951
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1006907674877946e-05,
+      "loss": 0.4724,
+      "step": 18952
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.099731727674191e-05,
+      "loss": 0.3855,
+      "step": 18953
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.098772747170162e-05,
+      "loss": 0.4675,
+      "step": 18954
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.097813825987326e-05,
+      "loss": 0.3871,
+      "step": 18955
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0968549641372845e-05,
+      "loss": 0.5283,
+      "step": 18956
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.095896161631642e-05,
+      "loss": 0.462,
+      "step": 18957
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.094937418482002e-05,
+      "loss": 0.4342,
+      "step": 18958
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.093978734699968e-05,
+      "loss": 0.4229,
+      "step": 18959
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.093020110297149e-05,
+      "loss": 0.4047,
+      "step": 18960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.092061545285143e-05,
+      "loss": 0.5919,
+      "step": 18961
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0911030396755535e-05,
+      "loss": 0.4259,
+      "step": 18962
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0901445934799794e-05,
+      "loss": 0.5168,
+      "step": 18963
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.089186206710024e-05,
+      "loss": 0.452,
+      "step": 18964
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.088227879377281e-05,
+      "loss": 0.5386,
+      "step": 18965
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.087269611493357e-05,
+      "loss": 0.4592,
+      "step": 18966
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0863114030698476e-05,
+      "loss": 0.486,
+      "step": 18967
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.08535325411835e-05,
+      "loss": 0.4323,
+      "step": 18968
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0843951646504615e-05,
+      "loss": 0.4042,
+      "step": 18969
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.083437134677772e-05,
+      "loss": 0.5239,
+      "step": 18970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.082479164211886e-05,
+      "loss": 0.5,
+      "step": 18971
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0815212532643964e-05,
+      "loss": 0.5172,
+      "step": 18972
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.080563401846894e-05,
+      "loss": 0.4821,
+      "step": 18973
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.079605609970972e-05,
+      "loss": 0.4171,
+      "step": 18974
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.078647877648222e-05,
+      "loss": 0.4633,
+      "step": 18975
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.077690204890241e-05,
+      "loss": 0.5115,
+      "step": 18976
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.076732591708617e-05,
+      "loss": 0.531,
+      "step": 18977
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.07577503811494e-05,
+      "loss": 0.4297,
+      "step": 18978
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.074817544120799e-05,
+      "loss": 0.5012,
+      "step": 18979
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.07386010973778e-05,
+      "loss": 0.3969,
+      "step": 18980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0729027349774784e-05,
+      "loss": 0.4372,
+      "step": 18981
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0719454198514784e-05,
+      "loss": 0.5041,
+      "step": 18982
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.070988164371366e-05,
+      "loss": 0.5282,
+      "step": 18983
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.070030968548728e-05,
+      "loss": 0.4143,
+      "step": 18984
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.069073832395145e-05,
+      "loss": 0.476,
+      "step": 18985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.06811675592221e-05,
+      "loss": 0.4361,
+      "step": 18986
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0671597391415005e-05,
+      "loss": 0.4338,
+      "step": 18987
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.066202782064603e-05,
+      "loss": 0.5311,
+      "step": 18988
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0652458847030984e-05,
+      "loss": 0.5773,
+      "step": 18989
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0642890470685647e-05,
+      "loss": 0.4938,
+      "step": 18990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0633322691725894e-05,
+      "loss": 0.5209,
+      "step": 18991
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.06237555102675e-05,
+      "loss": 0.5484,
+      "step": 18992
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.061418892642626e-05,
+      "loss": 0.3951,
+      "step": 18993
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.060462294031796e-05,
+      "loss": 0.4182,
+      "step": 18994
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.059505755205833e-05,
+      "loss": 0.4812,
+      "step": 18995
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0585492761763255e-05,
+      "loss": 0.456,
+      "step": 18996
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0575928569548425e-05,
+      "loss": 0.5377,
+      "step": 18997
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.05663649755296e-05,
+      "loss": 0.5401,
+      "step": 18998
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.055680197982255e-05,
+      "loss": 0.4542,
+      "step": 18999
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0547239582543036e-05,
+      "loss": 0.5628,
+      "step": 19000
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 0.45892348885536194,
+      "eval_runtime": 847.1291,
+      "eval_samples_per_second": 28.402,
+      "eval_steps_per_second": 14.201,
+      "step": 19000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0537677783806714e-05,
+      "loss": 0.4843,
+      "step": 19001
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.05281165837294e-05,
+      "loss": 0.4235,
+      "step": 19002
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0518555982426805e-05,
+      "loss": 0.424,
+      "step": 19003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.050899598001462e-05,
+      "loss": 0.4903,
+      "step": 19004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.049943657660855e-05,
+      "loss": 0.5635,
+      "step": 19005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.048987777232428e-05,
+      "loss": 0.3874,
+      "step": 19006
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.048031956727756e-05,
+      "loss": 0.4459,
+      "step": 19007
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.047076196158405e-05,
+      "loss": 0.5429,
+      "step": 19008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.046120495535942e-05,
+      "loss": 0.4221,
+      "step": 19009
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0451648548719335e-05,
+      "loss": 0.5052,
+      "step": 19010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.044209274177945e-05,
+      "loss": 0.4976,
+      "step": 19011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.043253753465547e-05,
+      "loss": 0.4084,
+      "step": 19012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0422982927463014e-05,
+      "loss": 0.4096,
+      "step": 19013
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.041342892031773e-05,
+      "loss": 0.4702,
+      "step": 19014
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.040387551333525e-05,
+      "loss": 0.5646,
+      "step": 19015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.039432270663116e-05,
+      "loss": 0.4886,
+      "step": 19016
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0384770500321176e-05,
+      "loss": 0.3828,
+      "step": 19017
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0375218894520834e-05,
+      "loss": 0.4792,
+      "step": 19018
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.036566788934579e-05,
+      "loss": 0.4458,
+      "step": 19019
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.03561174849116e-05,
+      "loss": 0.4355,
+      "step": 19020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0346567681333855e-05,
+      "loss": 0.5085,
+      "step": 19021
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.033701847872818e-05,
+      "loss": 0.4926,
+      "step": 19022
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.032746987721014e-05,
+      "loss": 0.4713,
+      "step": 19023
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.031792187689529e-05,
+      "loss": 0.4064,
+      "step": 19024
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.03083744778992e-05,
+      "loss": 0.4336,
+      "step": 19025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.029882768033738e-05,
+      "loss": 0.4144,
+      "step": 19026
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.028928148432548e-05,
+      "loss": 0.52,
+      "step": 19027
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.027973588997896e-05,
+      "loss": 0.5398,
+      "step": 19028
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.027019089741338e-05,
+      "loss": 0.4316,
+      "step": 19029
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.026064650674427e-05,
+      "loss": 0.3928,
+      "step": 19030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0251102718087126e-05,
+      "loss": 0.5218,
+      "step": 19031
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.024155953155745e-05,
+      "loss": 0.5461,
+      "step": 19032
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.02320169472708e-05,
+      "loss": 0.3912,
+      "step": 19033
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0222474965342645e-05,
+      "loss": 0.4682,
+      "step": 19034
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.021293358588849e-05,
+      "loss": 0.3724,
+      "step": 19035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0203392809023786e-05,
+      "loss": 0.421,
+      "step": 19036
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0193852634863983e-05,
+      "loss": 0.4612,
+      "step": 19037
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0184313063524634e-05,
+      "loss": 0.4381,
+      "step": 19038
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.017477409512116e-05,
+      "loss": 0.5201,
+      "step": 19039
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0165235729769e-05,
+      "loss": 0.4162,
+      "step": 19040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0155697967583615e-05,
+      "loss": 0.4411,
+      "step": 19041
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0146160808680395e-05,
+      "loss": 0.4005,
+      "step": 19042
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.013662425317486e-05,
+      "loss": 0.4558,
+      "step": 19043
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.012708830118238e-05,
+      "loss": 0.5118,
+      "step": 19044
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.011755295281838e-05,
+      "loss": 0.4411,
+      "step": 19045
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0108018208198284e-05,
+      "loss": 0.4653,
+      "step": 19046
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.009848406743742e-05,
+      "loss": 0.5456,
+      "step": 19047
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.00889505306513e-05,
+      "loss": 0.4063,
+      "step": 19048
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0079417597955245e-05,
+      "loss": 0.5882,
+      "step": 19049
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.006988526946466e-05,
+      "loss": 0.5474,
+      "step": 19050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.006035354529488e-05,
+      "loss": 0.4738,
+      "step": 19051
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.005082242556127e-05,
+      "loss": 0.5221,
+      "step": 19052
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.004129191037924e-05,
+      "loss": 0.4532,
+      "step": 19053
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.003176199986412e-05,
+      "loss": 0.4521,
+      "step": 19054
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0022232694131247e-05,
+      "loss": 0.4375,
+      "step": 19055
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.001270399329595e-05,
+      "loss": 0.4194,
+      "step": 19056
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.000317589747353e-05,
+      "loss": 0.4505,
+      "step": 19057
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.999364840677938e-05,
+      "loss": 0.4846,
+      "step": 19058
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.998412152132879e-05,
+      "loss": 0.4995,
+      "step": 19059
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9974595241237044e-05,
+      "loss": 0.4251,
+      "step": 19060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9965069566619447e-05,
+      "loss": 0.4089,
+      "step": 19061
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.995554449759127e-05,
+      "loss": 0.5128,
+      "step": 19062
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.994602003426786e-05,
+      "loss": 0.5186,
+      "step": 19063
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.993649617676446e-05,
+      "loss": 0.5045,
+      "step": 19064
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9926972925196336e-05,
+      "loss": 0.5068,
+      "step": 19065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.991745027967877e-05,
+      "loss": 0.4897,
+      "step": 19066
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9907928240326985e-05,
+      "loss": 0.4756,
+      "step": 19067
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9898406807256225e-05,
+      "loss": 0.3503,
+      "step": 19068
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.988888598058178e-05,
+      "loss": 0.4176,
+      "step": 19069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.987936576041886e-05,
+      "loss": 0.5249,
+      "step": 19070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.986984614688269e-05,
+      "loss": 0.4454,
+      "step": 19071
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.986032714008849e-05,
+      "loss": 0.4853,
+      "step": 19072
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.985080874015143e-05,
+      "loss": 0.4847,
+      "step": 19073
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.984129094718679e-05,
+      "loss": 0.4987,
+      "step": 19074
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.983177376130974e-05,
+      "loss": 0.4999,
+      "step": 19075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9822257182635457e-05,
+      "loss": 0.4195,
+      "step": 19076
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9812741211279125e-05,
+      "loss": 0.4543,
+      "step": 19077
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.980322584735588e-05,
+      "loss": 0.4782,
+      "step": 19078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9793711090980975e-05,
+      "loss": 0.5142,
+      "step": 19079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.978419694226956e-05,
+      "loss": 0.4284,
+      "step": 19080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9774683401336716e-05,
+      "loss": 0.5413,
+      "step": 19081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.976517046829758e-05,
+      "loss": 0.5345,
+      "step": 19082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.975565814326739e-05,
+      "loss": 0.4844,
+      "step": 19083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.974614642636122e-05,
+      "loss": 0.4425,
+      "step": 19084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.973663531769419e-05,
+      "loss": 0.3939,
+      "step": 19085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.972712481738143e-05,
+      "loss": 0.3935,
+      "step": 19086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9717614925538005e-05,
+      "loss": 0.4441,
+      "step": 19087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9708105642279076e-05,
+      "loss": 0.3682,
+      "step": 19088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9698596967719734e-05,
+      "loss": 0.4423,
+      "step": 19089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9689088901975033e-05,
+      "loss": 0.3898,
+      "step": 19090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9679581445160076e-05,
+      "loss": 0.4457,
+      "step": 19091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9670074597389874e-05,
+      "loss": 0.4449,
+      "step": 19092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.966056835877958e-05,
+      "loss": 0.4513,
+      "step": 19093
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.965106272944422e-05,
+      "loss": 0.528,
+      "step": 19094
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.964155770949883e-05,
+      "loss": 0.5271,
+      "step": 19095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.963205329905847e-05,
+      "loss": 0.6017,
+      "step": 19096
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9622549498238124e-05,
+      "loss": 0.4975,
+      "step": 19097
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.961304630715289e-05,
+      "loss": 0.4705,
+      "step": 19098
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.960354372591776e-05,
+      "loss": 0.5411,
+      "step": 19099
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.959404175464775e-05,
+      "loss": 0.4715,
+      "step": 19100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.958454039345786e-05,
+      "loss": 0.4199,
+      "step": 19101
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9575039642463085e-05,
+      "loss": 0.4848,
+      "step": 19102
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.956553950177838e-05,
+      "loss": 0.4403,
+      "step": 19103
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.955603997151881e-05,
+      "loss": 0.4403,
+      "step": 19104
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.95465410517993e-05,
+      "loss": 0.53,
+      "step": 19105
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.953704274273484e-05,
+      "loss": 0.5303,
+      "step": 19106
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.952754504444037e-05,
+      "loss": 0.4799,
+      "step": 19107
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.951804795703081e-05,
+      "loss": 0.4586,
+      "step": 19108
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.95085514806212e-05,
+      "loss": 0.4181,
+      "step": 19109
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.949905561532642e-05,
+      "loss": 0.4972,
+      "step": 19110
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9489560361261414e-05,
+      "loss": 0.6038,
+      "step": 19111
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.948006571854109e-05,
+      "loss": 0.5116,
+      "step": 19112
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.947057168728035e-05,
+      "loss": 0.5475,
+      "step": 19113
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.946107826759416e-05,
+      "loss": 0.4634,
+      "step": 19114
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9451585459597406e-05,
+      "loss": 0.4997,
+      "step": 19115
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.944209326340495e-05,
+      "loss": 0.3878,
+      "step": 19116
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.943260167913172e-05,
+      "loss": 0.4834,
+      "step": 19117
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.942311070689252e-05,
+      "loss": 0.5844,
+      "step": 19118
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.941362034680231e-05,
+      "loss": 0.4755,
+      "step": 19119
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9404130598975926e-05,
+      "loss": 0.4336,
+      "step": 19120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9394641463528225e-05,
+      "loss": 0.4677,
+      "step": 19121
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.938515294057405e-05,
+      "loss": 0.5592,
+      "step": 19122
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9375665030228214e-05,
+      "loss": 0.4422,
+      "step": 19123
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.936617773260561e-05,
+      "loss": 0.4513,
+      "step": 19124
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.935669104782106e-05,
+      "loss": 0.4092,
+      "step": 19125
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.934720497598935e-05,
+      "loss": 0.4781,
+      "step": 19126
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.933771951722531e-05,
+      "loss": 0.4285,
+      "step": 19127
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.93282346716437e-05,
+      "loss": 0.4592,
+      "step": 19128
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.931875043935942e-05,
+      "loss": 0.4802,
+      "step": 19129
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9309266820487185e-05,
+      "loss": 0.4915,
+      "step": 19130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9299783815141796e-05,
+      "loss": 0.4062,
+      "step": 19131
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9290301423438046e-05,
+      "loss": 0.4465,
+      "step": 19132
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.928081964549067e-05,
+      "loss": 0.4951,
+      "step": 19133
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.927133848141442e-05,
+      "loss": 0.4195,
+      "step": 19134
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.926185793132411e-05,
+      "loss": 0.4254,
+      "step": 19135
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9252377995334444e-05,
+      "loss": 0.4487,
+      "step": 19136
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9242898673560176e-05,
+      "loss": 0.4834,
+      "step": 19137
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9233419966116036e-05,
+      "loss": 0.4596,
+      "step": 19138
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.92239418731167e-05,
+      "loss": 0.4404,
+      "step": 19139
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.921446439467696e-05,
+      "loss": 0.4383,
+      "step": 19140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9204987530911495e-05,
+      "loss": 0.5913,
+      "step": 19141
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9195511281935e-05,
+      "loss": 0.4507,
+      "step": 19142
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9186035647862185e-05,
+      "loss": 0.4733,
+      "step": 19143
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9176560628807665e-05,
+      "loss": 0.4733,
+      "step": 19144
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.916708622488624e-05,
+      "loss": 0.4816,
+      "step": 19145
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9157612436212506e-05,
+      "loss": 0.4422,
+      "step": 19146
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.914813926290115e-05,
+      "loss": 0.4605,
+      "step": 19147
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9138666705066815e-05,
+      "loss": 0.4519,
+      "step": 19148
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.912919476282412e-05,
+      "loss": 0.4569,
+      "step": 19149
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.911972343628778e-05,
+      "loss": 0.4905,
+      "step": 19150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9110252725572405e-05,
+      "loss": 0.4842,
+      "step": 19151
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.91007826307926e-05,
+      "loss": 0.5598,
+      "step": 19152
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.909131315206299e-05,
+      "loss": 0.4274,
+      "step": 19153
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.908184428949816e-05,
+      "loss": 0.4107,
+      "step": 19154
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9072376043212784e-05,
+      "loss": 0.3659,
+      "step": 19155
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.906290841332143e-05,
+      "loss": 0.4197,
+      "step": 19156
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9053441399938674e-05,
+      "loss": 0.4122,
+      "step": 19157
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.904397500317911e-05,
+      "loss": 0.4824,
+      "step": 19158
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9034509223157265e-05,
+      "loss": 0.4548,
+      "step": 19159
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.902504405998778e-05,
+      "loss": 0.396,
+      "step": 19160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.901557951378519e-05,
+      "loss": 0.5096,
+      "step": 19161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.900611558466405e-05,
+      "loss": 0.4608,
+      "step": 19162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.89966522727389e-05,
+      "loss": 0.5631,
+      "step": 19163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.898718957812426e-05,
+      "loss": 0.4058,
+      "step": 19164
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8977727500934646e-05,
+      "loss": 0.3864,
+      "step": 19165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.896826604128464e-05,
+      "loss": 0.5176,
+      "step": 19166
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.895880519928873e-05,
+      "loss": 0.4163,
+      "step": 19167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.894934497506142e-05,
+      "loss": 0.4651,
+      "step": 19168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.893988536871721e-05,
+      "loss": 0.5596,
+      "step": 19169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8930426380370566e-05,
+      "loss": 0.524,
+      "step": 19170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8920968010136024e-05,
+      "loss": 0.4435,
+      "step": 19171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.891151025812805e-05,
+      "loss": 0.4809,
+      "step": 19172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.890205312446109e-05,
+      "loss": 0.4638,
+      "step": 19173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.889259660924963e-05,
+      "loss": 0.5534,
+      "step": 19174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.888314071260807e-05,
+      "loss": 0.4516,
+      "step": 19175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.887368543465095e-05,
+      "loss": 0.5044,
+      "step": 19176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.886423077549266e-05,
+      "loss": 0.4529,
+      "step": 19177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.885477673524763e-05,
+      "loss": 0.3968,
+      "step": 19178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.884532331403029e-05,
+      "loss": 0.4743,
+      "step": 19179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.883587051195503e-05,
+      "loss": 0.4234,
+      "step": 19180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.882641832913632e-05,
+      "loss": 0.4723,
+      "step": 19181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.881696676568854e-05,
+      "loss": 0.5186,
+      "step": 19182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.880751582172608e-05,
+      "loss": 0.4445,
+      "step": 19183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.879806549736332e-05,
+      "loss": 0.4822,
+      "step": 19184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.878861579271461e-05,
+      "loss": 0.4759,
+      "step": 19185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.87791667078944e-05,
+      "loss": 0.4948,
+      "step": 19186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8769718243017016e-05,
+      "loss": 0.5339,
+      "step": 19187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.876027039819681e-05,
+      "loss": 0.5422,
+      "step": 19188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.875082317354813e-05,
+      "loss": 0.5015,
+      "step": 19189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8741376569185305e-05,
+      "loss": 0.4309,
+      "step": 19190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8731930585222716e-05,
+      "loss": 0.6219,
+      "step": 19191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.872248522177467e-05,
+      "loss": 0.4132,
+      "step": 19192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8713040478955494e-05,
+      "loss": 0.479,
+      "step": 19193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.870359635687948e-05,
+      "loss": 0.5614,
+      "step": 19194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8694152855660914e-05,
+      "loss": 0.4843,
+      "step": 19195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8684709975414156e-05,
+      "loss": 0.4287,
+      "step": 19196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.867526771625346e-05,
+      "loss": 0.6004,
+      "step": 19197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.866582607829312e-05,
+      "loss": 0.407,
+      "step": 19198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.86563850616474e-05,
+      "loss": 0.4263,
+      "step": 19199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.864694466643057e-05,
+      "loss": 0.4372,
+      "step": 19200
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 0.45599257946014404,
+      "eval_runtime": 847.1101,
+      "eval_samples_per_second": 28.402,
+      "eval_steps_per_second": 14.201,
+      "step": 19200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.863750489275687e-05,
+      "loss": 0.4755,
+      "step": 19201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8628065740740606e-05,
+      "loss": 0.4282,
+      "step": 19202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8618627210495984e-05,
+      "loss": 0.474,
+      "step": 19203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.860918930213725e-05,
+      "loss": 0.4433,
+      "step": 19204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.859975201577864e-05,
+      "loss": 0.5168,
+      "step": 19205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8590315351534335e-05,
+      "loss": 0.52,
+      "step": 19206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.858087930951861e-05,
+      "loss": 0.4778,
+      "step": 19207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8571443889845656e-05,
+      "loss": 0.4786,
+      "step": 19208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.856200909262968e-05,
+      "loss": 0.4968,
+      "step": 19209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.855257491798484e-05,
+      "loss": 0.5584,
+      "step": 19210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.854314136602531e-05,
+      "loss": 0.4035,
+      "step": 19211
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.853370843686534e-05,
+      "loss": 0.3484,
+      "step": 19212
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.852427613061906e-05,
+      "loss": 0.4981,
+      "step": 19213
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.851484444740062e-05,
+      "loss": 0.5178,
+      "step": 19214
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.850541338732419e-05,
+      "loss": 0.4839,
+      "step": 19215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.849598295050391e-05,
+      "loss": 0.4755,
+      "step": 19216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.848655313705394e-05,
+      "loss": 0.5252,
+      "step": 19217
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8477123947088375e-05,
+      "loss": 0.4292,
+      "step": 19218
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.846769538072137e-05,
+      "loss": 0.5145,
+      "step": 19219
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.845826743806698e-05,
+      "loss": 0.4991,
+      "step": 19220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8448840119239415e-05,
+      "loss": 0.4966,
+      "step": 19221
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.843941342435272e-05,
+      "loss": 0.4945,
+      "step": 19222
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.842998735352101e-05,
+      "loss": 0.5414,
+      "step": 19223
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.842056190685835e-05,
+      "loss": 0.4566,
+      "step": 19224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8411137084478786e-05,
+      "loss": 0.4143,
+      "step": 19225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.840171288649648e-05,
+      "loss": 0.4881,
+      "step": 19226
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.839228931302544e-05,
+      "loss": 0.5115,
+      "step": 19227
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8382866364179736e-05,
+      "loss": 0.4804,
+      "step": 19228
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8373444040073414e-05,
+      "loss": 0.6531,
+      "step": 19229
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.836402234082047e-05,
+      "loss": 0.5446,
+      "step": 19230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8354601266535026e-05,
+      "loss": 0.477,
+      "step": 19231
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.834518081733106e-05,
+      "loss": 0.5167,
+      "step": 19232
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.83357609933226e-05,
+      "loss": 0.5334,
+      "step": 19233
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.832634179462364e-05,
+      "loss": 0.4792,
+      "step": 19234
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.831692322134821e-05,
+      "loss": 0.3714,
+      "step": 19235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.830750527361024e-05,
+      "loss": 0.4385,
+      "step": 19236
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.829808795152382e-05,
+      "loss": 0.5269,
+      "step": 19237
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.828867125520288e-05,
+      "loss": 0.4507,
+      "step": 19238
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8279255184761387e-05,
+      "loss": 0.5308,
+      "step": 19239
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.826983974031332e-05,
+      "loss": 0.3831,
+      "step": 19240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.826042492197259e-05,
+      "loss": 0.4171,
+      "step": 19241
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.825101072985324e-05,
+      "loss": 0.527,
+      "step": 19242
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.824159716406915e-05,
+      "loss": 0.4818,
+      "step": 19243
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.823218422473428e-05,
+      "loss": 0.5113,
+      "step": 19244
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.822277191196254e-05,
+      "loss": 0.5018,
+      "step": 19245
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.821336022586781e-05,
+      "loss": 0.5633,
+      "step": 19246
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.82039491665641e-05,
+      "loss": 0.3954,
+      "step": 19247
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.819453873416526e-05,
+      "loss": 0.4619,
+      "step": 19248
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8185128928785197e-05,
+      "loss": 0.5464,
+      "step": 19249
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.817571975053779e-05,
+      "loss": 0.5182,
+      "step": 19250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.816631119953688e-05,
+      "loss": 0.4771,
+      "step": 19251
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.815690327589645e-05,
+      "loss": 0.3956,
+      "step": 19252
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.814749597973029e-05,
+      "loss": 0.4556,
+      "step": 19253
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.813808931115228e-05,
+      "loss": 0.4847,
+      "step": 19254
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.812868327027627e-05,
+      "loss": 0.4802,
+      "step": 19255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8119277857216075e-05,
+      "loss": 0.4426,
+      "step": 19256
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.810987307208558e-05,
+      "loss": 0.5026,
+      "step": 19257
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.81004689149986e-05,
+      "loss": 0.4486,
+      "step": 19258
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.809106538606896e-05,
+      "loss": 0.3993,
+      "step": 19259
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.808166248541045e-05,
+      "loss": 0.5068,
+      "step": 19260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.807226021313687e-05,
+      "loss": 0.5271,
+      "step": 19261
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.806285856936206e-05,
+      "loss": 0.423,
+      "step": 19262
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.80534575541998e-05,
+      "loss": 0.4446,
+      "step": 19263
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8044057167763866e-05,
+      "loss": 0.5157,
+      "step": 19264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8034657410168026e-05,
+      "loss": 0.4244,
+      "step": 19265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8025258281526065e-05,
+      "loss": 0.4165,
+      "step": 19266
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.801585978195169e-05,
+      "loss": 0.4503,
+      "step": 19267
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.800646191155873e-05,
+      "loss": 0.4067,
+      "step": 19268
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.799706467046091e-05,
+      "loss": 0.5945,
+      "step": 19269
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.798766805877195e-05,
+      "loss": 0.398,
+      "step": 19270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.797827207660559e-05,
+      "loss": 0.4914,
+      "step": 19271
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.796887672407552e-05,
+      "loss": 0.4916,
+      "step": 19272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.795948200129551e-05,
+      "loss": 0.418,
+      "step": 19273
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7950087908379245e-05,
+      "loss": 0.4429,
+      "step": 19274
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.794069444544043e-05,
+      "loss": 0.5031,
+      "step": 19275
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7931301612592736e-05,
+      "loss": 0.4787,
+      "step": 19276
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.792190940994983e-05,
+      "loss": 0.4476,
+      "step": 19277
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.791251783762546e-05,
+      "loss": 0.4384,
+      "step": 19278
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.790312689573325e-05,
+      "loss": 0.3804,
+      "step": 19279
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.789373658438687e-05,
+      "loss": 0.4838,
+      "step": 19280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.788434690369997e-05,
+      "loss": 0.4619,
+      "step": 19281
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7874957853786165e-05,
+      "loss": 0.4842,
+      "step": 19282
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.786556943475915e-05,
+      "loss": 0.4815,
+      "step": 19283
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7856181646732544e-05,
+      "loss": 0.4838,
+      "step": 19284
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.784679448981996e-05,
+      "loss": 0.4977,
+      "step": 19285
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7837407964135004e-05,
+      "loss": 0.4214,
+      "step": 19286
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7828022069791256e-05,
+      "loss": 0.5338,
+      "step": 19287
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7818636806902405e-05,
+      "loss": 0.4365,
+      "step": 19288
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.780925217558197e-05,
+      "loss": 0.4412,
+      "step": 19289
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7799868175943585e-05,
+      "loss": 0.4326,
+      "step": 19290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.779048480810079e-05,
+      "loss": 0.472,
+      "step": 19291
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.778110207216713e-05,
+      "loss": 0.4703,
+      "step": 19292
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.777171996825624e-05,
+      "loss": 0.5391,
+      "step": 19293
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.776233849648163e-05,
+      "loss": 0.5161,
+      "step": 19294
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.775295765695687e-05,
+      "loss": 0.481,
+      "step": 19295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774357744979547e-05,
+      "loss": 0.4961,
+      "step": 19296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.773419787511095e-05,
+      "loss": 0.4489,
+      "step": 19297
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.77248189330169e-05,
+      "loss": 0.5246,
+      "step": 19298
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.77154406236268e-05,
+      "loss": 0.4417,
+      "step": 19299
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7706062947054145e-05,
+      "loss": 0.5099,
+      "step": 19300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7696685903412456e-05,
+      "loss": 0.4826,
+      "step": 19301
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.768730949281521e-05,
+      "loss": 0.529,
+      "step": 19302
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7677933715375875e-05,
+      "loss": 0.4015,
+      "step": 19303
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.766855857120799e-05,
+      "loss": 0.5466,
+      "step": 19304
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.765918406042499e-05,
+      "loss": 0.4808,
+      "step": 19305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7649810183140334e-05,
+      "loss": 0.4027,
+      "step": 19306
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.764043693946748e-05,
+      "loss": 0.4381,
+      "step": 19307
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.763106432951984e-05,
+      "loss": 0.4377,
+      "step": 19308
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.762169235341093e-05,
+      "loss": 0.5331,
+      "step": 19309
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.761232101125415e-05,
+      "loss": 0.4638,
+      "step": 19310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7602950303162905e-05,
+      "loss": 0.4536,
+      "step": 19311
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7593580229250636e-05,
+      "loss": 0.4531,
+      "step": 19312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.758421078963069e-05,
+      "loss": 0.3894,
+      "step": 19313
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.757484198441657e-05,
+      "loss": 0.4449,
+      "step": 19314
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.756547381372162e-05,
+      "loss": 0.4639,
+      "step": 19315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.755610627765921e-05,
+      "loss": 0.4979,
+      "step": 19316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.754673937634274e-05,
+      "loss": 0.4294,
+      "step": 19317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.753737310988554e-05,
+      "loss": 0.4624,
+      "step": 19318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.752800747840104e-05,
+      "loss": 0.4855,
+      "step": 19319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7518642482002565e-05,
+      "loss": 0.4688,
+      "step": 19320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7509278120803455e-05,
+      "loss": 0.5045,
+      "step": 19321
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.749991439491706e-05,
+      "loss": 0.5158,
+      "step": 19322
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7490551304456664e-05,
+      "loss": 0.4731,
+      "step": 19323
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.748118884953569e-05,
+      "loss": 0.4297,
+      "step": 19324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7471827030267376e-05,
+      "loss": 0.5484,
+      "step": 19325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.746246584676507e-05,
+      "loss": 0.4534,
+      "step": 19326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7453105299142054e-05,
+      "loss": 0.3928,
+      "step": 19327
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.744374538751159e-05,
+      "loss": 0.4565,
+      "step": 19328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.743438611198703e-05,
+      "loss": 0.3776,
+      "step": 19329
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.742502747268163e-05,
+      "loss": 0.3963,
+      "step": 19330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.741566946970866e-05,
+      "loss": 0.4547,
+      "step": 19331
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.740631210318136e-05,
+      "loss": 0.4633,
+      "step": 19332
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7396955373213014e-05,
+      "loss": 0.4802,
+      "step": 19333
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7387599279916814e-05,
+      "loss": 0.5149,
+      "step": 19334
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7378243823406085e-05,
+      "loss": 0.4591,
+      "step": 19335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7368889003794026e-05,
+      "loss": 0.4718,
+      "step": 19336
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.735953482119384e-05,
+      "loss": 0.4917,
+      "step": 19337
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7350181275718766e-05,
+      "loss": 0.5175,
+      "step": 19338
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.734082836748196e-05,
+      "loss": 0.5361,
+      "step": 19339
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.733147609659671e-05,
+      "loss": 0.4491,
+      "step": 19340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.732212446317617e-05,
+      "loss": 0.4495,
+      "step": 19341
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7312773467333506e-05,
+      "loss": 0.4543,
+      "step": 19342
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.730342310918192e-05,
+      "loss": 0.3939,
+      "step": 19343
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7294073388834545e-05,
+      "loss": 0.4482,
+      "step": 19344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.728472430640462e-05,
+      "loss": 0.4385,
+      "step": 19345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.727537586200525e-05,
+      "loss": 0.4766,
+      "step": 19346
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.726602805574957e-05,
+      "loss": 0.4196,
+      "step": 19347
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7256680887750746e-05,
+      "loss": 0.4358,
+      "step": 19348
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.724733435812186e-05,
+      "loss": 0.3754,
+      "step": 19349
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.723798846697617e-05,
+      "loss": 0.5923,
+      "step": 19350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.722864321442665e-05,
+      "loss": 0.527,
+      "step": 19351
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7219298600586446e-05,
+      "loss": 0.4437,
+      "step": 19352
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.720995462556865e-05,
+      "loss": 0.4946,
+      "step": 19353
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7200611289486405e-05,
+      "loss": 0.3968,
+      "step": 19354
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.719126859245277e-05,
+      "loss": 0.3774,
+      "step": 19355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7181926534580824e-05,
+      "loss": 0.4371,
+      "step": 19356
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.717258511598363e-05,
+      "loss": 0.5513,
+      "step": 19357
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.716324433677422e-05,
+      "loss": 0.4768,
+      "step": 19358
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.71539041970657e-05,
+      "loss": 0.5349,
+      "step": 19359
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.714456469697112e-05,
+      "loss": 0.4766,
+      "step": 19360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.713522583660349e-05,
+      "loss": 0.4649,
+      "step": 19361
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.712588761607586e-05,
+      "loss": 0.5229,
+      "step": 19362
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7116550035501196e-05,
+      "loss": 0.4779,
+      "step": 19363
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7107213094992594e-05,
+      "loss": 0.4579,
+      "step": 19364
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.709787679466303e-05,
+      "loss": 0.4442,
+      "step": 19365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7088541134625506e-05,
+      "loss": 0.4472,
+      "step": 19366
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.707920611499302e-05,
+      "loss": 0.5585,
+      "step": 19367
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7069871735878544e-05,
+      "loss": 0.4971,
+      "step": 19368
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.706053799739501e-05,
+      "loss": 0.5042,
+      "step": 19369
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.70512048996555e-05,
+      "loss": 0.4602,
+      "step": 19370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.704187244277289e-05,
+      "loss": 0.47,
+      "step": 19371
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.703254062686017e-05,
+      "loss": 0.4074,
+      "step": 19372
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.702320945203027e-05,
+      "loss": 0.4807,
+      "step": 19373
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.701387891839609e-05,
+      "loss": 0.4047,
+      "step": 19374
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7004549026070645e-05,
+      "loss": 0.4008,
+      "step": 19375
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.699521977516681e-05,
+      "loss": 0.4022,
+      "step": 19376
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.69858911657975e-05,
+      "loss": 0.4487,
+      "step": 19377
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.697656319807564e-05,
+      "loss": 0.4454,
+      "step": 19378
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.696723587211407e-05,
+      "loss": 0.4898,
+      "step": 19379
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.695790918802576e-05,
+      "loss": 0.4054,
+      "step": 19380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.694858314592358e-05,
+      "loss": 0.4487,
+      "step": 19381
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.693925774592036e-05,
+      "loss": 0.5623,
+      "step": 19382
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.692993298812902e-05,
+      "loss": 0.4473,
+      "step": 19383
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.692060887266235e-05,
+      "loss": 0.4728,
+      "step": 19384
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.691128539963329e-05,
+      "loss": 0.4588,
+      "step": 19385
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.690196256915464e-05,
+      "loss": 0.496,
+      "step": 19386
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6892640381339256e-05,
+      "loss": 0.4469,
+      "step": 19387
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.688331883629995e-05,
+      "loss": 0.4577,
+      "step": 19388
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6873997934149504e-05,
+      "loss": 0.4628,
+      "step": 19389
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.686467767500081e-05,
+      "loss": 0.5126,
+      "step": 19390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.685535805896665e-05,
+      "loss": 0.4367,
+      "step": 19391
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.684603908615981e-05,
+      "loss": 0.3913,
+      "step": 19392
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6836720756693075e-05,
+      "loss": 0.3983,
+      "step": 19393
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.682740307067921e-05,
+      "loss": 0.4806,
+      "step": 19394
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6818086028231046e-05,
+      "loss": 0.4311,
+      "step": 19395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.680876962946133e-05,
+      "loss": 0.5035,
+      "step": 19396
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.67994538744828e-05,
+      "loss": 0.5168,
+      "step": 19397
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.679013876340822e-05,
+      "loss": 0.5035,
+      "step": 19398
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.67808242963503e-05,
+      "loss": 0.5578,
+      "step": 19399
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.677151047342185e-05,
+      "loss": 0.4748,
+      "step": 19400
+    },
+    {
+      "epoch": 0.68,
+      "eval_loss": 0.4529370069503784,
+      "eval_runtime": 850.9234,
+      "eval_samples_per_second": 28.275,
+      "eval_steps_per_second": 14.138,
+      "step": 19400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.676219729473555e-05,
+      "loss": 0.4328,
+      "step": 19401
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.675288476040414e-05,
+      "loss": 0.4861,
+      "step": 19402
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.674357287054031e-05,
+      "loss": 0.5461,
+      "step": 19403
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.673426162525678e-05,
+      "loss": 0.4516,
+      "step": 19404
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6724951024666206e-05,
+      "loss": 0.512,
+      "step": 19405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.671564106888135e-05,
+      "loss": 0.4949,
+      "step": 19406
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.670633175801486e-05,
+      "loss": 0.4816,
+      "step": 19407
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.669702309217942e-05,
+      "loss": 0.5345,
+      "step": 19408
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.668771507148766e-05,
+      "loss": 0.4601,
+      "step": 19409
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.667840769605224e-05,
+      "loss": 0.4671,
+      "step": 19410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.666910096598586e-05,
+      "loss": 0.5308,
+      "step": 19411
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.665979488140113e-05,
+      "loss": 0.5397,
+      "step": 19412
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.66504894424107e-05,
+      "loss": 0.5107,
+      "step": 19413
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.664118464912718e-05,
+      "loss": 0.5043,
+      "step": 19414
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.663188050166316e-05,
+      "loss": 0.4332,
+      "step": 19415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.662257700013133e-05,
+      "loss": 0.4043,
+      "step": 19416
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.661327414464424e-05,
+      "loss": 0.4646,
+      "step": 19417
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.660397193531449e-05,
+      "loss": 0.4741,
+      "step": 19418
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.659467037225468e-05,
+      "loss": 0.4439,
+      "step": 19419
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6585369455577346e-05,
+      "loss": 0.5545,
+      "step": 19420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6576069185395134e-05,
+      "loss": 0.5009,
+      "step": 19421
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.656676956182057e-05,
+      "loss": 0.4461,
+      "step": 19422
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.655747058496621e-05,
+      "loss": 0.3714,
+      "step": 19423
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.65481722549446e-05,
+      "loss": 0.4115,
+      "step": 19424
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6538874571868255e-05,
+      "loss": 0.5331,
+      "step": 19425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6529577535849766e-05,
+      "loss": 0.4527,
+      "step": 19426
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6520281147001644e-05,
+      "loss": 0.4635,
+      "step": 19427
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.651098540543638e-05,
+      "loss": 0.4278,
+      "step": 19428
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.65016903112665e-05,
+      "loss": 0.4805,
+      "step": 19429
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6492395864604465e-05,
+      "loss": 0.4319,
+      "step": 19430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.648310206556283e-05,
+      "loss": 0.469,
+      "step": 19431
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.647380891425407e-05,
+      "loss": 0.4562,
+      "step": 19432
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.646451641079065e-05,
+      "loss": 0.4271,
+      "step": 19433
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.645522455528504e-05,
+      "loss": 0.4236,
+      "step": 19434
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.644593334784969e-05,
+      "loss": 0.4999,
+      "step": 19435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.643664278859703e-05,
+      "loss": 0.4241,
+      "step": 19436
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6427352877639586e-05,
+      "loss": 0.4667,
+      "step": 19437
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.641806361508976e-05,
+      "loss": 0.4721,
+      "step": 19438
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.640877500105997e-05,
+      "loss": 0.4205,
+      "step": 19439
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.639948703566265e-05,
+      "loss": 0.4307,
+      "step": 19440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6390199719010164e-05,
+      "loss": 0.5149,
+      "step": 19441
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.638091305121501e-05,
+      "loss": 0.4601,
+      "step": 19442
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.637162703238954e-05,
+      "loss": 0.4896,
+      "step": 19443
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.636234166264616e-05,
+      "loss": 0.4062,
+      "step": 19444
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.635305694209724e-05,
+      "loss": 0.5025,
+      "step": 19445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.634377287085513e-05,
+      "loss": 0.5099,
+      "step": 19446
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.633448944903226e-05,
+      "loss": 0.4709,
+      "step": 19447
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.632520667674095e-05,
+      "loss": 0.4654,
+      "step": 19448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.631592455409357e-05,
+      "loss": 0.4913,
+      "step": 19449
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.630664308120246e-05,
+      "loss": 0.5281,
+      "step": 19450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.629736225817991e-05,
+      "loss": 0.4628,
+      "step": 19451
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.628808208513833e-05,
+      "loss": 0.387,
+      "step": 19452
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6278802562190004e-05,
+      "loss": 0.398,
+      "step": 19453
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.626952368944725e-05,
+      "loss": 0.4543,
+      "step": 19454
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6260245467022354e-05,
+      "loss": 0.4361,
+      "step": 19455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6250967895027594e-05,
+      "loss": 0.4315,
+      "step": 19456
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6241690973575334e-05,
+      "loss": 0.522,
+      "step": 19457
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.62324147027778e-05,
+      "loss": 0.4693,
+      "step": 19458
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.62231390827473e-05,
+      "loss": 0.3884,
+      "step": 19459
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.621386411359607e-05,
+      "loss": 0.5284,
+      "step": 19460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.620458979543634e-05,
+      "loss": 0.5198,
+      "step": 19461
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.619531612838044e-05,
+      "loss": 0.4413,
+      "step": 19462
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.618604311254056e-05,
+      "loss": 0.5816,
+      "step": 19463
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.617677074802895e-05,
+      "loss": 0.5006,
+      "step": 19464
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.616749903495783e-05,
+      "loss": 0.4302,
+      "step": 19465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.615822797343938e-05,
+      "loss": 0.5533,
+      "step": 19466
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.614895756358589e-05,
+      "loss": 0.5193,
+      "step": 19467
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6139687805509535e-05,
+      "loss": 0.4648,
+      "step": 19468
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.613041869932249e-05,
+      "loss": 0.4551,
+      "step": 19469
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.612115024513694e-05,
+      "loss": 0.5479,
+      "step": 19470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.611188244306508e-05,
+      "loss": 0.4614,
+      "step": 19471
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6102615293219044e-05,
+      "loss": 0.5412,
+      "step": 19472
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.609334879571106e-05,
+      "loss": 0.5416,
+      "step": 19473
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.608408295065325e-05,
+      "loss": 0.5115,
+      "step": 19474
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.607481775815776e-05,
+      "loss": 0.5663,
+      "step": 19475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.606555321833674e-05,
+      "loss": 0.605,
+      "step": 19476
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.605628933130227e-05,
+      "loss": 0.4519,
+      "step": 19477
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.604702609716657e-05,
+      "loss": 0.4027,
+      "step": 19478
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6037763516041684e-05,
+      "loss": 0.4054,
+      "step": 19479
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.602850158803975e-05,
+      "loss": 0.6487,
+      "step": 19480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.601924031327285e-05,
+      "loss": 0.5089,
+      "step": 19481
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.600997969185305e-05,
+      "loss": 0.5085,
+      "step": 19482
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.600071972389252e-05,
+      "loss": 0.4834,
+      "step": 19483
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.599146040950331e-05,
+      "loss": 0.4986,
+      "step": 19484
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5982201748797425e-05,
+      "loss": 0.4183,
+      "step": 19485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.597294374188693e-05,
+      "loss": 0.4315,
+      "step": 19486
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5963686388883956e-05,
+      "loss": 0.4061,
+      "step": 19487
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.595442968990049e-05,
+      "loss": 0.4416,
+      "step": 19488
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.594517364504859e-05,
+      "loss": 0.4902,
+      "step": 19489
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.593591825444028e-05,
+      "loss": 0.4088,
+      "step": 19490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.592666351818754e-05,
+      "loss": 0.441,
+      "step": 19491
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.591740943640246e-05,
+      "loss": 0.4451,
+      "step": 19492
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5908156009197004e-05,
+      "loss": 0.3525,
+      "step": 19493
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5898903236683174e-05,
+      "loss": 0.4316,
+      "step": 19494
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.588965111897296e-05,
+      "loss": 0.6206,
+      "step": 19495
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.588039965617831e-05,
+      "loss": 0.4195,
+      "step": 19496
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.587114884841126e-05,
+      "loss": 0.5163,
+      "step": 19497
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5861898695783746e-05,
+      "loss": 0.4902,
+      "step": 19498
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.585264919840771e-05,
+      "loss": 0.5094,
+      "step": 19499
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.584340035639514e-05,
+      "loss": 0.4114,
+      "step": 19500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.58341521698579e-05,
+      "loss": 0.4004,
+      "step": 19501
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.582490463890803e-05,
+      "loss": 0.4177,
+      "step": 19502
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.581565776365739e-05,
+      "loss": 0.4685,
+      "step": 19503
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.580641154421792e-05,
+      "loss": 0.44,
+      "step": 19504
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5797165980701516e-05,
+      "loss": 0.4701,
+      "step": 19505
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5787921073220096e-05,
+      "loss": 0.4708,
+      "step": 19506
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5778676821885505e-05,
+      "loss": 0.4386,
+      "step": 19507
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.57694332268097e-05,
+      "loss": 0.4576,
+      "step": 19508
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5760190288104524e-05,
+      "loss": 0.5574,
+      "step": 19509
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.575094800588186e-05,
+      "loss": 0.4557,
+      "step": 19510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.574170638025356e-05,
+      "loss": 0.3682,
+      "step": 19511
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.573246541133144e-05,
+      "loss": 0.4548,
+      "step": 19512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.572322509922742e-05,
+      "loss": 0.4462,
+      "step": 19513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.571398544405331e-05,
+      "loss": 0.4964,
+      "step": 19514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.570474644592093e-05,
+      "loss": 0.4345,
+      "step": 19515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.569550810494212e-05,
+      "loss": 0.3687,
+      "step": 19516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.568627042122864e-05,
+      "loss": 0.3901,
+      "step": 19517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.567703339489238e-05,
+      "loss": 0.5257,
+      "step": 19518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.566779702604509e-05,
+      "loss": 0.4212,
+      "step": 19519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.565856131479858e-05,
+      "loss": 0.5673,
+      "step": 19520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.564932626126461e-05,
+      "loss": 0.515,
+      "step": 19521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.564009186555495e-05,
+      "loss": 0.4669,
+      "step": 19522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.563085812778141e-05,
+      "loss": 0.4463,
+      "step": 19523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.562162504805573e-05,
+      "loss": 0.5254,
+      "step": 19524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.561239262648964e-05,
+      "loss": 0.499,
+      "step": 19525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5603160863194906e-05,
+      "loss": 0.5499,
+      "step": 19526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.559392975828322e-05,
+      "loss": 0.4569,
+      "step": 19527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.558469931186637e-05,
+      "loss": 0.4426,
+      "step": 19528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.557546952405607e-05,
+      "loss": 0.4506,
+      "step": 19529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.556624039496399e-05,
+      "loss": 0.4355,
+      "step": 19530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5557011924701865e-05,
+      "loss": 0.4281,
+      "step": 19531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.554778411338133e-05,
+      "loss": 0.4896,
+      "step": 19532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5538556961114175e-05,
+      "loss": 0.3894,
+      "step": 19533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.552933046801201e-05,
+      "loss": 0.4763,
+      "step": 19534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.552010463418653e-05,
+      "loss": 0.5552,
+      "step": 19535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5510879459749386e-05,
+      "loss": 0.4004,
+      "step": 19536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.550165494481224e-05,
+      "loss": 0.4692,
+      "step": 19537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5492431089486696e-05,
+      "loss": 0.4321,
+      "step": 19538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.548320789388447e-05,
+      "loss": 0.5508,
+      "step": 19539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5473985358117166e-05,
+      "loss": 0.4643,
+      "step": 19540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.546476348229639e-05,
+      "loss": 0.4939,
+      "step": 19541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.545554226653378e-05,
+      "loss": 0.5105,
+      "step": 19542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.544632171094088e-05,
+      "loss": 0.4195,
+      "step": 19543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.543710181562938e-05,
+      "loss": 0.4926,
+      "step": 19544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.542788258071084e-05,
+      "loss": 0.3664,
+      "step": 19545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.541866400629682e-05,
+      "loss": 0.501,
+      "step": 19546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.540944609249891e-05,
+      "loss": 0.474,
+      "step": 19547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.540022883942864e-05,
+      "loss": 0.4715,
+      "step": 19548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5391012247197654e-05,
+      "loss": 0.4868,
+      "step": 19549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.538179631591745e-05,
+      "loss": 0.5661,
+      "step": 19550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.537258104569957e-05,
+      "loss": 0.5543,
+      "step": 19551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.536336643665556e-05,
+      "loss": 0.5099,
+      "step": 19552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5354152488896916e-05,
+      "loss": 0.4785,
+      "step": 19553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.534493920253522e-05,
+      "loss": 0.4941,
+      "step": 19554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.533572657768195e-05,
+      "loss": 0.425,
+      "step": 19555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5326514614448615e-05,
+      "loss": 0.4566,
+      "step": 19556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.531730331294669e-05,
+      "loss": 0.4671,
+      "step": 19557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.530809267328765e-05,
+      "loss": 0.4487,
+      "step": 19558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5298882695583034e-05,
+      "loss": 0.4545,
+      "step": 19559
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5289673379944286e-05,
+      "loss": 0.4449,
+      "step": 19560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.528046472648286e-05,
+      "loss": 0.454,
+      "step": 19561
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.527125673531022e-05,
+      "loss": 0.4521,
+      "step": 19562
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.526204940653777e-05,
+      "loss": 0.4441,
+      "step": 19563
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.525284274027701e-05,
+      "loss": 0.4316,
+      "step": 19564
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5243636736639374e-05,
+      "loss": 0.4423,
+      "step": 19565
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5234431395736246e-05,
+      "loss": 0.5114,
+      "step": 19566
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5225226717679056e-05,
+      "loss": 0.3996,
+      "step": 19567
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.521602270257922e-05,
+      "loss": 0.3973,
+      "step": 19568
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.520681935054809e-05,
+      "loss": 0.4803,
+      "step": 19569
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.519761666169714e-05,
+      "loss": 0.4817,
+      "step": 19570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.51884146361377e-05,
+      "loss": 0.4226,
+      "step": 19571
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5179213273981147e-05,
+      "loss": 0.4553,
+      "step": 19572
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.517001257533887e-05,
+      "loss": 0.3936,
+      "step": 19573
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.516081254032217e-05,
+      "loss": 0.5115,
+      "step": 19574
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5151613169042486e-05,
+      "loss": 0.4378,
+      "step": 19575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5142414461611104e-05,
+      "loss": 0.5235,
+      "step": 19576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.513321641813938e-05,
+      "loss": 0.4915,
+      "step": 19577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.512401903873863e-05,
+      "loss": 0.4233,
+      "step": 19578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5114822323520145e-05,
+      "loss": 0.5291,
+      "step": 19579
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.51056262725953e-05,
+      "loss": 0.4129,
+      "step": 19580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.509643088607537e-05,
+      "loss": 0.4851,
+      "step": 19581
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.508723616407163e-05,
+      "loss": 0.4482,
+      "step": 19582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507804210669539e-05,
+      "loss": 0.5133,
+      "step": 19583
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5068848714057884e-05,
+      "loss": 0.4734,
+      "step": 19584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.505965598627047e-05,
+      "loss": 0.4733,
+      "step": 19585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.505046392344434e-05,
+      "loss": 0.4793,
+      "step": 19586
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.504127252569077e-05,
+      "loss": 0.4383,
+      "step": 19587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.503208179312101e-05,
+      "loss": 0.49,
+      "step": 19588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.502289172584625e-05,
+      "loss": 0.4487,
+      "step": 19589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5013702323977815e-05,
+      "loss": 0.5199,
+      "step": 19590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.500451358762686e-05,
+      "loss": 0.4702,
+      "step": 19591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4995325516904615e-05,
+      "loss": 0.4222,
+      "step": 19592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4986138111922295e-05,
+      "loss": 0.4338,
+      "step": 19593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497695137279105e-05,
+      "loss": 0.4977,
+      "step": 19594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4967765299622155e-05,
+      "loss": 0.3381,
+      "step": 19595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.495857989252673e-05,
+      "loss": 0.4551,
+      "step": 19596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4949395151615984e-05,
+      "loss": 0.4499,
+      "step": 19597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.494021107700105e-05,
+      "loss": 0.4797,
+      "step": 19598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4931027668793066e-05,
+      "loss": 0.4693,
+      "step": 19599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4921844927103264e-05,
+      "loss": 0.5461,
+      "step": 19600
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 0.4498661160469055,
+      "eval_runtime": 854.1745,
+      "eval_samples_per_second": 28.168,
+      "eval_steps_per_second": 14.084,
+      "step": 19600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.491266285204273e-05,
+      "loss": 0.4602,
+      "step": 19601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.490348144372262e-05,
+      "loss": 0.4933,
+      "step": 19602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.489430070225403e-05,
+      "loss": 0.4046,
+      "step": 19603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.48851206277481e-05,
+      "loss": 0.4016,
+      "step": 19604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.48759412203159e-05,
+      "loss": 0.4212,
+      "step": 19605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.486676248006859e-05,
+      "loss": 0.4806,
+      "step": 19606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.485758440711725e-05,
+      "loss": 0.5048,
+      "step": 19607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.484840700157295e-05,
+      "loss": 0.4677,
+      "step": 19608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4839230263546764e-05,
+      "loss": 0.4364,
+      "step": 19609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4830054193149726e-05,
+      "loss": 0.4372,
+      "step": 19610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.482087879049299e-05,
+      "loss": 0.5561,
+      "step": 19611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.481170405568755e-05,
+      "loss": 0.4787,
+      "step": 19612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4802529988844466e-05,
+      "loss": 0.5174,
+      "step": 19613
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.479335659007475e-05,
+      "loss": 0.4463,
+      "step": 19614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4784183859489425e-05,
+      "loss": 0.5034,
+      "step": 19615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4775011797199574e-05,
+      "loss": 0.4975,
+      "step": 19616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.476584040331616e-05,
+      "loss": 0.504,
+      "step": 19617
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.47566696779502e-05,
+      "loss": 0.3811,
+      "step": 19618
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.47474996212127e-05,
+      "loss": 0.467,
+      "step": 19619
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.473833023321462e-05,
+      "loss": 0.3753,
+      "step": 19620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4729161514066964e-05,
+      "loss": 0.5139,
+      "step": 19621
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.47199934638807e-05,
+      "loss": 0.4379,
+      "step": 19622
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.471082608276678e-05,
+      "loss": 0.4551,
+      "step": 19623
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.470165937083613e-05,
+      "loss": 0.4525,
+      "step": 19624
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.469249332819977e-05,
+      "loss": 0.4067,
+      "step": 19625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4683327954968614e-05,
+      "loss": 0.424,
+      "step": 19626
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4674163251253586e-05,
+      "loss": 0.564,
+      "step": 19627
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4664999217165604e-05,
+      "loss": 0.4201,
+      "step": 19628
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4655835852815556e-05,
+      "loss": 0.3905,
+      "step": 19629
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.464667315831441e-05,
+      "loss": 0.499,
+      "step": 19630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4637511133773036e-05,
+      "loss": 0.4105,
+      "step": 19631
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.462834977930234e-05,
+      "loss": 0.4007,
+      "step": 19632
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.461918909501318e-05,
+      "loss": 0.5115,
+      "step": 19633
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4610029081016405e-05,
+      "loss": 0.4494,
+      "step": 19634
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4600869737422944e-05,
+      "loss": 0.4625,
+      "step": 19635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4591711064343644e-05,
+      "loss": 0.375,
+      "step": 19636
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4582553061889344e-05,
+      "loss": 0.4354,
+      "step": 19637
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4573395730170877e-05,
+      "loss": 0.4601,
+      "step": 19638
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4564239069299075e-05,
+      "loss": 0.4562,
+      "step": 19639
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.455508307938474e-05,
+      "loss": 0.426,
+      "step": 19640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4545927760538764e-05,
+      "loss": 0.3765,
+      "step": 19641
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.453677311287192e-05,
+      "loss": 0.4446,
+      "step": 19642
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.452761913649499e-05,
+      "loss": 0.4249,
+      "step": 19643
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4518465831518795e-05,
+      "loss": 0.4358,
+      "step": 19644
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.450931319805407e-05,
+      "loss": 0.4851,
+      "step": 19645
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.450016123621167e-05,
+      "loss": 0.4709,
+      "step": 19646
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4491009946102317e-05,
+      "loss": 0.4032,
+      "step": 19647
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4481859327836783e-05,
+      "loss": 0.3752,
+      "step": 19648
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.447270938152582e-05,
+      "loss": 0.3667,
+      "step": 19649
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4463560107280134e-05,
+      "loss": 0.513,
+      "step": 19650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4454411505210534e-05,
+      "loss": 0.5272,
+      "step": 19651
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.444526357542772e-05,
+      "loss": 0.4332,
+      "step": 19652
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4436116318042396e-05,
+      "loss": 0.4593,
+      "step": 19653
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4426969733165294e-05,
+      "loss": 0.4581,
+      "step": 19654
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.441782382090706e-05,
+      "loss": 0.406,
+      "step": 19655
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4408678581378485e-05,
+      "loss": 0.3983,
+      "step": 19656
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.43995340146902e-05,
+      "loss": 0.4194,
+      "step": 19657
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.43903901209529e-05,
+      "loss": 0.4671,
+      "step": 19658
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.438124690027725e-05,
+      "loss": 0.4126,
+      "step": 19659
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.437210435277388e-05,
+      "loss": 0.4633,
+      "step": 19660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4362962478553515e-05,
+      "loss": 0.4392,
+      "step": 19661
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.435382127772677e-05,
+      "loss": 0.6059,
+      "step": 19662
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.434468075040428e-05,
+      "loss": 0.5245,
+      "step": 19663
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.433554089669668e-05,
+      "loss": 0.4521,
+      "step": 19664
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.432640171671455e-05,
+      "loss": 0.4149,
+      "step": 19665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4317263210568585e-05,
+      "loss": 0.4339,
+      "step": 19666
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.430812537836934e-05,
+      "loss": 0.4481,
+      "step": 19667
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.429898822022743e-05,
+      "loss": 0.5585,
+      "step": 19668
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.428985173625344e-05,
+      "loss": 0.5391,
+      "step": 19669
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.428071592655795e-05,
+      "loss": 0.5151,
+      "step": 19670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.42715807912515e-05,
+      "loss": 0.5103,
+      "step": 19671
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.426244633044473e-05,
+      "loss": 0.4126,
+      "step": 19672
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.425331254424815e-05,
+      "loss": 0.494,
+      "step": 19673
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.424417943277232e-05,
+      "loss": 0.4603,
+      "step": 19674
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.423504699612778e-05,
+      "loss": 0.3622,
+      "step": 19675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.422591523442502e-05,
+      "loss": 0.4225,
+      "step": 19676
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4216784147774645e-05,
+      "loss": 0.5121,
+      "step": 19677
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4207653736287135e-05,
+      "loss": 0.461,
+      "step": 19678
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.419852400007299e-05,
+      "loss": 0.4784,
+      "step": 19679
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.418939493924272e-05,
+      "loss": 0.5112,
+      "step": 19680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.418026655390678e-05,
+      "loss": 0.4982,
+      "step": 19681
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.417113884417572e-05,
+      "loss": 0.3966,
+      "step": 19682
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.416201181015999e-05,
+      "loss": 0.3477,
+      "step": 19683
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.415288545197004e-05,
+      "loss": 0.4576,
+      "step": 19684
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.414375976971634e-05,
+      "loss": 0.4905,
+      "step": 19685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.413463476350931e-05,
+      "loss": 0.4928,
+      "step": 19686
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.412551043345945e-05,
+      "loss": 0.4971,
+      "step": 19687
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.411638677967718e-05,
+      "loss": 0.4797,
+      "step": 19688
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.410726380227292e-05,
+      "loss": 0.3864,
+      "step": 19689
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.409814150135706e-05,
+      "loss": 0.4105,
+      "step": 19690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4089019877040014e-05,
+      "loss": 0.4723,
+      "step": 19691
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.407989892943224e-05,
+      "loss": 0.4348,
+      "step": 19692
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.40707786586441e-05,
+      "loss": 0.4608,
+      "step": 19693
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4061659064785965e-05,
+      "loss": 0.4728,
+      "step": 19694
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.405254014796822e-05,
+      "loss": 0.4766,
+      "step": 19695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.404342190830122e-05,
+      "loss": 0.4797,
+      "step": 19696
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.403430434589536e-05,
+      "loss": 0.4602,
+      "step": 19697
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.402518746086098e-05,
+      "loss": 0.4607,
+      "step": 19698
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4016071253308436e-05,
+      "loss": 0.5205,
+      "step": 19699
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.400695572334803e-05,
+      "loss": 0.4735,
+      "step": 19700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.399784087109009e-05,
+      "loss": 0.4772,
+      "step": 19701
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.398872669664498e-05,
+      "loss": 0.4653,
+      "step": 19702
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3979613200122996e-05,
+      "loss": 0.4306,
+      "step": 19703
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.397050038163443e-05,
+      "loss": 0.4626,
+      "step": 19704
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.396138824128958e-05,
+      "loss": 0.4318,
+      "step": 19705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.395227677919873e-05,
+      "loss": 0.3652,
+      "step": 19706
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.394316599547214e-05,
+      "loss": 0.3999,
+      "step": 19707
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.393405589022013e-05,
+      "loss": 0.4,
+      "step": 19708
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.392494646355294e-05,
+      "loss": 0.4157,
+      "step": 19709
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.391583771558082e-05,
+      "loss": 0.4573,
+      "step": 19710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.390672964641401e-05,
+      "loss": 0.5322,
+      "step": 19711
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.389762225616273e-05,
+      "loss": 0.4923,
+      "step": 19712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.388851554493727e-05,
+      "loss": 0.425,
+      "step": 19713
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.387940951284781e-05,
+      "loss": 0.4972,
+      "step": 19714
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3870304160004575e-05,
+      "loss": 0.4281,
+      "step": 19715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3861199486517764e-05,
+      "loss": 0.5189,
+      "step": 19716
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.385209549249755e-05,
+      "loss": 0.6074,
+      "step": 19717
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3842992178054164e-05,
+      "loss": 0.4162,
+      "step": 19718
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.383388954329778e-05,
+      "loss": 0.4356,
+      "step": 19719
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3824787588338554e-05,
+      "loss": 0.5383,
+      "step": 19720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3815686313286665e-05,
+      "loss": 0.4349,
+      "step": 19721
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3806585718252213e-05,
+      "loss": 0.4471,
+      "step": 19722
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.379748580334543e-05,
+      "loss": 0.4566,
+      "step": 19723
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3788386568676424e-05,
+      "loss": 0.4211,
+      "step": 19724
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3779288014355314e-05,
+      "loss": 0.4593,
+      "step": 19725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.377019014049223e-05,
+      "loss": 0.4307,
+      "step": 19726
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.376109294719724e-05,
+      "loss": 0.43,
+      "step": 19727
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.375199643458053e-05,
+      "loss": 0.5153,
+      "step": 19728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.374290060275218e-05,
+      "loss": 0.4632,
+      "step": 19729
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.373380545182225e-05,
+      "loss": 0.5155,
+      "step": 19730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.372471098190084e-05,
+      "loss": 0.4941,
+      "step": 19731
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.371561719309798e-05,
+      "loss": 0.4449,
+      "step": 19732
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.37065240855238e-05,
+      "loss": 0.5043,
+      "step": 19733
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.369743165928834e-05,
+      "loss": 0.3933,
+      "step": 19734
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.368833991450164e-05,
+      "loss": 0.4302,
+      "step": 19735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.367924885127375e-05,
+      "loss": 0.424,
+      "step": 19736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.367015846971467e-05,
+      "loss": 0.4061,
+      "step": 19737
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.366106876993442e-05,
+      "loss": 0.5618,
+      "step": 19738
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.365197975204307e-05,
+      "loss": 0.5007,
+      "step": 19739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.36428914161506e-05,
+      "loss": 0.43,
+      "step": 19740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.363380376236701e-05,
+      "loss": 0.5277,
+      "step": 19741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.362471679080229e-05,
+      "loss": 0.473,
+      "step": 19742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3615630501566384e-05,
+      "loss": 0.4596,
+      "step": 19743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.360654489476934e-05,
+      "loss": 0.4197,
+      "step": 19744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.359745997052109e-05,
+      "loss": 0.4821,
+      "step": 19745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3588375728931586e-05,
+      "loss": 0.4677,
+      "step": 19746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3579292170110775e-05,
+      "loss": 0.4093,
+      "step": 19747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3570209294168576e-05,
+      "loss": 0.4712,
+      "step": 19748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.356112710121498e-05,
+      "loss": 0.5381,
+      "step": 19749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.355204559135989e-05,
+      "loss": 0.4077,
+      "step": 19750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.354296476471321e-05,
+      "loss": 0.42,
+      "step": 19751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.353388462138486e-05,
+      "loss": 0.515,
+      "step": 19752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.352480516148469e-05,
+      "loss": 0.427,
+      "step": 19753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.351572638512271e-05,
+      "loss": 0.4368,
+      "step": 19754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3506648292408695e-05,
+      "loss": 0.4457,
+      "step": 19755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3497570883452545e-05,
+      "loss": 0.4292,
+      "step": 19756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3488494158364105e-05,
+      "loss": 0.473,
+      "step": 19757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.34794181172533e-05,
+      "loss": 0.4075,
+      "step": 19758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3470342760229944e-05,
+      "loss": 0.4537,
+      "step": 19759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3461268087403876e-05,
+      "loss": 0.4723,
+      "step": 19760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.345219409888494e-05,
+      "loss": 0.4075,
+      "step": 19761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.344312079478291e-05,
+      "loss": 0.4086,
+      "step": 19762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3434048175207683e-05,
+      "loss": 0.4598,
+      "step": 19763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.342497624026903e-05,
+      "loss": 0.3866,
+      "step": 19764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.341590499007677e-05,
+      "loss": 0.3727,
+      "step": 19765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.340683442474067e-05,
+      "loss": 0.5046,
+      "step": 19766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.339776454437049e-05,
+      "loss": 0.4151,
+      "step": 19767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.338869534907608e-05,
+      "loss": 0.4655,
+      "step": 19768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.337962683896716e-05,
+      "loss": 0.5576,
+      "step": 19769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.33705590141535e-05,
+      "loss": 0.4376,
+      "step": 19770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3361491874744853e-05,
+      "loss": 0.5137,
+      "step": 19771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.335242542085094e-05,
+      "loss": 0.5019,
+      "step": 19772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3343359652581495e-05,
+      "loss": 0.4236,
+      "step": 19773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.333429457004629e-05,
+      "loss": 0.4885,
+      "step": 19774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3325230173355e-05,
+      "loss": 0.4132,
+      "step": 19775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3316166462617355e-05,
+      "loss": 0.4228,
+      "step": 19776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.330710343794305e-05,
+      "loss": 0.4097,
+      "step": 19777
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3298041099441746e-05,
+      "loss": 0.3497,
+      "step": 19778
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3288979447223185e-05,
+      "loss": 0.4583,
+      "step": 19779
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.327991848139702e-05,
+      "loss": 0.4424,
+      "step": 19780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3270858202072914e-05,
+      "loss": 0.4745,
+      "step": 19781
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3261798609360524e-05,
+      "loss": 0.3684,
+      "step": 19782
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.325273970336947e-05,
+      "loss": 0.4683,
+      "step": 19783
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3243681484209476e-05,
+      "loss": 0.4553,
+      "step": 19784
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.323462395199012e-05,
+      "loss": 0.5037,
+      "step": 19785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3225567106821044e-05,
+      "loss": 0.4468,
+      "step": 19786
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.321651094881186e-05,
+      "loss": 0.5117,
+      "step": 19787
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.320745547807214e-05,
+      "loss": 0.4787,
+      "step": 19788
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.319840069471156e-05,
+      "loss": 0.48,
+      "step": 19789
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.318934659883967e-05,
+      "loss": 0.3904,
+      "step": 19790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.318029319056607e-05,
+      "loss": 0.3754,
+      "step": 19791
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.317124047000033e-05,
+      "loss": 0.471,
+      "step": 19792
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.316218843725197e-05,
+      "loss": 0.4262,
+      "step": 19793
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3153137092430654e-05,
+      "loss": 0.4748,
+      "step": 19794
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.314408643564585e-05,
+      "loss": 0.4959,
+      "step": 19795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.313503646700714e-05,
+      "loss": 0.503,
+      "step": 19796
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3125987186624036e-05,
+      "loss": 0.4153,
+      "step": 19797
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.311693859460605e-05,
+      "loss": 0.407,
+      "step": 19798
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.310789069106276e-05,
+      "loss": 0.4423,
+      "step": 19799
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3098843476103626e-05,
+      "loss": 0.4313,
+      "step": 19800
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 0.4471443295478821,
+      "eval_runtime": 856.0559,
+      "eval_samples_per_second": 28.106,
+      "eval_steps_per_second": 14.053,
+      "step": 19800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.308979694983817e-05,
+      "loss": 0.4578,
+      "step": 19801
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.308075111237588e-05,
+      "loss": 0.4219,
+      "step": 19802
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3071705963826204e-05,
+      "loss": 0.4522,
+      "step": 19803
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.306266150429869e-05,
+      "loss": 0.4588,
+      "step": 19804
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.305361773390276e-05,
+      "loss": 0.4452,
+      "step": 19805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3044574652747896e-05,
+      "loss": 0.4615,
+      "step": 19806
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.303553226094352e-05,
+      "loss": 0.459,
+      "step": 19807
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.302649055859911e-05,
+      "loss": 0.4766,
+      "step": 19808
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.301744954582404e-05,
+      "loss": 0.4845,
+      "step": 19809
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3008409222727806e-05,
+      "loss": 0.4904,
+      "step": 19810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.299936958941981e-05,
+      "loss": 0.4124,
+      "step": 19811
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2990330646009446e-05,
+      "loss": 0.4263,
+      "step": 19812
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.298129239260612e-05,
+      "loss": 0.4536,
+      "step": 19813
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2972254829319194e-05,
+      "loss": 0.5284,
+      "step": 19814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.296321795625812e-05,
+      "loss": 0.482,
+      "step": 19815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.295418177353223e-05,
+      "loss": 0.4532,
+      "step": 19816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2945146281250916e-05,
+      "loss": 0.4449,
+      "step": 19817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.29361114795235e-05,
+      "loss": 0.4773,
+      "step": 19818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2927077368459334e-05,
+      "loss": 0.4394,
+      "step": 19819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.291804394816781e-05,
+      "loss": 0.441,
+      "step": 19820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2909011218758234e-05,
+      "loss": 0.5325,
+      "step": 19821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.289997918033993e-05,
+      "loss": 0.4519,
+      "step": 19822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.289094783302223e-05,
+      "loss": 0.4386,
+      "step": 19823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2881917176914385e-05,
+      "loss": 0.5086,
+      "step": 19824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.287288721212578e-05,
+      "loss": 0.4478,
+      "step": 19825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.286385793876567e-05,
+      "loss": 0.4344,
+      "step": 19826
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.285482935694335e-05,
+      "loss": 0.4525,
+      "step": 19827
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2845801466768075e-05,
+      "loss": 0.457,
+      "step": 19828
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.283677426834909e-05,
+      "loss": 0.4585,
+      "step": 19829
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.282774776179572e-05,
+      "loss": 0.4299,
+      "step": 19830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.281872194721718e-05,
+      "loss": 0.4483,
+      "step": 19831
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.280969682472272e-05,
+      "loss": 0.4305,
+      "step": 19832
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.280067239442157e-05,
+      "loss": 0.452,
+      "step": 19833
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2791648656422924e-05,
+      "loss": 0.537,
+      "step": 19834
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2782625610836056e-05,
+      "loss": 0.3853,
+      "step": 19835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.277360325777016e-05,
+      "loss": 0.5445,
+      "step": 19836
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2764581597334406e-05,
+      "loss": 0.4875,
+      "step": 19837
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.275556062963802e-05,
+      "loss": 0.3891,
+      "step": 19838
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2746540354790176e-05,
+      "loss": 0.4104,
+      "step": 19839
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.273752077290001e-05,
+      "loss": 0.4229,
+      "step": 19840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.272850188407675e-05,
+      "loss": 0.4025,
+      "step": 19841
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.271948368842954e-05,
+      "loss": 0.4828,
+      "step": 19842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.271046618606751e-05,
+      "loss": 0.4862,
+      "step": 19843
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.270144937709981e-05,
+      "loss": 0.4336,
+      "step": 19844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.269243326163554e-05,
+      "loss": 0.4865,
+      "step": 19845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.26834178397839e-05,
+      "loss": 0.4469,
+      "step": 19846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.267440311165397e-05,
+      "loss": 0.3517,
+      "step": 19847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2665389077354847e-05,
+      "loss": 0.48,
+      "step": 19848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2656375736995634e-05,
+      "loss": 0.3805,
+      "step": 19849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.26473630906854e-05,
+      "loss": 0.4986,
+      "step": 19850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2638351138533296e-05,
+      "loss": 0.526,
+      "step": 19851
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2629339880648346e-05,
+      "loss": 0.3907,
+      "step": 19852
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.262032931713962e-05,
+      "loss": 0.4863,
+      "step": 19853
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2611319448116196e-05,
+      "loss": 0.3915,
+      "step": 19854
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.260231027368705e-05,
+      "loss": 0.4956,
+      "step": 19855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.259330179396134e-05,
+      "loss": 0.4049,
+      "step": 19856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.258429400904803e-05,
+      "loss": 0.4254,
+      "step": 19857
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.257528691905616e-05,
+      "loss": 0.3976,
+      "step": 19858
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2566280524094736e-05,
+      "loss": 0.4237,
+      "step": 19859
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.255727482427274e-05,
+      "loss": 0.5381,
+      "step": 19860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.254826981969924e-05,
+      "loss": 0.481,
+      "step": 19861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2539265510483175e-05,
+      "loss": 0.396,
+      "step": 19862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.253026189673355e-05,
+      "loss": 0.439,
+      "step": 19863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.252125897855932e-05,
+      "loss": 0.5432,
+      "step": 19864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.251225675606944e-05,
+      "loss": 0.4997,
+      "step": 19865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.25032552293729e-05,
+      "loss": 0.6141,
+      "step": 19866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.249425439857865e-05,
+      "loss": 0.5287,
+      "step": 19867
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.24852542637956e-05,
+      "loss": 0.3819,
+      "step": 19868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.247625482513271e-05,
+      "loss": 0.4397,
+      "step": 19869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.246725608269885e-05,
+      "loss": 0.4415,
+      "step": 19870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2458258036603e-05,
+      "loss": 0.4521,
+      "step": 19871
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.244926068695405e-05,
+      "loss": 0.4812,
+      "step": 19872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.244026403386089e-05,
+      "loss": 0.4654,
+      "step": 19873
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.24312680774324e-05,
+      "loss": 0.4241,
+      "step": 19874
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.242227281777748e-05,
+      "loss": 0.4628,
+      "step": 19875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.241327825500494e-05,
+      "loss": 0.4642,
+      "step": 19876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2404284389223734e-05,
+      "loss": 0.4372,
+      "step": 19877
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2395291220542686e-05,
+      "loss": 0.4101,
+      "step": 19878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.238629874907064e-05,
+      "loss": 0.5927,
+      "step": 19879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237730697491642e-05,
+      "loss": 0.4586,
+      "step": 19880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.236831589818884e-05,
+      "loss": 0.4699,
+      "step": 19881
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2359325518996787e-05,
+      "loss": 0.4563,
+      "step": 19882
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.235033583744904e-05,
+      "loss": 0.4638,
+      "step": 19883
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.234134685365439e-05,
+      "loss": 0.5086,
+      "step": 19884
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.233235856772165e-05,
+      "loss": 0.4701,
+      "step": 19885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.232337097975957e-05,
+      "loss": 0.4329,
+      "step": 19886
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2314384089877005e-05,
+      "loss": 0.4495,
+      "step": 19887
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230539789818271e-05,
+      "loss": 0.4572,
+      "step": 19888
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2296412404785394e-05,
+      "loss": 0.4213,
+      "step": 19889
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2287427609793796e-05,
+      "loss": 0.4685,
+      "step": 19890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.227844351331675e-05,
+      "loss": 0.4079,
+      "step": 19891
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.226946011546295e-05,
+      "loss": 0.4337,
+      "step": 19892
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.226047741634112e-05,
+      "loss": 0.4601,
+      "step": 19893
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.225149541605998e-05,
+      "loss": 0.3995,
+      "step": 19894
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.224251411472821e-05,
+      "loss": 0.4391,
+      "step": 19895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2233533512454606e-05,
+      "loss": 0.5096,
+      "step": 19896
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.222455360934779e-05,
+      "loss": 0.5377,
+      "step": 19897
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.221557440551648e-05,
+      "loss": 0.3958,
+      "step": 19898
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2206595901069336e-05,
+      "loss": 0.4574,
+      "step": 19899
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.219761809611499e-05,
+      "loss": 0.411,
+      "step": 19900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.218864099076219e-05,
+      "loss": 0.3748,
+      "step": 19901
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.217966458511954e-05,
+      "loss": 0.4944,
+      "step": 19902
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2170688879295693e-05,
+      "loss": 0.5148,
+      "step": 19903
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2161713873399276e-05,
+      "loss": 0.4238,
+      "step": 19904
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.215273956753889e-05,
+      "loss": 0.502,
+      "step": 19905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2143765961823224e-05,
+      "loss": 0.4973,
+      "step": 19906
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.213479305636086e-05,
+      "loss": 0.4308,
+      "step": 19907
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.212582085126038e-05,
+      "loss": 0.4321,
+      "step": 19908
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.211684934663038e-05,
+      "loss": 0.439,
+      "step": 19909
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2107878542579474e-05,
+      "loss": 0.4579,
+      "step": 19910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.209890843921617e-05,
+      "loss": 0.5307,
+      "step": 19911
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2089939036649115e-05,
+      "loss": 0.4455,
+      "step": 19912
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2080970334986843e-05,
+      "loss": 0.3866,
+      "step": 19913
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.20720023343379e-05,
+      "loss": 0.5566,
+      "step": 19914
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.206303503481084e-05,
+      "loss": 0.4153,
+      "step": 19915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.205406843651414e-05,
+      "loss": 0.4388,
+      "step": 19916
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.20451025395564e-05,
+      "loss": 0.4611,
+      "step": 19917
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2036137344046125e-05,
+      "loss": 0.433,
+      "step": 19918
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2027172850091786e-05,
+      "loss": 0.4603,
+      "step": 19919
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.201820905780193e-05,
+      "loss": 0.5258,
+      "step": 19920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.200924596728497e-05,
+      "loss": 0.4779,
+      "step": 19921
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.200028357864949e-05,
+      "loss": 0.5071,
+      "step": 19922
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.199132189200391e-05,
+      "loss": 0.5954,
+      "step": 19923
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.198236090745671e-05,
+      "loss": 0.5059,
+      "step": 19924
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.197340062511633e-05,
+      "loss": 0.4747,
+      "step": 19925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.196444104509121e-05,
+      "loss": 0.4356,
+      "step": 19926
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.195548216748985e-05,
+      "loss": 0.4604,
+      "step": 19927
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.194652399242063e-05,
+      "loss": 0.5631,
+      "step": 19928
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1937566519992e-05,
+      "loss": 0.455,
+      "step": 19929
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.192860975031235e-05,
+      "loss": 0.4298,
+      "step": 19930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.191965368349008e-05,
+      "loss": 0.4273,
+      "step": 19931
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1910698319633635e-05,
+      "loss": 0.4863,
+      "step": 19932
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.190174365885138e-05,
+      "loss": 0.4851,
+      "step": 19933
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.189278970125169e-05,
+      "loss": 0.5441,
+      "step": 19934
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.188383644694295e-05,
+      "loss": 0.4363,
+      "step": 19935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.187488389603348e-05,
+      "loss": 0.4482,
+      "step": 19936
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.186593204863171e-05,
+      "loss": 0.478,
+      "step": 19937
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1856980904845935e-05,
+      "loss": 0.4391,
+      "step": 19938
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1848030464784514e-05,
+      "loss": 0.4012,
+      "step": 19939
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.183908072855578e-05,
+      "loss": 0.4457,
+      "step": 19940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.183013169626804e-05,
+      "loss": 0.4251,
+      "step": 19941
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1821183368029573e-05,
+      "loss": 0.409,
+      "step": 19942
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1812235743948755e-05,
+      "loss": 0.4849,
+      "step": 19943
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1803288824133866e-05,
+      "loss": 0.4664,
+      "step": 19944
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.179434260869317e-05,
+      "loss": 0.484,
+      "step": 19945
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.178539709773496e-05,
+      "loss": 0.5454,
+      "step": 19946
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.177645229136745e-05,
+      "loss": 0.4759,
+      "step": 19947
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1767508189699e-05,
+      "loss": 0.3658,
+      "step": 19948
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.175856479283782e-05,
+      "loss": 0.5391,
+      "step": 19949
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.174962210089214e-05,
+      "loss": 0.5499,
+      "step": 19950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.174068011397021e-05,
+      "loss": 0.4107,
+      "step": 19951
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.173173883218022e-05,
+      "loss": 0.4097,
+      "step": 19952
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.172279825563047e-05,
+      "loss": 0.4191,
+      "step": 19953
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.171385838442911e-05,
+      "loss": 0.5271,
+      "step": 19954
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.170491921868437e-05,
+      "loss": 0.5277,
+      "step": 19955
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.169598075850442e-05,
+      "loss": 0.441,
+      "step": 19956
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1687043003997426e-05,
+      "loss": 0.4292,
+      "step": 19957
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.167810595527163e-05,
+      "loss": 0.5154,
+      "step": 19958
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1669169612435164e-05,
+      "loss": 0.3791,
+      "step": 19959
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.16602339755962e-05,
+      "loss": 0.4027,
+      "step": 19960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.165129904486286e-05,
+      "loss": 0.5091,
+      "step": 19961
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.164236482034327e-05,
+      "loss": 0.4974,
+      "step": 19962
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.163343130214564e-05,
+      "loss": 0.4912,
+      "step": 19963
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.162449849037806e-05,
+      "loss": 0.4699,
+      "step": 19964
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.161556638514863e-05,
+      "loss": 0.4814,
+      "step": 19965
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.160663498656547e-05,
+      "loss": 0.4308,
+      "step": 19966
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.159770429473665e-05,
+      "loss": 0.4226,
+      "step": 19967
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1588774309770316e-05,
+      "loss": 0.4531,
+      "step": 19968
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1579845031774524e-05,
+      "loss": 0.4222,
+      "step": 19969
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1570916460857354e-05,
+      "loss": 0.4995,
+      "step": 19970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1561988597126854e-05,
+      "loss": 0.4312,
+      "step": 19971
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.155306144069111e-05,
+      "loss": 0.4486,
+      "step": 19972
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1544134991658104e-05,
+      "loss": 0.4854,
+      "step": 19973
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1535209250135964e-05,
+      "loss": 0.4424,
+      "step": 19974
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1526284216232684e-05,
+      "loss": 0.5187,
+      "step": 19975
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1517359890056276e-05,
+      "loss": 0.474,
+      "step": 19976
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1508436271714755e-05,
+      "loss": 0.5142,
+      "step": 19977
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1499513361316114e-05,
+      "loss": 0.5056,
+      "step": 19978
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.149059115896841e-05,
+      "loss": 0.5042,
+      "step": 19979
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.148166966477957e-05,
+      "loss": 0.4721,
+      "step": 19980
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.147274887885761e-05,
+      "loss": 0.4347,
+      "step": 19981
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.146382880131049e-05,
+      "loss": 0.4582,
+      "step": 19982
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1454909432246124e-05,
+      "loss": 0.5147,
+      "step": 19983
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.144599077177255e-05,
+      "loss": 0.4251,
+      "step": 19984
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.143707281999767e-05,
+      "loss": 0.5027,
+      "step": 19985
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1428155577029426e-05,
+      "loss": 0.3796,
+      "step": 19986
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.141923904297575e-05,
+      "loss": 0.3873,
+      "step": 19987
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.141032321794453e-05,
+      "loss": 0.4269,
+      "step": 19988
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.140140810204373e-05,
+      "loss": 0.4968,
+      "step": 19989
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.139249369538123e-05,
+      "loss": 0.4415,
+      "step": 19990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1383579998064916e-05,
+      "loss": 0.4345,
+      "step": 19991
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1374667010202687e-05,
+      "loss": 0.4559,
+      "step": 19992
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1365754731902376e-05,
+      "loss": 0.3816,
+      "step": 19993
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.135684316327191e-05,
+      "loss": 0.4378,
+      "step": 19994
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1347932304419134e-05,
+      "loss": 0.3124,
+      "step": 19995
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.133902215545189e-05,
+      "loss": 0.4063,
+      "step": 19996
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.133011271647803e-05,
+      "loss": 0.3971,
+      "step": 19997
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.132120398760533e-05,
+      "loss": 0.5209,
+      "step": 19998
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1312295968941704e-05,
+      "loss": 0.4294,
+      "step": 19999
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.130338866059493e-05,
+      "loss": 0.4353,
+      "step": 20000
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 0.4445178806781769,
+      "eval_runtime": 851.846,
+      "eval_samples_per_second": 28.245,
+      "eval_steps_per_second": 14.122,
+      "step": 20000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.129448206267282e-05,
+      "loss": 0.5116,
+      "step": 20001
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1285576175283166e-05,
+      "loss": 0.4967,
+      "step": 20002
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.127667099853373e-05,
+      "loss": 0.3923,
+      "step": 20003
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.126776653253236e-05,
+      "loss": 0.4556,
+      "step": 20004
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1258862777386797e-05,
+      "loss": 0.5023,
+      "step": 20005
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.12499597332048e-05,
+      "loss": 0.4336,
+      "step": 20006
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.124105740009414e-05,
+      "loss": 0.4596,
+      "step": 20007
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.123215577816254e-05,
+      "loss": 0.4605,
+      "step": 20008
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.122325486751774e-05,
+      "loss": 0.4865,
+      "step": 20009
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1214354668267495e-05,
+      "loss": 0.467,
+      "step": 20010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1205455180519534e-05,
+      "loss": 0.4352,
+      "step": 20011
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1196556404381556e-05,
+      "loss": 0.5359,
+      "step": 20012
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.118765833996126e-05,
+      "loss": 0.4342,
+      "step": 20013
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1178760987366307e-05,
+      "loss": 0.4412,
+      "step": 20014
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.116986434670446e-05,
+      "loss": 0.459,
+      "step": 20015
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.116096841808337e-05,
+      "loss": 0.5077,
+      "step": 20016
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.115207320161069e-05,
+      "loss": 0.4664,
+      "step": 20017
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.114317869739409e-05,
+      "loss": 0.4852,
+      "step": 20018
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.11342849055412e-05,
+      "loss": 0.4368,
+      "step": 20019
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.112539182615971e-05,
+      "loss": 0.3917,
+      "step": 20020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.111649945935724e-05,
+      "loss": 0.4336,
+      "step": 20021
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.110760780524141e-05,
+      "loss": 0.4814,
+      "step": 20022
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.109871686391985e-05,
+      "loss": 0.3743,
+      "step": 20023
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.108982663550015e-05,
+      "loss": 0.4634,
+      "step": 20024
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.108093712008992e-05,
+      "loss": 0.5054,
+      "step": 20025
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.107204831779676e-05,
+      "loss": 0.4587,
+      "step": 20026
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.106316022872825e-05,
+      "loss": 0.4799,
+      "step": 20027
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.105427285299193e-05,
+      "loss": 0.4976,
+      "step": 20028
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1045386190695435e-05,
+      "loss": 0.5435,
+      "step": 20029
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.103650024194629e-05,
+      "loss": 0.47,
+      "step": 20030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.102761500685204e-05,
+      "loss": 0.4896,
+      "step": 20031
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.101873048552023e-05,
+      "loss": 0.4329,
+      "step": 20032
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.100984667805836e-05,
+      "loss": 0.4132,
+      "step": 20033
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1000963584574017e-05,
+      "loss": 0.4257,
+      "step": 20034
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0992081205174695e-05,
+      "loss": 0.5044,
+      "step": 20035
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.098319953996787e-05,
+      "loss": 0.3908,
+      "step": 20036
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0974318589061075e-05,
+      "loss": 0.4475,
+      "step": 20037
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.096543835256176e-05,
+      "loss": 0.4081,
+      "step": 20038
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0956558830577454e-05,
+      "loss": 0.4084,
+      "step": 20039
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0947680023215594e-05,
+      "loss": 0.3475,
+      "step": 20040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.093880193058366e-05,
+      "loss": 0.4849,
+      "step": 20041
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.092992455278911e-05,
+      "loss": 0.5369,
+      "step": 20042
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0921047889939366e-05,
+      "loss": 0.5,
+      "step": 20043
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.091217194214185e-05,
+      "loss": 0.423,
+      "step": 20044
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.090329670950406e-05,
+      "loss": 0.4897,
+      "step": 20045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.089442219213338e-05,
+      "loss": 0.4053,
+      "step": 20046
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0885548390137194e-05,
+      "loss": 0.4578,
+      "step": 20047
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0876675303622944e-05,
+      "loss": 0.3545,
+      "step": 20048
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.086780293269797e-05,
+      "loss": 0.4425,
+      "step": 20049
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.085893127746974e-05,
+      "loss": 0.4197,
+      "step": 20050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.085006033804557e-05,
+      "loss": 0.3848,
+      "step": 20051
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0841190114532845e-05,
+      "loss": 0.4727,
+      "step": 20052
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0832320607038935e-05,
+      "loss": 0.365,
+      "step": 20053
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.082345181567113e-05,
+      "loss": 0.4858,
+      "step": 20054
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.081458374053686e-05,
+      "loss": 0.4755,
+      "step": 20055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.080571638174342e-05,
+      "loss": 0.4625,
+      "step": 20056
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.079684973939812e-05,
+      "loss": 0.436,
+      "step": 20057
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.078798381360831e-05,
+      "loss": 0.4773,
+      "step": 20058
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.077911860448123e-05,
+      "loss": 0.4731,
+      "step": 20059
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0770254112124264e-05,
+      "loss": 0.4935,
+      "step": 20060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.076139033664466e-05,
+      "loss": 0.4784,
+      "step": 20061
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0752527278149696e-05,
+      "loss": 0.4859,
+      "step": 20062
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.074366493674666e-05,
+      "loss": 0.4614,
+      "step": 20063
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.073480331254277e-05,
+      "loss": 0.4608,
+      "step": 20064
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.072594240564534e-05,
+      "loss": 0.4399,
+      "step": 20065
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.07170822161616e-05,
+      "loss": 0.5436,
+      "step": 20066
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.070822274419879e-05,
+      "loss": 0.3976,
+      "step": 20067
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0699363989864126e-05,
+      "loss": 0.4911,
+      "step": 20068
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0690505953264804e-05,
+      "loss": 0.4419,
+      "step": 20069
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0681648634508096e-05,
+      "loss": 0.3881,
+      "step": 20070
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.067279203370117e-05,
+      "loss": 0.4553,
+      "step": 20071
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.066393615095123e-05,
+      "loss": 0.4399,
+      "step": 20072
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0655080986365444e-05,
+      "loss": 0.4551,
+      "step": 20073
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.064622654005101e-05,
+      "loss": 0.4813,
+      "step": 20074
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.063737281211506e-05,
+      "loss": 0.5225,
+      "step": 20075
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.062851980266481e-05,
+      "loss": 0.4635,
+      "step": 20076
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.061966751180738e-05,
+      "loss": 0.5063,
+      "step": 20077
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0610815939649924e-05,
+      "loss": 0.466,
+      "step": 20078
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.060196508629955e-05,
+      "loss": 0.4249,
+      "step": 20079
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.059311495186338e-05,
+      "loss": 0.4432,
+      "step": 20080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.058426553644857e-05,
+      "loss": 0.3775,
+      "step": 20081
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.057541684016222e-05,
+      "loss": 0.5509,
+      "step": 20082
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.056656886311141e-05,
+      "loss": 0.4038,
+      "step": 20083
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.055772160540323e-05,
+      "loss": 0.4701,
+      "step": 20084
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.054887506714474e-05,
+      "loss": 0.4758,
+      "step": 20085
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0540029248443065e-05,
+      "loss": 0.4485,
+      "step": 20086
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0531184149405254e-05,
+      "loss": 0.5393,
+      "step": 20087
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.052233977013834e-05,
+      "loss": 0.4565,
+      "step": 20088
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.051349611074938e-05,
+      "loss": 0.5715,
+      "step": 20089
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0504653171345384e-05,
+      "loss": 0.3383,
+      "step": 20090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.049581095203344e-05,
+      "loss": 0.4559,
+      "step": 20091
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.048696945292054e-05,
+      "loss": 0.5106,
+      "step": 20092
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.04781286741137e-05,
+      "loss": 0.3721,
+      "step": 20093
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0469288615719914e-05,
+      "loss": 0.4169,
+      "step": 20094
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0460449277846136e-05,
+      "loss": 0.4686,
+      "step": 20095
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.045161066059943e-05,
+      "loss": 0.6032,
+      "step": 20096
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.044277276408673e-05,
+      "loss": 0.477,
+      "step": 20097
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.043393558841503e-05,
+      "loss": 0.4661,
+      "step": 20098
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.042509913369126e-05,
+      "loss": 0.4073,
+      "step": 20099
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0416263400022334e-05,
+      "loss": 0.4667,
+      "step": 20100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.040742838751529e-05,
+      "loss": 0.3842,
+      "step": 20101
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.039859409627701e-05,
+      "loss": 0.4294,
+      "step": 20102
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0389760526414424e-05,
+      "loss": 0.4832,
+      "step": 20103
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.038092767803443e-05,
+      "loss": 0.452,
+      "step": 20104
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.037209555124392e-05,
+      "loss": 0.5162,
+      "step": 20105
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.036326414614985e-05,
+      "loss": 0.4262,
+      "step": 20106
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0354433462859085e-05,
+      "loss": 0.4948,
+      "step": 20107
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.03456035014785e-05,
+      "loss": 0.4875,
+      "step": 20108
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0336774262114974e-05,
+      "loss": 0.4749,
+      "step": 20109
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.032794574487536e-05,
+      "loss": 0.5131,
+      "step": 20110
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.031911794986647e-05,
+      "loss": 0.5392,
+      "step": 20111
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.031029087719525e-05,
+      "loss": 0.4397,
+      "step": 20112
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.030146452696847e-05,
+      "loss": 0.4245,
+      "step": 20113
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.029263889929298e-05,
+      "loss": 0.4113,
+      "step": 20114
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0283813994275576e-05,
+      "loss": 0.421,
+      "step": 20115
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0274989812023056e-05,
+      "loss": 0.5198,
+      "step": 20116
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.026616635264229e-05,
+      "loss": 0.4062,
+      "step": 20117
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0257343616240026e-05,
+      "loss": 0.4624,
+      "step": 20118
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.024852160292306e-05,
+      "loss": 0.479,
+      "step": 20119
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0239700312798155e-05,
+      "loss": 0.5004,
+      "step": 20120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.023087974597205e-05,
+      "loss": 0.4842,
+      "step": 20121
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.022205990255157e-05,
+      "loss": 0.5107,
+      "step": 20122
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.021324078264343e-05,
+      "loss": 0.3841,
+      "step": 20123
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.020442238635438e-05,
+      "loss": 0.4822,
+      "step": 20124
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0195604713791144e-05,
+      "loss": 0.4733,
+      "step": 20125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.018678776506041e-05,
+      "loss": 0.4639,
+      "step": 20126
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.017797154026897e-05,
+      "loss": 0.4939,
+      "step": 20127
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.016915603952347e-05,
+      "loss": 0.4519,
+      "step": 20128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0160341262930636e-05,
+      "loss": 0.4479,
+      "step": 20129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.015152721059715e-05,
+      "loss": 0.4375,
+      "step": 20130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.014271388262964e-05,
+      "loss": 0.443,
+      "step": 20131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0133901279134864e-05,
+      "loss": 0.4701,
+      "step": 20132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.012508940021945e-05,
+      "loss": 0.4757,
+      "step": 20133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.011627824599004e-05,
+      "loss": 0.3747,
+      "step": 20134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.010746781655328e-05,
+      "loss": 0.4369,
+      "step": 20135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.009865811201577e-05,
+      "loss": 0.3851,
+      "step": 20136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.008984913248421e-05,
+      "loss": 0.4004,
+      "step": 20137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.008104087806519e-05,
+      "loss": 0.5322,
+      "step": 20138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.007223334886531e-05,
+      "loss": 0.352,
+      "step": 20139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.006342654499117e-05,
+      "loss": 0.4701,
+      "step": 20140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.005462046654936e-05,
+      "loss": 0.4887,
+      "step": 20141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.004581511364645e-05,
+      "loss": 0.3891,
+      "step": 20142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.003701048638905e-05,
+      "loss": 0.4264,
+      "step": 20143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.002820658488371e-05,
+      "loss": 0.396,
+      "step": 20144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.001940340923699e-05,
+      "loss": 0.4636,
+      "step": 20145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.001060095955542e-05,
+      "loss": 0.4326,
+      "step": 20146
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0001799235945525e-05,
+      "loss": 0.4642,
+      "step": 20147
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.999299823851389e-05,
+      "loss": 0.4008,
+      "step": 20148
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.998419796736701e-05,
+      "loss": 0.517,
+      "step": 20149
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9975398422611386e-05,
+      "loss": 0.4814,
+      "step": 20150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.996659960435354e-05,
+      "loss": 0.4402,
+      "step": 20151
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9957801512699924e-05,
+      "loss": 0.4305,
+      "step": 20152
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9949004147757075e-05,
+      "loss": 0.4541,
+      "step": 20153
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.994020750963147e-05,
+      "loss": 0.4563,
+      "step": 20154
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9931411598429555e-05,
+      "loss": 0.4155,
+      "step": 20155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9922616414257794e-05,
+      "loss": 0.4786,
+      "step": 20156
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.991382195722261e-05,
+      "loss": 0.4606,
+      "step": 20157
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9905028227430543e-05,
+      "loss": 0.4305,
+      "step": 20158
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.989623522498791e-05,
+      "loss": 0.504,
+      "step": 20159
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9887442950001195e-05,
+      "loss": 0.4875,
+      "step": 20160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.987865140257677e-05,
+      "loss": 0.4707,
+      "step": 20161
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9869860582821104e-05,
+      "loss": 0.5082,
+      "step": 20162
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.986107049084057e-05,
+      "loss": 0.5365,
+      "step": 20163
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.985228112674155e-05,
+      "loss": 0.4772,
+      "step": 20164
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984349249063043e-05,
+      "loss": 0.4136,
+      "step": 20165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.983470458261354e-05,
+      "loss": 0.3812,
+      "step": 20166
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982591740279732e-05,
+      "loss": 0.4401,
+      "step": 20167
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9817130951288084e-05,
+      "loss": 0.509,
+      "step": 20168
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9808345228192187e-05,
+      "loss": 0.4308,
+      "step": 20169
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979956023361595e-05,
+      "loss": 0.3861,
+      "step": 20170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979077596766568e-05,
+      "loss": 0.4141,
+      "step": 20171
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.978199243044777e-05,
+      "loss": 0.4106,
+      "step": 20172
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.977320962206846e-05,
+      "loss": 0.4637,
+      "step": 20173
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9764427542634096e-05,
+      "loss": 0.4734,
+      "step": 20174
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.975564619225095e-05,
+      "loss": 0.5055,
+      "step": 20175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9746865571025305e-05,
+      "loss": 0.442,
+      "step": 20176
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.97380856790634e-05,
+      "loss": 0.5583,
+      "step": 20177
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.972930651647158e-05,
+      "loss": 0.4646,
+      "step": 20178
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.972052808335605e-05,
+      "loss": 0.547,
+      "step": 20179
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.971175037982306e-05,
+      "loss": 0.4266,
+      "step": 20180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.970297340597888e-05,
+      "loss": 0.422,
+      "step": 20181
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.969419716192968e-05,
+      "loss": 0.4505,
+      "step": 20182
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.968542164778175e-05,
+      "loss": 0.3515,
+      "step": 20183
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9676646863641276e-05,
+      "loss": 0.4883,
+      "step": 20184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.966787280961446e-05,
+      "loss": 0.4764,
+      "step": 20185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.965909948580749e-05,
+      "loss": 0.4793,
+      "step": 20186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.965032689232653e-05,
+      "loss": 0.4439,
+      "step": 20187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.964155502927782e-05,
+      "loss": 0.4425,
+      "step": 20188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.96327838967675e-05,
+      "loss": 0.5235,
+      "step": 20189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.962401349490172e-05,
+      "loss": 0.4913,
+      "step": 20190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9615243823786654e-05,
+      "loss": 0.4399,
+      "step": 20191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9606474883528375e-05,
+      "loss": 0.4236,
+      "step": 20192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.95977066742331e-05,
+      "loss": 0.4379,
+      "step": 20193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.958893919600694e-05,
+      "loss": 0.4378,
+      "step": 20194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.958017244895598e-05,
+      "loss": 0.4174,
+      "step": 20195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.957140643318634e-05,
+      "loss": 0.5513,
+      "step": 20196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.956264114880409e-05,
+      "loss": 0.4691,
+      "step": 20197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9553876595915375e-05,
+      "loss": 0.4797,
+      "step": 20198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.954511277462626e-05,
+      "loss": 0.4993,
+      "step": 20199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9536349685042785e-05,
+      "loss": 0.4988,
+      "step": 20200
+    },
+    {
+      "epoch": 0.71,
+      "eval_loss": 0.44190868735313416,
+      "eval_runtime": 845.132,
+      "eval_samples_per_second": 28.469,
+      "eval_steps_per_second": 14.234,
+      "step": 20200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.952758732727104e-05,
+      "loss": 0.4601,
+      "step": 20201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.951882570141704e-05,
+      "loss": 0.4463,
+      "step": 20202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.951006480758689e-05,
+      "loss": 0.4219,
+      "step": 20203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9501304645886595e-05,
+      "loss": 0.5392,
+      "step": 20204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.949254521642218e-05,
+      "loss": 0.5803,
+      "step": 20205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.948378651929968e-05,
+      "loss": 0.4637,
+      "step": 20206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.947502855462505e-05,
+      "loss": 0.5142,
+      "step": 20207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.946627132250436e-05,
+      "loss": 0.4997,
+      "step": 20208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9457514823043576e-05,
+      "loss": 0.5014,
+      "step": 20209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9448759056348686e-05,
+      "loss": 0.3944,
+      "step": 20210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.944000402252565e-05,
+      "loss": 0.4843,
+      "step": 20211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.943124972168043e-05,
+      "loss": 0.4564,
+      "step": 20212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9422496153918965e-05,
+      "loss": 0.3932,
+      "step": 20213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.941374331934726e-05,
+      "loss": 0.5238,
+      "step": 20214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.940499121807123e-05,
+      "loss": 0.408,
+      "step": 20215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.939623985019679e-05,
+      "loss": 0.42,
+      "step": 20216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.938748921582986e-05,
+      "loss": 0.4005,
+      "step": 20217
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9378739315076344e-05,
+      "loss": 0.4265,
+      "step": 20218
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.936999014804218e-05,
+      "loss": 0.4396,
+      "step": 20219
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.936124171483325e-05,
+      "loss": 0.4385,
+      "step": 20220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.935249401555543e-05,
+      "loss": 0.4482,
+      "step": 20221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.934374705031459e-05,
+      "loss": 0.4352,
+      "step": 20222
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9335000819216584e-05,
+      "loss": 0.4561,
+      "step": 20223
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.932625532236731e-05,
+      "loss": 0.4087,
+      "step": 20224
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9317510559872615e-05,
+      "loss": 0.5336,
+      "step": 20225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.930876653183831e-05,
+      "loss": 0.4389,
+      "step": 20226
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.930002323837025e-05,
+      "loss": 0.3907,
+      "step": 20227
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9291280679574215e-05,
+      "loss": 0.5043,
+      "step": 20228
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.92825388555561e-05,
+      "loss": 0.3814,
+      "step": 20229
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9273797766421637e-05,
+      "loss": 0.4399,
+      "step": 20230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9265057412276665e-05,
+      "loss": 0.4556,
+      "step": 20231
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9256317793226947e-05,
+      "loss": 0.4411,
+      "step": 20232
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9247578909378245e-05,
+      "loss": 0.4572,
+      "step": 20233
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9238840760836385e-05,
+      "loss": 0.4098,
+      "step": 20234
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.923010334770709e-05,
+      "loss": 0.4707,
+      "step": 20235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9221366670096124e-05,
+      "loss": 0.5093,
+      "step": 20236
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.921263072810922e-05,
+      "loss": 0.4075,
+      "step": 20237
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9203895521852085e-05,
+      "loss": 0.5515,
+      "step": 20238
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9195161051430506e-05,
+      "loss": 0.5248,
+      "step": 20239
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.918642731695017e-05,
+      "loss": 0.4821,
+      "step": 20240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.917769431851678e-05,
+      "loss": 0.435,
+      "step": 20241
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.916896205623605e-05,
+      "loss": 0.4179,
+      "step": 20242
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9160230530213636e-05,
+      "loss": 0.3923,
+      "step": 20243
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.915149974055522e-05,
+      "loss": 0.3473,
+      "step": 20244
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.914276968736652e-05,
+      "loss": 0.6048,
+      "step": 20245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.913404037075317e-05,
+      "loss": 0.4726,
+      "step": 20246
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.912531179082083e-05,
+      "loss": 0.517,
+      "step": 20247
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9116583947675136e-05,
+      "loss": 0.4902,
+      "step": 20248
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9107856841421684e-05,
+      "loss": 0.41,
+      "step": 20249
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.90991304721662e-05,
+      "loss": 0.4549,
+      "step": 20250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.909040484001424e-05,
+      "loss": 0.4677,
+      "step": 20251
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.908167994507141e-05,
+      "loss": 0.4174,
+      "step": 20252
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.907295578744333e-05,
+      "loss": 0.4301,
+      "step": 20253
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.906423236723555e-05,
+      "loss": 0.5151,
+      "step": 20254
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9055509684553715e-05,
+      "loss": 0.4157,
+      "step": 20255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9046787739503366e-05,
+      "loss": 0.4505,
+      "step": 20256
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.903806653219008e-05,
+      "loss": 0.3986,
+      "step": 20257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9029346062719397e-05,
+      "loss": 0.4955,
+      "step": 20258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.902062633119683e-05,
+      "loss": 0.4443,
+      "step": 20259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.901190733772799e-05,
+      "loss": 0.4084,
+      "step": 20260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.900318908241838e-05,
+      "loss": 0.5052,
+      "step": 20261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.89944715653735e-05,
+      "loss": 0.5014,
+      "step": 20262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.898575478669889e-05,
+      "loss": 0.4345,
+      "step": 20263
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8977038746499996e-05,
+      "loss": 0.3956,
+      "step": 20264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.896832344488237e-05,
+      "loss": 0.4462,
+      "step": 20265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8959608881951495e-05,
+      "loss": 0.4269,
+      "step": 20266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.895089505781282e-05,
+      "loss": 0.4145,
+      "step": 20267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.894218197257181e-05,
+      "loss": 0.4676,
+      "step": 20268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.893346962633391e-05,
+      "loss": 0.5852,
+      "step": 20269
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8924758019204624e-05,
+      "loss": 0.4435,
+      "step": 20270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.891604715128935e-05,
+      "loss": 0.5057,
+      "step": 20271
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8907337022693524e-05,
+      "loss": 0.4309,
+      "step": 20272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8898627633522566e-05,
+      "loss": 0.424,
+      "step": 20273
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8889918983881865e-05,
+      "loss": 0.3752,
+      "step": 20274
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8881211073876885e-05,
+      "loss": 0.3861,
+      "step": 20275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8872503903613e-05,
+      "loss": 0.4915,
+      "step": 20276
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.886379747319556e-05,
+      "loss": 0.4598,
+      "step": 20277
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.885509178272998e-05,
+      "loss": 0.4359,
+      "step": 20278
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8846386832321614e-05,
+      "loss": 0.4382,
+      "step": 20279
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8837682622075775e-05,
+      "loss": 0.4383,
+      "step": 20280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.88289791520979e-05,
+      "loss": 0.4171,
+      "step": 20281
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8820276422493284e-05,
+      "loss": 0.4758,
+      "step": 20282
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.881157443336727e-05,
+      "loss": 0.3599,
+      "step": 20283
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.880287318482517e-05,
+      "loss": 0.4203,
+      "step": 20284
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8794172676972264e-05,
+      "loss": 0.4485,
+      "step": 20285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8785472909913936e-05,
+      "loss": 0.4294,
+      "step": 20286
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8776773883755435e-05,
+      "loss": 0.4074,
+      "step": 20287
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.876807559860206e-05,
+      "loss": 0.421,
+      "step": 20288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.875937805455908e-05,
+      "loss": 0.4129,
+      "step": 20289
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.875068125173174e-05,
+      "loss": 0.4906,
+      "step": 20290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8741985190225346e-05,
+      "loss": 0.3351,
+      "step": 20291
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.873328987014515e-05,
+      "loss": 0.4492,
+      "step": 20292
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8724595291596396e-05,
+      "loss": 0.4037,
+      "step": 20293
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8715901454684246e-05,
+      "loss": 0.4242,
+      "step": 20294
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.870720835951399e-05,
+      "loss": 0.4406,
+      "step": 20295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.869851600619084e-05,
+      "loss": 0.3999,
+      "step": 20296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8689824394819996e-05,
+      "loss": 0.4288,
+      "step": 20297
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.868113352550664e-05,
+      "loss": 0.4347,
+      "step": 20298
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.867244339835595e-05,
+      "loss": 0.4192,
+      "step": 20299
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.866375401347315e-05,
+      "loss": 0.3773,
+      "step": 20300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.86550653709634e-05,
+      "loss": 0.4085,
+      "step": 20301
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.864637747093184e-05,
+      "loss": 0.4183,
+      "step": 20302
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.863769031348363e-05,
+      "loss": 0.5143,
+      "step": 20303
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.862900389872389e-05,
+      "loss": 0.4563,
+      "step": 20304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.862031822675781e-05,
+      "loss": 0.3941,
+      "step": 20305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8611633297690487e-05,
+      "loss": 0.4789,
+      "step": 20306
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8602949111627026e-05,
+      "loss": 0.4651,
+      "step": 20307
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.859426566867256e-05,
+      "loss": 0.5798,
+      "step": 20308
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.858558296893213e-05,
+      "loss": 0.4988,
+      "step": 20309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.85769010125109e-05,
+      "loss": 0.4635,
+      "step": 20310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8568219799513924e-05,
+      "loss": 0.5627,
+      "step": 20311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.855953933004627e-05,
+      "loss": 0.5211,
+      "step": 20312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8550859604212994e-05,
+      "loss": 0.4107,
+      "step": 20313
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.854218062211915e-05,
+      "loss": 0.4437,
+      "step": 20314
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.853350238386977e-05,
+      "loss": 0.4749,
+      "step": 20315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.852482488956992e-05,
+      "loss": 0.4414,
+      "step": 20316
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.851614813932463e-05,
+      "loss": 0.3987,
+      "step": 20317
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8507472133238886e-05,
+      "loss": 0.4238,
+      "step": 20318
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.849879687141772e-05,
+      "loss": 0.3406,
+      "step": 20319
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.849012235396608e-05,
+      "loss": 0.4086,
+      "step": 20320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.848144858098905e-05,
+      "loss": 0.482,
+      "step": 20321
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8472775552591556e-05,
+      "loss": 0.4776,
+      "step": 20322
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.846410326887857e-05,
+      "loss": 0.4374,
+      "step": 20323
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.845543172995506e-05,
+      "loss": 0.4261,
+      "step": 20324
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8446760935925955e-05,
+      "loss": 0.4725,
+      "step": 20325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.843809088689626e-05,
+      "loss": 0.5272,
+      "step": 20326
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.842942158297087e-05,
+      "loss": 0.5502,
+      "step": 20327
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.842075302425473e-05,
+      "loss": 0.4241,
+      "step": 20328
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.841208521085274e-05,
+      "loss": 0.4927,
+      "step": 20329
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8403418142869794e-05,
+      "loss": 0.4751,
+      "step": 20330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.839475182041086e-05,
+      "loss": 0.515,
+      "step": 20331
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.838608624358077e-05,
+      "loss": 0.4402,
+      "step": 20332
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.837742141248444e-05,
+      "loss": 0.4705,
+      "step": 20333
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.836875732722671e-05,
+      "loss": 0.4428,
+      "step": 20334
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8360093987912436e-05,
+      "loss": 0.5941,
+      "step": 20335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.835143139464653e-05,
+      "loss": 0.5016,
+      "step": 20336
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.834276954753381e-05,
+      "loss": 0.4269,
+      "step": 20337
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.833410844667911e-05,
+      "loss": 0.432,
+      "step": 20338
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.832544809218725e-05,
+      "loss": 0.4551,
+      "step": 20339
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.831678848416302e-05,
+      "loss": 0.4491,
+      "step": 20340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.830812962271132e-05,
+      "loss": 0.3948,
+      "step": 20341
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.829947150793688e-05,
+      "loss": 0.4385,
+      "step": 20342
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.829081413994452e-05,
+      "loss": 0.5037,
+      "step": 20343
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8282157518839014e-05,
+      "loss": 0.4319,
+      "step": 20344
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8273501644725126e-05,
+      "loss": 0.4644,
+      "step": 20345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.826484651770761e-05,
+      "loss": 0.5449,
+      "step": 20346
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8256192137891264e-05,
+      "loss": 0.3728,
+      "step": 20347
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824753850538082e-05,
+      "loss": 0.484,
+      "step": 20348
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8238885620281004e-05,
+      "loss": 0.4503,
+      "step": 20349
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8230233482696554e-05,
+      "loss": 0.4181,
+      "step": 20350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8221582092732155e-05,
+      "loss": 0.4674,
+      "step": 20351
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8212931450492575e-05,
+      "loss": 0.4182,
+      "step": 20352
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820428155608249e-05,
+      "loss": 0.3968,
+      "step": 20353
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.81956324096066e-05,
+      "loss": 0.5079,
+      "step": 20354
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.818698401116956e-05,
+      "loss": 0.4258,
+      "step": 20355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.817833636087604e-05,
+      "loss": 0.4849,
+      "step": 20356
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.816968945883077e-05,
+      "loss": 0.4664,
+      "step": 20357
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8161043305138354e-05,
+      "loss": 0.4319,
+      "step": 20358
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.815239789990346e-05,
+      "loss": 0.4384,
+      "step": 20359
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.814375324323072e-05,
+      "loss": 0.3981,
+      "step": 20360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.813510933522472e-05,
+      "loss": 0.3544,
+      "step": 20361
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.812646617599015e-05,
+      "loss": 0.4246,
+      "step": 20362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.81178237656316e-05,
+      "loss": 0.4356,
+      "step": 20363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.810918210425366e-05,
+      "loss": 0.5007,
+      "step": 20364
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8100541191960926e-05,
+      "loss": 0.4319,
+      "step": 20365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8091901028857955e-05,
+      "loss": 0.411,
+      "step": 20366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.808326161504937e-05,
+      "loss": 0.4251,
+      "step": 20367
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8074622950639723e-05,
+      "loss": 0.5263,
+      "step": 20368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8065985035733566e-05,
+      "loss": 0.4427,
+      "step": 20369
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8057347870435445e-05,
+      "loss": 0.385,
+      "step": 20370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.804871145484986e-05,
+      "loss": 0.4411,
+      "step": 20371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8040075789081406e-05,
+      "loss": 0.4816,
+      "step": 20372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.803144087323459e-05,
+      "loss": 0.4443,
+      "step": 20373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8022806707413905e-05,
+      "loss": 0.4705,
+      "step": 20374
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8014173291723845e-05,
+      "loss": 0.4489,
+      "step": 20375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8005540626268924e-05,
+      "loss": 0.4417,
+      "step": 20376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.799690871115358e-05,
+      "loss": 0.4473,
+      "step": 20377
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7988277546482355e-05,
+      "loss": 0.357,
+      "step": 20378
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.79796471323597e-05,
+      "loss": 0.3754,
+      "step": 20379
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7971017468890036e-05,
+      "loss": 0.3808,
+      "step": 20380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.796238855617784e-05,
+      "loss": 0.4656,
+      "step": 20381
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.795376039432751e-05,
+      "loss": 0.3993,
+      "step": 20382
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.794513298344354e-05,
+      "loss": 0.3891,
+      "step": 20383
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.793650632363032e-05,
+      "loss": 0.4135,
+      "step": 20384
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.792788041499226e-05,
+      "loss": 0.4255,
+      "step": 20385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7919255257633756e-05,
+      "loss": 0.3564,
+      "step": 20386
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.791063085165917e-05,
+      "loss": 0.4275,
+      "step": 20387
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.790200719717296e-05,
+      "loss": 0.4908,
+      "step": 20388
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.789338429427947e-05,
+      "loss": 0.5403,
+      "step": 20389
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.788476214308305e-05,
+      "loss": 0.4185,
+      "step": 20390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7876140743688074e-05,
+      "loss": 0.4727,
+      "step": 20391
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.786752009619884e-05,
+      "loss": 0.5008,
+      "step": 20392
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.785890020071976e-05,
+      "loss": 0.4036,
+      "step": 20393
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.785028105735514e-05,
+      "loss": 0.421,
+      "step": 20394
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7841662666209285e-05,
+      "loss": 0.4216,
+      "step": 20395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7833045027386525e-05,
+      "loss": 0.4874,
+      "step": 20396
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.782442814099111e-05,
+      "loss": 0.4581,
+      "step": 20397
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.781581200712741e-05,
+      "loss": 0.4514,
+      "step": 20398
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7807196625899664e-05,
+      "loss": 0.4452,
+      "step": 20399
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7798581997412164e-05,
+      "loss": 0.4037,
+      "step": 20400
+    },
+    {
+      "epoch": 0.71,
+      "eval_loss": 0.4393974542617798,
+      "eval_runtime": 846.2989,
+      "eval_samples_per_second": 28.43,
+      "eval_steps_per_second": 14.215,
+      "step": 20400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.778996812176916e-05,
+      "loss": 0.3926,
+      "step": 20401
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.778135499907489e-05,
+      "loss": 0.4399,
+      "step": 20402
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.777274262943365e-05,
+      "loss": 0.315,
+      "step": 20403
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.776413101294967e-05,
+      "loss": 0.4466,
+      "step": 20404
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.775552014972715e-05,
+      "loss": 0.4011,
+      "step": 20405
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7746910039870346e-05,
+      "loss": 0.4719,
+      "step": 20406
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.77383006834834e-05,
+      "loss": 0.4343,
+      "step": 20407
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7729692080670595e-05,
+      "loss": 0.5849,
+      "step": 20408
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.772108423153609e-05,
+      "loss": 0.3972,
+      "step": 20409
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.771247713618408e-05,
+      "loss": 0.4275,
+      "step": 20410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.770387079471871e-05,
+      "loss": 0.4541,
+      "step": 20411
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.769526520724418e-05,
+      "loss": 0.5123,
+      "step": 20412
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.768666037386459e-05,
+      "loss": 0.4988,
+      "step": 20413
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.767805629468417e-05,
+      "loss": 0.4681,
+      "step": 20414
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.766945296980701e-05,
+      "loss": 0.4908,
+      "step": 20415
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.766085039933724e-05,
+      "loss": 0.4603,
+      "step": 20416
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.765224858337899e-05,
+      "loss": 0.4531,
+      "step": 20417
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.764364752203633e-05,
+      "loss": 0.6634,
+      "step": 20418
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.763504721541343e-05,
+      "loss": 0.4717,
+      "step": 20419
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.762644766361435e-05,
+      "loss": 0.5527,
+      "step": 20420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7617848866743176e-05,
+      "loss": 0.4587,
+      "step": 20421
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.760925082490398e-05,
+      "loss": 0.456,
+      "step": 20422
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.760065353820078e-05,
+      "loss": 0.3895,
+      "step": 20423
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7592057006737725e-05,
+      "loss": 0.404,
+      "step": 20424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.758346123061881e-05,
+      "loss": 0.4631,
+      "step": 20425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7574866209948065e-05,
+      "loss": 0.3991,
+      "step": 20426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.756627194482954e-05,
+      "loss": 0.4374,
+      "step": 20427
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.755767843536725e-05,
+      "loss": 0.4266,
+      "step": 20428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.754908568166519e-05,
+      "loss": 0.4243,
+      "step": 20429
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.754049368382737e-05,
+      "loss": 0.4539,
+      "step": 20430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.753190244195779e-05,
+      "loss": 0.4926,
+      "step": 20431
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.752331195616038e-05,
+      "loss": 0.4895,
+      "step": 20432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.75147222265392e-05,
+      "loss": 0.3839,
+      "step": 20433
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.750613325319817e-05,
+      "loss": 0.5098,
+      "step": 20434
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.749754503624124e-05,
+      "loss": 0.5123,
+      "step": 20435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7488957575772376e-05,
+      "loss": 0.4454,
+      "step": 20436
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7480370871895464e-05,
+      "loss": 0.4548,
+      "step": 20437
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.74717849247145e-05,
+      "loss": 0.3766,
+      "step": 20438
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7463199734333385e-05,
+      "loss": 0.5208,
+      "step": 20439
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.745461530085601e-05,
+      "loss": 0.4608,
+      "step": 20440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7446031624386276e-05,
+      "loss": 0.5135,
+      "step": 20441
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7437448705028056e-05,
+      "loss": 0.4293,
+      "step": 20442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.742886654288529e-05,
+      "loss": 0.4502,
+      "step": 20443
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.742028513806181e-05,
+      "loss": 0.4565,
+      "step": 20444
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.741170449066149e-05,
+      "loss": 0.4396,
+      "step": 20445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.740312460078818e-05,
+      "loss": 0.4551,
+      "step": 20446
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.739454546854571e-05,
+      "loss": 0.5253,
+      "step": 20447
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7385967094037924e-05,
+      "loss": 0.4281,
+      "step": 20448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7377389477368676e-05,
+      "loss": 0.4962,
+      "step": 20449
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.736881261864177e-05,
+      "loss": 0.4184,
+      "step": 20450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7360236517961013e-05,
+      "loss": 0.4506,
+      "step": 20451
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.735166117543018e-05,
+      "loss": 0.531,
+      "step": 20452
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7343086591153076e-05,
+      "loss": 0.4818,
+      "step": 20453
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.73345127652335e-05,
+      "loss": 0.4285,
+      "step": 20454
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7325939697775226e-05,
+      "loss": 0.4813,
+      "step": 20455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7317367388881994e-05,
+      "loss": 0.3958,
+      "step": 20456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.730879583865757e-05,
+      "loss": 0.4024,
+      "step": 20457
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.730022504720566e-05,
+      "loss": 0.4017,
+      "step": 20458
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7291655014630066e-05,
+      "loss": 0.4886,
+      "step": 20459
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7283085741034486e-05,
+      "loss": 0.4483,
+      "step": 20460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.727451722652262e-05,
+      "loss": 0.4522,
+      "step": 20461
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.72659494711982e-05,
+      "loss": 0.387,
+      "step": 20462
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.725738247516487e-05,
+      "loss": 0.4039,
+      "step": 20463
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.724881623852641e-05,
+      "loss": 0.4474,
+      "step": 20464
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7240250761386444e-05,
+      "loss": 0.3807,
+      "step": 20465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.723168604384863e-05,
+      "loss": 0.4361,
+      "step": 20466
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.722312208601668e-05,
+      "loss": 0.3875,
+      "step": 20467
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7214558887994166e-05,
+      "loss": 0.4774,
+      "step": 20468
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.720599644988482e-05,
+      "loss": 0.4262,
+      "step": 20469
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.719743477179224e-05,
+      "loss": 0.451,
+      "step": 20470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.718887385382004e-05,
+      "loss": 0.5187,
+      "step": 20471
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.718031369607186e-05,
+      "loss": 0.4088,
+      "step": 20472
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7171754298651243e-05,
+      "loss": 0.4806,
+      "step": 20473
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7163195661661864e-05,
+      "loss": 0.4106,
+      "step": 20474
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7154637785207294e-05,
+      "loss": 0.4262,
+      "step": 20475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7146080669391095e-05,
+      "loss": 0.5079,
+      "step": 20476
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.713752431431683e-05,
+      "loss": 0.4302,
+      "step": 20477
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7128968720088074e-05,
+      "loss": 0.3922,
+      "step": 20478
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.712041388680835e-05,
+      "loss": 0.4232,
+      "step": 20479
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.711185981458125e-05,
+      "loss": 0.4082,
+      "step": 20480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.710330650351028e-05,
+      "loss": 0.4098,
+      "step": 20481
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.709475395369897e-05,
+      "loss": 0.4398,
+      "step": 20482
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7086202165250825e-05,
+      "loss": 0.461,
+      "step": 20483
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.707765113826932e-05,
+      "loss": 0.491,
+      "step": 20484
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.706910087285802e-05,
+      "loss": 0.4159,
+      "step": 20485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7060551369120365e-05,
+      "loss": 0.4872,
+      "step": 20486
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7052002627159866e-05,
+      "loss": 0.4223,
+      "step": 20487
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.704345464707996e-05,
+      "loss": 0.5128,
+      "step": 20488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7034907428984064e-05,
+      "loss": 0.4695,
+      "step": 20489
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.702636097297574e-05,
+      "loss": 0.4667,
+      "step": 20490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7017815279158354e-05,
+      "loss": 0.4039,
+      "step": 20491
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.700927034763535e-05,
+      "loss": 0.431,
+      "step": 20492
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7000726178510145e-05,
+      "loss": 0.4034,
+      "step": 20493
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.699218277188614e-05,
+      "loss": 0.4788,
+      "step": 20494
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.698364012786678e-05,
+      "loss": 0.4143,
+      "step": 20495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.697509824655544e-05,
+      "loss": 0.4223,
+      "step": 20496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.69665571280555e-05,
+      "loss": 0.4328,
+      "step": 20497
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6958016772470336e-05,
+      "loss": 0.4047,
+      "step": 20498
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.694947717990329e-05,
+      "loss": 0.4361,
+      "step": 20499
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6940938350457776e-05,
+      "loss": 0.4055,
+      "step": 20500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.693240028423711e-05,
+      "loss": 0.4415,
+      "step": 20501
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6923862981344634e-05,
+      "loss": 0.4331,
+      "step": 20502
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.691532644188368e-05,
+      "loss": 0.5087,
+      "step": 20503
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.690679066595753e-05,
+      "loss": 0.4211,
+      "step": 20504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6898255653669565e-05,
+      "loss": 0.4189,
+      "step": 20505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6889721405123056e-05,
+      "loss": 0.4535,
+      "step": 20506
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.688118792042129e-05,
+      "loss": 0.4464,
+      "step": 20507
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.687265519966755e-05,
+      "loss": 0.4145,
+      "step": 20508
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.686412324296507e-05,
+      "loss": 0.4346,
+      "step": 20509
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.685559205041722e-05,
+      "loss": 0.5001,
+      "step": 20510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6847061622127175e-05,
+      "loss": 0.4834,
+      "step": 20511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.683853195819821e-05,
+      "loss": 0.4645,
+      "step": 20512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6830003058733556e-05,
+      "loss": 0.4128,
+      "step": 20513
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.682147492383643e-05,
+      "loss": 0.3967,
+      "step": 20514
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.681294755361003e-05,
+      "loss": 0.4351,
+      "step": 20515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6804420948157627e-05,
+      "loss": 0.5221,
+      "step": 20516
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.679589510758239e-05,
+      "loss": 0.4893,
+      "step": 20517
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6787370031987526e-05,
+      "loss": 0.5275,
+      "step": 20518
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6778845721476173e-05,
+      "loss": 0.4253,
+      "step": 20519
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6770322176151516e-05,
+      "loss": 0.4011,
+      "step": 20520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6761799396116755e-05,
+      "loss": 0.5275,
+      "step": 20521
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.675327738147504e-05,
+      "loss": 0.4409,
+      "step": 20522
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.674475613232949e-05,
+      "loss": 0.518,
+      "step": 20523
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6736235648783244e-05,
+      "loss": 0.482,
+      "step": 20524
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.672771593093941e-05,
+      "loss": 0.4825,
+      "step": 20525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.671919697890115e-05,
+      "loss": 0.5031,
+      "step": 20526
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6710678792771556e-05,
+      "loss": 0.4072,
+      "step": 20527
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.670216137265372e-05,
+      "loss": 0.5087,
+      "step": 20528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6693644718650736e-05,
+      "loss": 0.3887,
+      "step": 20529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.668512883086563e-05,
+      "loss": 0.4895,
+      "step": 20530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6676613709401575e-05,
+      "loss": 0.42,
+      "step": 20531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.666809935436157e-05,
+      "loss": 0.4844,
+      "step": 20532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.665958576584868e-05,
+      "loss": 0.5273,
+      "step": 20533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.665107294396595e-05,
+      "loss": 0.4587,
+      "step": 20534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6642560888816355e-05,
+      "loss": 0.5153,
+      "step": 20535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6634049600503015e-05,
+      "loss": 0.3874,
+      "step": 20536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6625539079128914e-05,
+      "loss": 0.4064,
+      "step": 20537
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.661702932479704e-05,
+      "loss": 0.4747,
+      "step": 20538
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.660852033761039e-05,
+      "loss": 0.4646,
+      "step": 20539
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.660001211767192e-05,
+      "loss": 0.551,
+      "step": 20540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6591504665084674e-05,
+      "loss": 0.4436,
+      "step": 20541
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6582997979951604e-05,
+      "loss": 0.4198,
+      "step": 20542
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6574492062375654e-05,
+      "loss": 0.4716,
+      "step": 20543
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.656598691245977e-05,
+      "loss": 0.4094,
+      "step": 20544
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.65574825303069e-05,
+      "loss": 0.4174,
+      "step": 20545
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.654897891601994e-05,
+      "loss": 0.4137,
+      "step": 20546
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.654047606970188e-05,
+      "loss": 0.4628,
+      "step": 20547
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.653197399145559e-05,
+      "loss": 0.4704,
+      "step": 20548
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6523472681384e-05,
+      "loss": 0.4836,
+      "step": 20549
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6514972139589975e-05,
+      "loss": 0.4895,
+      "step": 20550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.650647236617638e-05,
+      "loss": 0.5064,
+      "step": 20551
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.649797336124615e-05,
+      "loss": 0.3778,
+      "step": 20552
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.648947512490214e-05,
+      "loss": 0.4907,
+      "step": 20553
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6480977657247174e-05,
+      "loss": 0.4356,
+      "step": 20554
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.647248095838413e-05,
+      "loss": 0.4179,
+      "step": 20555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.646398502841578e-05,
+      "loss": 0.4611,
+      "step": 20556
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6455489867445045e-05,
+      "loss": 0.5243,
+      "step": 20557
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.644699547557471e-05,
+      "loss": 0.4433,
+      "step": 20558
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.643850185290758e-05,
+      "loss": 0.5224,
+      "step": 20559
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.643000899954645e-05,
+      "loss": 0.417,
+      "step": 20560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.642151691559409e-05,
+      "loss": 0.4268,
+      "step": 20561
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6413025601153384e-05,
+      "loss": 0.3949,
+      "step": 20562
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6404535056326996e-05,
+      "loss": 0.4553,
+      "step": 20563
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6396045281217715e-05,
+      "loss": 0.3854,
+      "step": 20564
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6387556275928284e-05,
+      "loss": 0.4477,
+      "step": 20565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6379068040561505e-05,
+      "loss": 0.455,
+      "step": 20566
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.637058057522008e-05,
+      "loss": 0.4072,
+      "step": 20567
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.636209388000673e-05,
+      "loss": 0.476,
+      "step": 20568
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.635360795502418e-05,
+      "loss": 0.4182,
+      "step": 20569
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.63451228003751e-05,
+      "loss": 0.3718,
+      "step": 20570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.633663841616226e-05,
+      "loss": 0.4614,
+      "step": 20571
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6328154802488314e-05,
+      "loss": 0.4121,
+      "step": 20572
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.631967195945595e-05,
+      "loss": 0.4579,
+      "step": 20573
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.631118988716782e-05,
+      "loss": 0.4821,
+      "step": 20574
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6302708585726565e-05,
+      "loss": 0.4296,
+      "step": 20575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.629422805523489e-05,
+      "loss": 0.4109,
+      "step": 20576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.628574829579543e-05,
+      "loss": 0.3986,
+      "step": 20577
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.62772693075108e-05,
+      "loss": 0.5334,
+      "step": 20578
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.626879109048363e-05,
+      "loss": 0.4562,
+      "step": 20579
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6260313644816535e-05,
+      "loss": 0.3961,
+      "step": 20580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6251836970612075e-05,
+      "loss": 0.4947,
+      "step": 20581
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.624336106797292e-05,
+      "loss": 0.3817,
+      "step": 20582
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6234885937001636e-05,
+      "loss": 0.328,
+      "step": 20583
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.622641157780079e-05,
+      "loss": 0.4711,
+      "step": 20584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.621793799047295e-05,
+      "loss": 0.4851,
+      "step": 20585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.620946517512064e-05,
+      "loss": 0.4631,
+      "step": 20586
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6200993131846474e-05,
+      "loss": 0.4609,
+      "step": 20587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.619252186075297e-05,
+      "loss": 0.4779,
+      "step": 20588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.618405136194264e-05,
+      "loss": 0.4211,
+      "step": 20589
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.617558163551802e-05,
+      "loss": 0.4059,
+      "step": 20590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.616711268158158e-05,
+      "loss": 0.4495,
+      "step": 20591
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6158644500235904e-05,
+      "loss": 0.5145,
+      "step": 20592
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6150177091583434e-05,
+      "loss": 0.4599,
+      "step": 20593
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6141710455726664e-05,
+      "loss": 0.4858,
+      "step": 20594
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6133244592768065e-05,
+      "loss": 0.427,
+      "step": 20595
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6124779502810066e-05,
+      "loss": 0.5584,
+      "step": 20596
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6116315185955194e-05,
+      "loss": 0.5343,
+      "step": 20597
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.610785164230587e-05,
+      "loss": 0.3886,
+      "step": 20598
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6099388871964524e-05,
+      "loss": 0.5615,
+      "step": 20599
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.609092687503357e-05,
+      "loss": 0.446,
+      "step": 20600
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 0.43681588768959045,
+      "eval_runtime": 853.819,
+      "eval_samples_per_second": 28.179,
+      "eval_steps_per_second": 14.09,
+      "step": 20600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.60824656516154e-05,
+      "loss": 0.4192,
+      "step": 20601
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.607400520181251e-05,
+      "loss": 0.436,
+      "step": 20602
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6065545525727254e-05,
+      "loss": 0.3904,
+      "step": 20603
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.605708662346201e-05,
+      "loss": 0.4626,
+      "step": 20604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6048628495119166e-05,
+      "loss": 0.425,
+      "step": 20605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.604017114080107e-05,
+      "loss": 0.4439,
+      "step": 20606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.603171456061014e-05,
+      "loss": 0.5274,
+      "step": 20607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.602325875464869e-05,
+      "loss": 0.4991,
+      "step": 20608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6014803723019085e-05,
+      "loss": 0.4515,
+      "step": 20609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6006349465823643e-05,
+      "loss": 0.4281,
+      "step": 20610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5997895983164656e-05,
+      "loss": 0.5294,
+      "step": 20611
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.598944327514452e-05,
+      "loss": 0.4815,
+      "step": 20612
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5980991341865474e-05,
+      "loss": 0.4878,
+      "step": 20613
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.597254018342986e-05,
+      "loss": 0.5533,
+      "step": 20614
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.596408979993993e-05,
+      "loss": 0.4254,
+      "step": 20615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5955640191497976e-05,
+      "loss": 0.4778,
+      "step": 20616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.594719135820623e-05,
+      "loss": 0.4761,
+      "step": 20617
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.593874330016702e-05,
+      "loss": 0.4314,
+      "step": 20618
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5930296017482556e-05,
+      "loss": 0.4616,
+      "step": 20619
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.592184951025509e-05,
+      "loss": 0.5037,
+      "step": 20620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.591340377858685e-05,
+      "loss": 0.4138,
+      "step": 20621
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5904958822580015e-05,
+      "loss": 0.3442,
+      "step": 20622
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.589651464233687e-05,
+      "loss": 0.5204,
+      "step": 20623
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5888071237959575e-05,
+      "loss": 0.5668,
+      "step": 20624
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5879628609550345e-05,
+      "loss": 0.4337,
+      "step": 20625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.587118675721135e-05,
+      "loss": 0.4712,
+      "step": 20626
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.586274568104472e-05,
+      "loss": 0.4462,
+      "step": 20627
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.585430538115271e-05,
+      "loss": 0.4523,
+      "step": 20628
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5845865857637426e-05,
+      "loss": 0.3602,
+      "step": 20629
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.583742711060102e-05,
+      "loss": 0.4256,
+      "step": 20630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.582898914014563e-05,
+      "loss": 0.4393,
+      "step": 20631
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.582055194637335e-05,
+      "loss": 0.4162,
+      "step": 20632
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.581211552938637e-05,
+      "loss": 0.4114,
+      "step": 20633
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5803679889286745e-05,
+      "loss": 0.3774,
+      "step": 20634
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.579524502617659e-05,
+      "loss": 0.4284,
+      "step": 20635
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.578681094015801e-05,
+      "loss": 0.5194,
+      "step": 20636
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5778377631333026e-05,
+      "loss": 0.516,
+      "step": 20637
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.576994509980378e-05,
+      "loss": 0.4547,
+      "step": 20638
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.576151334567232e-05,
+      "loss": 0.3298,
+      "step": 20639
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.575308236904067e-05,
+      "loss": 0.4123,
+      "step": 20640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.574465217001088e-05,
+      "loss": 0.4462,
+      "step": 20641
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.573622274868498e-05,
+      "loss": 0.4093,
+      "step": 20642
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5727794105165026e-05,
+      "loss": 0.4243,
+      "step": 20643
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.571936623955301e-05,
+      "loss": 0.4897,
+      "step": 20644
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.571093915195094e-05,
+      "loss": 0.4581,
+      "step": 20645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5702512842460825e-05,
+      "loss": 0.4922,
+      "step": 20646
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.569408731118462e-05,
+      "loss": 0.5082,
+      "step": 20647
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.568566255822429e-05,
+      "loss": 0.4468,
+      "step": 20648
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.567723858368186e-05,
+      "loss": 0.4738,
+      "step": 20649
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.566881538765927e-05,
+      "loss": 0.4509,
+      "step": 20650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5660392970258446e-05,
+      "loss": 0.3886,
+      "step": 20651
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.565197133158136e-05,
+      "loss": 0.5096,
+      "step": 20652
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.564355047172987e-05,
+      "loss": 0.5043,
+      "step": 20653
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.563513039080598e-05,
+      "loss": 0.4631,
+      "step": 20654
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.562671108891158e-05,
+      "loss": 0.4151,
+      "step": 20655
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.561829256614856e-05,
+      "loss": 0.5025,
+      "step": 20656
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5609874822618817e-05,
+      "loss": 0.3864,
+      "step": 20657
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.560145785842419e-05,
+      "loss": 0.5367,
+      "step": 20658
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.559304167366663e-05,
+      "loss": 0.3655,
+      "step": 20659
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.558462626844795e-05,
+      "loss": 0.4874,
+      "step": 20660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5576211642870024e-05,
+      "loss": 0.4283,
+      "step": 20661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.556779779703469e-05,
+      "loss": 0.4477,
+      "step": 20662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5559384731043766e-05,
+      "loss": 0.4333,
+      "step": 20663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.555097244499912e-05,
+      "loss": 0.4963,
+      "step": 20664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5542560939002536e-05,
+      "loss": 0.3508,
+      "step": 20665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5534150213155834e-05,
+      "loss": 0.3531,
+      "step": 20666
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5525740267560815e-05,
+      "loss": 0.4721,
+      "step": 20667
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5517331102319226e-05,
+      "loss": 0.4453,
+      "step": 20668
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5508922717532914e-05,
+      "loss": 0.4255,
+      "step": 20669
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.550051511330361e-05,
+      "loss": 0.4268,
+      "step": 20670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.549210828973308e-05,
+      "loss": 0.4543,
+      "step": 20671
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.548370224692308e-05,
+      "loss": 0.4196,
+      "step": 20672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.54752969849753e-05,
+      "loss": 0.5116,
+      "step": 20673
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.546689250399156e-05,
+      "loss": 0.4354,
+      "step": 20674
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.545848880407353e-05,
+      "loss": 0.4403,
+      "step": 20675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5450085885322936e-05,
+      "loss": 0.5013,
+      "step": 20676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.544168374784148e-05,
+      "loss": 0.3517,
+      "step": 20677
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.543328239173081e-05,
+      "loss": 0.3467,
+      "step": 20678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.542488181709269e-05,
+      "loss": 0.4407,
+      "step": 20679
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5416482024028766e-05,
+      "loss": 0.4419,
+      "step": 20680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.540808301264069e-05,
+      "loss": 0.4508,
+      "step": 20681
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.539968478303012e-05,
+      "loss": 0.4392,
+      "step": 20682
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5391287335298716e-05,
+      "loss": 0.4801,
+      "step": 20683
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5382890669548065e-05,
+      "loss": 0.5424,
+      "step": 20684
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.537449478587986e-05,
+      "loss": 0.3725,
+      "step": 20685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.536609968439571e-05,
+      "loss": 0.4434,
+      "step": 20686
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5357705365197194e-05,
+      "loss": 0.5335,
+      "step": 20687
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.534931182838593e-05,
+      "loss": 0.4633,
+      "step": 20688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5340919074063474e-05,
+      "loss": 0.4415,
+      "step": 20689
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.533252710233146e-05,
+      "loss": 0.4778,
+      "step": 20690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5324135913291434e-05,
+      "loss": 0.4204,
+      "step": 20691
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.531574550704496e-05,
+      "loss": 0.42,
+      "step": 20692
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5307355883693594e-05,
+      "loss": 0.4223,
+      "step": 20693
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.529896704333884e-05,
+      "loss": 0.374,
+      "step": 20694
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.529057898608228e-05,
+      "loss": 0.4654,
+      "step": 20695
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5282191712025424e-05,
+      "loss": 0.4124,
+      "step": 20696
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.527380522126983e-05,
+      "loss": 0.4531,
+      "step": 20697
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.526541951391687e-05,
+      "loss": 0.4443,
+      "step": 20698
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.525703459006816e-05,
+      "loss": 0.475,
+      "step": 20699
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.524865044982516e-05,
+      "loss": 0.4648,
+      "step": 20700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.524026709328933e-05,
+      "loss": 0.4806,
+      "step": 20701
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.523188452056213e-05,
+      "loss": 0.4104,
+      "step": 20702
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5223502731745005e-05,
+      "loss": 0.4425,
+      "step": 20703
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.521512172693946e-05,
+      "loss": 0.4795,
+      "step": 20704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.52067415062469e-05,
+      "loss": 0.4859,
+      "step": 20705
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5198362069768756e-05,
+      "loss": 0.4688,
+      "step": 20706
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5189983417606434e-05,
+      "loss": 0.4527,
+      "step": 20707
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.518160554986133e-05,
+      "loss": 0.43,
+      "step": 20708
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.517322846663489e-05,
+      "loss": 0.4309,
+      "step": 20709
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.516485216802848e-05,
+      "loss": 0.4286,
+      "step": 20710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5156476654143497e-05,
+      "loss": 0.5014,
+      "step": 20711
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.51481019250813e-05,
+      "loss": 0.5005,
+      "step": 20712
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5139727980943204e-05,
+      "loss": 0.476,
+      "step": 20713
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5131354821830656e-05,
+      "loss": 0.4481,
+      "step": 20714
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.512298244784494e-05,
+      "loss": 0.4067,
+      "step": 20715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.51146108590874e-05,
+      "loss": 0.4591,
+      "step": 20716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.510624005565937e-05,
+      "loss": 0.4383,
+      "step": 20717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5097870037662154e-05,
+      "loss": 0.5485,
+      "step": 20718
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.508950080519702e-05,
+      "loss": 0.4517,
+      "step": 20719
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5081132358365335e-05,
+      "loss": 0.3603,
+      "step": 20720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5072764697268356e-05,
+      "loss": 0.476,
+      "step": 20721
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.506439782200735e-05,
+      "loss": 0.3979,
+      "step": 20722
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.505603173268359e-05,
+      "loss": 0.4717,
+      "step": 20723
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5047666429398305e-05,
+      "loss": 0.4406,
+      "step": 20724
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.503930191225281e-05,
+      "loss": 0.4303,
+      "step": 20725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.503093818134829e-05,
+      "loss": 0.3634,
+      "step": 20726
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5022575236786005e-05,
+      "loss": 0.5144,
+      "step": 20727
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.501421307866716e-05,
+      "loss": 0.4596,
+      "step": 20728
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5005851707092915e-05,
+      "loss": 0.4366,
+      "step": 20729
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.499749112216456e-05,
+      "loss": 0.4031,
+      "step": 20730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.498913132398325e-05,
+      "loss": 0.5333,
+      "step": 20731
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4980772312650146e-05,
+      "loss": 0.4868,
+      "step": 20732
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4972414088266446e-05,
+      "loss": 0.3795,
+      "step": 20733
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.496405665093327e-05,
+      "loss": 0.4664,
+      "step": 20734
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4955700000751835e-05,
+      "loss": 0.4286,
+      "step": 20735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.494734413782325e-05,
+      "loss": 0.4401,
+      "step": 20736
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.493898906224865e-05,
+      "loss": 0.357,
+      "step": 20737
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.493063477412917e-05,
+      "loss": 0.4284,
+      "step": 20738
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.492228127356586e-05,
+      "loss": 0.4943,
+      "step": 20739
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4913928560659925e-05,
+      "loss": 0.4764,
+      "step": 20740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.490557663551243e-05,
+      "loss": 0.4554,
+      "step": 20741
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.489722549822443e-05,
+      "loss": 0.4166,
+      "step": 20742
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.488887514889702e-05,
+      "loss": 0.4063,
+      "step": 20743
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.488052558763123e-05,
+      "loss": 0.4073,
+      "step": 20744
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.487217681452819e-05,
+      "loss": 0.4514,
+      "step": 20745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.486382882968892e-05,
+      "loss": 0.4008,
+      "step": 20746
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.485548163321443e-05,
+      "loss": 0.4812,
+      "step": 20747
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4847135225205775e-05,
+      "loss": 0.3269,
+      "step": 20748
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.483878960576398e-05,
+      "loss": 0.4429,
+      "step": 20749
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.483044477498999e-05,
+      "loss": 0.5142,
+      "step": 20750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.482210073298491e-05,
+      "loss": 0.4503,
+      "step": 20751
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4813757479849665e-05,
+      "loss": 0.4367,
+      "step": 20752
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.480541501568525e-05,
+      "loss": 0.4182,
+      "step": 20753
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4797073340592645e-05,
+      "loss": 0.3905,
+      "step": 20754
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.478873245467277e-05,
+      "loss": 0.4061,
+      "step": 20755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.478039235802664e-05,
+      "loss": 0.517,
+      "step": 20756
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.477205305075516e-05,
+      "loss": 0.4066,
+      "step": 20757
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.476371453295928e-05,
+      "loss": 0.405,
+      "step": 20758
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.475537680473991e-05,
+      "loss": 0.479,
+      "step": 20759
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.474703986619795e-05,
+      "loss": 0.4097,
+      "step": 20760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.473870371743435e-05,
+      "loss": 0.4799,
+      "step": 20761
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.473036835854998e-05,
+      "loss": 0.4201,
+      "step": 20762
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.472203378964573e-05,
+      "loss": 0.4535,
+      "step": 20763
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.471370001082247e-05,
+      "loss": 0.4355,
+      "step": 20764
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.470536702218103e-05,
+      "loss": 0.3925,
+      "step": 20765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.469703482382235e-05,
+      "loss": 0.4278,
+      "step": 20766
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.468870341584722e-05,
+      "loss": 0.3824,
+      "step": 20767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4680372798356496e-05,
+      "loss": 0.4623,
+      "step": 20768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4672042971451e-05,
+      "loss": 0.4418,
+      "step": 20769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.466371393523152e-05,
+      "loss": 0.4269,
+      "step": 20770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.465538568979894e-05,
+      "loss": 0.3858,
+      "step": 20771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4647058235254e-05,
+      "loss": 0.5175,
+      "step": 20772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.463873157169751e-05,
+      "loss": 0.4751,
+      "step": 20773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.463040569923025e-05,
+      "loss": 0.4717,
+      "step": 20774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4622080617952955e-05,
+      "loss": 0.4483,
+      "step": 20775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.461375632796645e-05,
+      "loss": 0.4412,
+      "step": 20776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4605432829371456e-05,
+      "loss": 0.5003,
+      "step": 20777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.459711012226872e-05,
+      "loss": 0.6107,
+      "step": 20778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.458878820675896e-05,
+      "loss": 0.4274,
+      "step": 20779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.458046708294287e-05,
+      "loss": 0.5554,
+      "step": 20780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.457214675092124e-05,
+      "loss": 0.4592,
+      "step": 20781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4563827210794743e-05,
+      "loss": 0.5347,
+      "step": 20782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.455550846266405e-05,
+      "loss": 0.4645,
+      "step": 20783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.454719050662987e-05,
+      "loss": 0.4227,
+      "step": 20784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.453887334279287e-05,
+      "loss": 0.4239,
+      "step": 20785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.453055697125367e-05,
+      "loss": 0.4946,
+      "step": 20786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.452224139211301e-05,
+      "loss": 0.4414,
+      "step": 20787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.45139266054715e-05,
+      "loss": 0.4688,
+      "step": 20788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4505612611429774e-05,
+      "loss": 0.3689,
+      "step": 20789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.449729941008844e-05,
+      "loss": 0.4976,
+      "step": 20790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4488987001548125e-05,
+      "loss": 0.4262,
+      "step": 20791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.448067538590945e-05,
+      "loss": 0.385,
+      "step": 20792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.447236456327303e-05,
+      "loss": 0.484,
+      "step": 20793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.446405453373943e-05,
+      "loss": 0.3907,
+      "step": 20794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4455745297409225e-05,
+      "loss": 0.5672,
+      "step": 20795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4447436854382954e-05,
+      "loss": 0.4905,
+      "step": 20796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4439129204761256e-05,
+      "loss": 0.4336,
+      "step": 20797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.443082234864463e-05,
+      "loss": 0.4392,
+      "step": 20798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.442251628613363e-05,
+      "loss": 0.4069,
+      "step": 20799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.441421101732878e-05,
+      "loss": 0.4381,
+      "step": 20800
+    },
+    {
+      "epoch": 0.73,
+      "eval_loss": 0.43474748730659485,
+      "eval_runtime": 854.1051,
+      "eval_samples_per_second": 28.17,
+      "eval_steps_per_second": 14.085,
+      "step": 20800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.440590654233057e-05,
+      "loss": 0.4407,
+      "step": 20801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.439760286123958e-05,
+      "loss": 0.4448,
+      "step": 20802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4389299974156286e-05,
+      "loss": 0.3829,
+      "step": 20803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4380997881181155e-05,
+      "loss": 0.4144,
+      "step": 20804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.43726965824147e-05,
+      "loss": 0.4543,
+      "step": 20805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.436439607795735e-05,
+      "loss": 0.4883,
+      "step": 20806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.435609636790962e-05,
+      "loss": 0.4314,
+      "step": 20807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4347797452371954e-05,
+      "loss": 0.5156,
+      "step": 20808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.433949933144479e-05,
+      "loss": 0.4838,
+      "step": 20809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.433120200522856e-05,
+      "loss": 0.4802,
+      "step": 20810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4322905473823654e-05,
+      "loss": 0.4485,
+      "step": 20811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.431460973733055e-05,
+      "loss": 0.4689,
+      "step": 20812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4306314795849625e-05,
+      "loss": 0.4696,
+      "step": 20813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4298020649481287e-05,
+      "loss": 0.4577,
+      "step": 20814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.428972729832591e-05,
+      "loss": 0.4568,
+      "step": 20815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4281434742483874e-05,
+      "loss": 0.4207,
+      "step": 20816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.427314298205552e-05,
+      "loss": 0.4171,
+      "step": 20817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4264852017141257e-05,
+      "loss": 0.4274,
+      "step": 20818
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4256561847841416e-05,
+      "loss": 0.3851,
+      "step": 20819
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.424827247425633e-05,
+      "loss": 0.4746,
+      "step": 20820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4239983896486327e-05,
+      "loss": 0.3939,
+      "step": 20821
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.423169611463168e-05,
+      "loss": 0.4427,
+      "step": 20822
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4223409128792796e-05,
+      "loss": 0.4663,
+      "step": 20823
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.421512293906992e-05,
+      "loss": 0.4911,
+      "step": 20824
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4206837545563344e-05,
+      "loss": 0.3873,
+      "step": 20825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4198552948373355e-05,
+      "loss": 0.4978,
+      "step": 20826
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4190269147600186e-05,
+      "loss": 0.5614,
+      "step": 20827
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4181986143344166e-05,
+      "loss": 0.5131,
+      "step": 20828
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4173703935705516e-05,
+      "loss": 0.5009,
+      "step": 20829
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.416542252478448e-05,
+      "loss": 0.3893,
+      "step": 20830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.415714191068128e-05,
+      "loss": 0.4276,
+      "step": 20831
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.414886209349615e-05,
+      "loss": 0.4063,
+      "step": 20832
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.41405830733293e-05,
+      "loss": 0.4292,
+      "step": 20833
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4132304850280935e-05,
+      "loss": 0.4354,
+      "step": 20834
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4124027424451254e-05,
+      "loss": 0.4419,
+      "step": 20835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4115750795940394e-05,
+      "loss": 0.5396,
+      "step": 20836
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4107474964848604e-05,
+      "loss": 0.4322,
+      "step": 20837
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4099199931276016e-05,
+      "loss": 0.5646,
+      "step": 20838
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.40909256953228e-05,
+      "loss": 0.5035,
+      "step": 20839
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.408265225708906e-05,
+      "loss": 0.3796,
+      "step": 20840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4074379616674954e-05,
+      "loss": 0.4408,
+      "step": 20841
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4066107774180634e-05,
+      "loss": 0.4706,
+      "step": 20842
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.40578367297062e-05,
+      "loss": 0.4396,
+      "step": 20843
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4049566483351756e-05,
+      "loss": 0.4702,
+      "step": 20844
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.40412970352174e-05,
+      "loss": 0.4065,
+      "step": 20845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.403302838540319e-05,
+      "loss": 0.4483,
+      "step": 20846
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4024760534009256e-05,
+      "loss": 0.4732,
+      "step": 20847
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4016493481135656e-05,
+      "loss": 0.4331,
+      "step": 20848
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.400822722688244e-05,
+      "loss": 0.4156,
+      "step": 20849
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.399996177134965e-05,
+      "loss": 0.5003,
+      "step": 20850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.399169711463733e-05,
+      "loss": 0.4118,
+      "step": 20851
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3983433256845476e-05,
+      "loss": 0.457,
+      "step": 20852
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.397517019807418e-05,
+      "loss": 0.4382,
+      "step": 20853
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.396690793842341e-05,
+      "loss": 0.3849,
+      "step": 20854
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.395864647799317e-05,
+      "loss": 0.4867,
+      "step": 20855
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.395038581688346e-05,
+      "loss": 0.4353,
+      "step": 20856
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.394212595519421e-05,
+      "loss": 0.4042,
+      "step": 20857
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.393386689302549e-05,
+      "loss": 0.457,
+      "step": 20858
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.392560863047719e-05,
+      "loss": 0.4912,
+      "step": 20859
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.391735116764929e-05,
+      "loss": 0.4541,
+      "step": 20860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3909094504641726e-05,
+      "loss": 0.4036,
+      "step": 20861
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.390083864155439e-05,
+      "loss": 0.4508,
+      "step": 20862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.389258357848728e-05,
+      "loss": 0.4432,
+      "step": 20863
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3884329315540274e-05,
+      "loss": 0.3935,
+      "step": 20864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.387607585281328e-05,
+      "loss": 0.4818,
+      "step": 20865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.386782319040619e-05,
+      "loss": 0.5009,
+      "step": 20866
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.385957132841885e-05,
+      "loss": 0.4811,
+      "step": 20867
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.38513202669512e-05,
+      "loss": 0.464,
+      "step": 20868
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.384307000610308e-05,
+      "loss": 0.5256,
+      "step": 20869
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3834820545974356e-05,
+      "loss": 0.4401,
+      "step": 20870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3826571886664846e-05,
+      "loss": 0.4992,
+      "step": 20871
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.381832402827439e-05,
+      "loss": 0.4445,
+      "step": 20872
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.381007697090284e-05,
+      "loss": 0.4586,
+      "step": 20873
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3801830714650006e-05,
+      "loss": 0.4216,
+      "step": 20874
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3793585259615687e-05,
+      "loss": 0.4231,
+      "step": 20875
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.378534060589968e-05,
+      "loss": 0.431,
+      "step": 20876
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.377709675360176e-05,
+      "loss": 0.4894,
+      "step": 20877
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.376885370282173e-05,
+      "loss": 0.4237,
+      "step": 20878
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.376061145365936e-05,
+      "loss": 0.3895,
+      "step": 20879
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.375237000621439e-05,
+      "loss": 0.4409,
+      "step": 20880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.374412936058659e-05,
+      "loss": 0.5148,
+      "step": 20881
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.373588951687567e-05,
+      "loss": 0.5077,
+      "step": 20882
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.372765047518135e-05,
+      "loss": 0.493,
+      "step": 20883
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.37194122356034e-05,
+      "loss": 0.4554,
+      "step": 20884
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3711174798241504e-05,
+      "loss": 0.4573,
+      "step": 20885
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3702938163195374e-05,
+      "loss": 0.3865,
+      "step": 20886
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3694702330564684e-05,
+      "loss": 0.4732,
+      "step": 20887
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.36864673004491e-05,
+      "loss": 0.5643,
+      "step": 20888
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.367823307294834e-05,
+      "loss": 0.4331,
+      "step": 20889
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.366999964816203e-05,
+      "loss": 0.4027,
+      "step": 20890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.366176702618984e-05,
+      "loss": 0.4577,
+      "step": 20891
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3653535207131405e-05,
+      "loss": 0.5977,
+      "step": 20892
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.364530419108632e-05,
+      "loss": 0.4713,
+      "step": 20893
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.363707397815429e-05,
+      "loss": 0.4648,
+      "step": 20894
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.362884456843487e-05,
+      "loss": 0.415,
+      "step": 20895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.362061596202768e-05,
+      "loss": 0.4657,
+      "step": 20896
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3612388159032303e-05,
+      "loss": 0.4241,
+      "step": 20897
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.36041611595483e-05,
+      "loss": 0.4312,
+      "step": 20898
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.35959349636753e-05,
+      "loss": 0.4234,
+      "step": 20899
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.358770957151286e-05,
+      "loss": 0.4217,
+      "step": 20900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.35794849831605e-05,
+      "loss": 0.4479,
+      "step": 20901
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.357126119871779e-05,
+      "loss": 0.4925,
+      "step": 20902
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.356303821828421e-05,
+      "loss": 0.431,
+      "step": 20903
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.355481604195938e-05,
+      "loss": 0.5026,
+      "step": 20904
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.354659466984277e-05,
+      "loss": 0.4553,
+      "step": 20905
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3538374102033866e-05,
+      "loss": 0.4564,
+      "step": 20906
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.353015433863219e-05,
+      "loss": 0.5034,
+      "step": 20907
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3521935379737194e-05,
+      "loss": 0.4214,
+      "step": 20908
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.351371722544841e-05,
+      "loss": 0.3866,
+      "step": 20909
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3505499875865275e-05,
+      "loss": 0.4566,
+      "step": 20910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.349728333108725e-05,
+      "loss": 0.3816,
+      "step": 20911
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.348906759121377e-05,
+      "loss": 0.4915,
+      "step": 20912
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3480852656344264e-05,
+      "loss": 0.4091,
+      "step": 20913
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3472638526578204e-05,
+      "loss": 0.4266,
+      "step": 20914
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.346442520201499e-05,
+      "loss": 0.452,
+      "step": 20915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.345621268275402e-05,
+      "loss": 0.4354,
+      "step": 20916
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.34480009688947e-05,
+      "loss": 0.4092,
+      "step": 20917
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.343979006053641e-05,
+      "loss": 0.4241,
+      "step": 20918
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.34315799577785e-05,
+      "loss": 0.4293,
+      "step": 20919
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.342337066072042e-05,
+      "loss": 0.4591,
+      "step": 20920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.341516216946148e-05,
+      "loss": 0.4918,
+      "step": 20921
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.340695448410103e-05,
+      "loss": 0.4726,
+      "step": 20922
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.339874760473841e-05,
+      "loss": 0.4598,
+      "step": 20923
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.339054153147294e-05,
+      "loss": 0.4447,
+      "step": 20924
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3382336264403976e-05,
+      "loss": 0.4334,
+      "step": 20925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.337413180363082e-05,
+      "loss": 0.4363,
+      "step": 20926
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.336592814925276e-05,
+      "loss": 0.4128,
+      "step": 20927
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.335772530136908e-05,
+      "loss": 0.5936,
+      "step": 20928
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3349523260079053e-05,
+      "loss": 0.3982,
+      "step": 20929
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3341322025482e-05,
+      "loss": 0.4184,
+      "step": 20930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.333312159767715e-05,
+      "loss": 0.4207,
+      "step": 20931
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.332492197676377e-05,
+      "loss": 0.4786,
+      "step": 20932
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3316723162841076e-05,
+      "loss": 0.4512,
+      "step": 20933
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.33085251560083e-05,
+      "loss": 0.4611,
+      "step": 20934
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3300327956364705e-05,
+      "loss": 0.4788,
+      "step": 20935
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.329213156400949e-05,
+      "loss": 0.538,
+      "step": 20936
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.328393597904184e-05,
+      "loss": 0.3592,
+      "step": 20937
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.327574120156096e-05,
+      "loss": 0.4682,
+      "step": 20938
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3267547231666006e-05,
+      "loss": 0.4374,
+      "step": 20939
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.325935406945622e-05,
+      "loss": 0.508,
+      "step": 20940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.325116171503071e-05,
+      "loss": 0.406,
+      "step": 20941
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3242970168488654e-05,
+      "loss": 0.5155,
+      "step": 20942
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3234779429929196e-05,
+      "loss": 0.4107,
+      "step": 20943
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.322658949945142e-05,
+      "loss": 0.4061,
+      "step": 20944
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3218400377154546e-05,
+      "loss": 0.4088,
+      "step": 20945
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.321021206313762e-05,
+      "loss": 0.4122,
+      "step": 20946
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.320202455749979e-05,
+      "loss": 0.4409,
+      "step": 20947
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3193837860340114e-05,
+      "loss": 0.4122,
+      "step": 20948
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3185651971757704e-05,
+      "loss": 0.4099,
+      "step": 20949
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3177466891851594e-05,
+      "loss": 0.4268,
+      "step": 20950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.316928262072092e-05,
+      "loss": 0.422,
+      "step": 20951
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.316109915846469e-05,
+      "loss": 0.4241,
+      "step": 20952
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.315291650518197e-05,
+      "loss": 0.5145,
+      "step": 20953
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.31447346609718e-05,
+      "loss": 0.4532,
+      "step": 20954
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3136553625933166e-05,
+      "loss": 0.4529,
+      "step": 20955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.312837340016515e-05,
+      "loss": 0.4685,
+      "step": 20956
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.312019398376672e-05,
+      "loss": 0.3984,
+      "step": 20957
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.311201537683689e-05,
+      "loss": 0.4931,
+      "step": 20958
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.310383757947464e-05,
+      "loss": 0.4129,
+      "step": 20959
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.309566059177892e-05,
+      "loss": 0.4554,
+      "step": 20960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.308748441384876e-05,
+      "loss": 0.4383,
+      "step": 20961
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.307930904578309e-05,
+      "loss": 0.379,
+      "step": 20962
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3071134487680855e-05,
+      "loss": 0.4439,
+      "step": 20963
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.306296073964099e-05,
+      "loss": 0.3338,
+      "step": 20964
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.305478780176241e-05,
+      "loss": 0.3655,
+      "step": 20965
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.304661567414412e-05,
+      "loss": 0.535,
+      "step": 20966
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.303844435688494e-05,
+      "loss": 0.4307,
+      "step": 20967
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3030273850083784e-05,
+      "loss": 0.4326,
+      "step": 20968
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3022104153839514e-05,
+      "loss": 0.3885,
+      "step": 20969
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.301393526825111e-05,
+      "loss": 0.4765,
+      "step": 20970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3005767193417356e-05,
+      "loss": 0.413,
+      "step": 20971
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.299759992943716e-05,
+      "loss": 0.5206,
+      "step": 20972
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.298943347640935e-05,
+      "loss": 0.4158,
+      "step": 20973
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.298126783443273e-05,
+      "loss": 0.4742,
+      "step": 20974
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.297310300360622e-05,
+      "loss": 0.4038,
+      "step": 20975
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2964938984028584e-05,
+      "loss": 0.4365,
+      "step": 20976
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.295677577579864e-05,
+      "loss": 0.4568,
+      "step": 20977
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2948613379015206e-05,
+      "loss": 0.4243,
+      "step": 20978
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2940451793777025e-05,
+      "loss": 0.4849,
+      "step": 20979
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.293229102018295e-05,
+      "loss": 0.3738,
+      "step": 20980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.292413105833172e-05,
+      "loss": 0.4481,
+      "step": 20981
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.29159719083221e-05,
+      "loss": 0.4218,
+      "step": 20982
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.290781357025284e-05,
+      "loss": 0.4039,
+      "step": 20983
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.289965604422268e-05,
+      "loss": 0.4799,
+      "step": 20984
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.289149933033033e-05,
+      "loss": 0.5363,
+      "step": 20985
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2883343428674574e-05,
+      "loss": 0.5013,
+      "step": 20986
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.287518833935409e-05,
+      "loss": 0.3719,
+      "step": 20987
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2867034062467584e-05,
+      "loss": 0.5244,
+      "step": 20988
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.285888059811375e-05,
+      "loss": 0.3699,
+      "step": 20989
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.285072794639125e-05,
+      "loss": 0.3846,
+      "step": 20990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2842576107398794e-05,
+      "loss": 0.4369,
+      "step": 20991
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.283442508123504e-05,
+      "loss": 0.4071,
+      "step": 20992
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.282627486799864e-05,
+      "loss": 0.4404,
+      "step": 20993
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.281812546778824e-05,
+      "loss": 0.4263,
+      "step": 20994
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.280997688070243e-05,
+      "loss": 0.4809,
+      "step": 20995
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.280182910683991e-05,
+      "loss": 0.4422,
+      "step": 20996
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2793682146299254e-05,
+      "loss": 0.4633,
+      "step": 20997
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2785535999179074e-05,
+      "loss": 0.433,
+      "step": 20998
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2777390665577966e-05,
+      "loss": 0.4198,
+      "step": 20999
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.276924614559448e-05,
+      "loss": 0.4849,
+      "step": 21000
+    },
+    {
+      "epoch": 0.73,
+      "eval_loss": 0.432409405708313,
+      "eval_runtime": 845.3607,
+      "eval_samples_per_second": 28.461,
+      "eval_steps_per_second": 14.231,
+      "step": 21000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2761102439327264e-05,
+      "loss": 0.496,
+      "step": 21001
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.275295954687484e-05,
+      "loss": 0.4612,
+      "step": 21002
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.274481746833578e-05,
+      "loss": 0.4924,
+      "step": 21003
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2736676203808604e-05,
+      "loss": 0.4218,
+      "step": 21004
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.272853575339184e-05,
+      "loss": 0.4785,
+      "step": 21005
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.272039611718407e-05,
+      "loss": 0.4439,
+      "step": 21006
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.271225729528379e-05,
+      "loss": 0.4132,
+      "step": 21007
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.270411928778948e-05,
+      "loss": 0.4515,
+      "step": 21008
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2695982094799674e-05,
+      "loss": 0.5411,
+      "step": 21009
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.268784571641279e-05,
+      "loss": 0.4215,
+      "step": 21010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.267971015272739e-05,
+      "loss": 0.4776,
+      "step": 21011
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2671575403841917e-05,
+      "loss": 0.4056,
+      "step": 21012
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.26634414698548e-05,
+      "loss": 0.4091,
+      "step": 21013
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.265530835086452e-05,
+      "loss": 0.4974,
+      "step": 21014
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2647176046969463e-05,
+      "loss": 0.4127,
+      "step": 21015
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.263904455826813e-05,
+      "loss": 0.4478,
+      "step": 21016
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.263091388485889e-05,
+      "loss": 0.3712,
+      "step": 21017
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.262278402684017e-05,
+      "loss": 0.3942,
+      "step": 21018
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2614654984310356e-05,
+      "loss": 0.4702,
+      "step": 21019
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.260652675736785e-05,
+      "loss": 0.4769,
+      "step": 21020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2598399346110984e-05,
+      "loss": 0.4327,
+      "step": 21021
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.259027275063821e-05,
+      "loss": 0.4241,
+      "step": 21022
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2582146971047834e-05,
+      "loss": 0.5115,
+      "step": 21023
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.257402200743821e-05,
+      "loss": 0.4515,
+      "step": 21024
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.256589785990769e-05,
+      "loss": 0.4375,
+      "step": 21025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2557774528554553e-05,
+      "loss": 0.5452,
+      "step": 21026
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.254965201347721e-05,
+      "loss": 0.4297,
+      "step": 21027
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.254153031477389e-05,
+      "loss": 0.481,
+      "step": 21028
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.253340943254295e-05,
+      "loss": 0.4452,
+      "step": 21029
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.252528936688264e-05,
+      "loss": 0.4107,
+      "step": 21030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.251717011789122e-05,
+      "loss": 0.3839,
+      "step": 21031
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.250905168566703e-05,
+      "loss": 0.4922,
+      "step": 21032
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2500934070308274e-05,
+      "loss": 0.4197,
+      "step": 21033
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2492817271913244e-05,
+      "loss": 0.417,
+      "step": 21034
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.248470129058014e-05,
+      "loss": 0.4445,
+      "step": 21035
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.247658612640718e-05,
+      "loss": 0.3859,
+      "step": 21036
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.246847177949266e-05,
+      "loss": 0.4592,
+      "step": 21037
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2460358249934734e-05,
+      "loss": 0.4927,
+      "step": 21038
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.245224553783162e-05,
+      "loss": 0.4821,
+      "step": 21039
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2444133643281496e-05,
+      "loss": 0.5006,
+      "step": 21040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2436022566382505e-05,
+      "loss": 0.4191,
+      "step": 21041
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.242791230723291e-05,
+      "loss": 0.3697,
+      "step": 21042
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.241980286593082e-05,
+      "loss": 0.4419,
+      "step": 21043
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.241169424257438e-05,
+      "loss": 0.4376,
+      "step": 21044
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.240358643726177e-05,
+      "loss": 0.4351,
+      "step": 21045
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2395479450091036e-05,
+      "loss": 0.3984,
+      "step": 21046
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.23873732811604e-05,
+      "loss": 0.46,
+      "step": 21047
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2379267930567926e-05,
+      "loss": 0.398,
+      "step": 21048
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2371163398411716e-05,
+      "loss": 0.4842,
+      "step": 21049
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2363059684789874e-05,
+      "loss": 0.365,
+      "step": 21050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.235495678980047e-05,
+      "loss": 0.4448,
+      "step": 21051
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.234685471354155e-05,
+      "loss": 0.3978,
+      "step": 21052
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.233875345611125e-05,
+      "loss": 0.5421,
+      "step": 21053
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.233065301760756e-05,
+      "loss": 0.4438,
+      "step": 21054
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.232255339812855e-05,
+      "loss": 0.4279,
+      "step": 21055
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.231445459777225e-05,
+      "loss": 0.4835,
+      "step": 21056
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2306356616636624e-05,
+      "loss": 0.4311,
+      "step": 21057
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.229825945481978e-05,
+      "loss": 0.3927,
+      "step": 21058
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.229016311241968e-05,
+      "loss": 0.3913,
+      "step": 21059
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.228206758953432e-05,
+      "loss": 0.4068,
+      "step": 21060
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.227397288626166e-05,
+      "loss": 0.5036,
+      "step": 21061
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2265879002699665e-05,
+      "loss": 0.4051,
+      "step": 21062
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.225778593894635e-05,
+      "loss": 0.4921,
+      "step": 21063
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.224969369509965e-05,
+      "loss": 0.4706,
+      "step": 21064
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.224160227125749e-05,
+      "loss": 0.4745,
+      "step": 21065
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.223351166751781e-05,
+      "loss": 0.4768,
+      "step": 21066
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.22254218839785e-05,
+      "loss": 0.4655,
+      "step": 21067
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2217332920737555e-05,
+      "loss": 0.421,
+      "step": 21068
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.220924477789282e-05,
+      "loss": 0.4284,
+      "step": 21069
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.220115745554221e-05,
+      "loss": 0.3563,
+      "step": 21070
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.219307095378359e-05,
+      "loss": 0.4068,
+      "step": 21071
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.218498527271482e-05,
+      "loss": 0.4219,
+      "step": 21072
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.217690041243381e-05,
+      "loss": 0.4191,
+      "step": 21073
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.216881637303839e-05,
+      "loss": 0.5091,
+      "step": 21074
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.216073315462641e-05,
+      "loss": 0.4581,
+      "step": 21075
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2152650757295686e-05,
+      "loss": 0.4518,
+      "step": 21076
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.214456918114403e-05,
+      "loss": 0.4904,
+      "step": 21077
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.213648842626932e-05,
+      "loss": 0.43,
+      "step": 21078
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.21284084927693e-05,
+      "loss": 0.3693,
+      "step": 21079
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2120329380741795e-05,
+      "loss": 0.4232,
+      "step": 21080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2112251090284585e-05,
+      "loss": 0.4946,
+      "step": 21081
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.210417362149539e-05,
+      "loss": 0.5209,
+      "step": 21082
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.209609697447206e-05,
+      "loss": 0.4078,
+      "step": 21083
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.208802114931231e-05,
+      "loss": 0.4562,
+      "step": 21084
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.207994614611389e-05,
+      "loss": 0.3895,
+      "step": 21085
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2071871964974534e-05,
+      "loss": 0.4949,
+      "step": 21086
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2063798605991945e-05,
+      "loss": 0.4006,
+      "step": 21087
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.205572606926384e-05,
+      "loss": 0.4162,
+      "step": 21088
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2047654354887956e-05,
+      "loss": 0.4788,
+      "step": 21089
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2039583462961976e-05,
+      "loss": 0.4865,
+      "step": 21090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.203151339358358e-05,
+      "loss": 0.485,
+      "step": 21091
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2023444146850444e-05,
+      "loss": 0.4344,
+      "step": 21092
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.20153757228602e-05,
+      "loss": 0.4331,
+      "step": 21093
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.200730812171055e-05,
+      "loss": 0.4098,
+      "step": 21094
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1999241343499144e-05,
+      "loss": 0.4271,
+      "step": 21095
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.199117538832358e-05,
+      "loss": 0.4472,
+      "step": 21096
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.19831102562815e-05,
+      "loss": 0.4703,
+      "step": 21097
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.197504594747048e-05,
+      "loss": 0.3963,
+      "step": 21098
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.196698246198821e-05,
+      "loss": 0.4278,
+      "step": 21099
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.195891979993222e-05,
+      "loss": 0.4258,
+      "step": 21100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.195085796140015e-05,
+      "loss": 0.4417,
+      "step": 21101
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.194279694648947e-05,
+      "loss": 0.4112,
+      "step": 21102
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.193473675529786e-05,
+      "loss": 0.3747,
+      "step": 21103
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.192667738792281e-05,
+      "loss": 0.4096,
+      "step": 21104
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.191861884446189e-05,
+      "loss": 0.4085,
+      "step": 21105
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.191056112501263e-05,
+      "loss": 0.5267,
+      "step": 21106
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1902504229672515e-05,
+      "loss": 0.4089,
+      "step": 21107
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1894448158539125e-05,
+      "loss": 0.4779,
+      "step": 21108
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.188639291170995e-05,
+      "loss": 0.4612,
+      "step": 21109
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.187833848928247e-05,
+      "loss": 0.3469,
+      "step": 21110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.187028489135417e-05,
+      "loss": 0.4418,
+      "step": 21111
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1862232118022503e-05,
+      "loss": 0.4237,
+      "step": 21112
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1854180169384996e-05,
+      "loss": 0.4854,
+      "step": 21113
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1846129045539065e-05,
+      "loss": 0.5103,
+      "step": 21114
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.183807874658217e-05,
+      "loss": 0.4165,
+      "step": 21115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.183002927261173e-05,
+      "loss": 0.4473,
+      "step": 21116
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.182198062372513e-05,
+      "loss": 0.3943,
+      "step": 21117
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1813932800019886e-05,
+      "loss": 0.3604,
+      "step": 21118
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1805885801593335e-05,
+      "loss": 0.4457,
+      "step": 21119
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1797839628542895e-05,
+      "loss": 0.4254,
+      "step": 21120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.178979428096595e-05,
+      "loss": 0.3971,
+      "step": 21121
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.178174975895985e-05,
+      "loss": 0.389,
+      "step": 21122
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1773706062621966e-05,
+      "loss": 0.4706,
+      "step": 21123
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1765663192049696e-05,
+      "loss": 0.4109,
+      "step": 21124
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.175762114734036e-05,
+      "loss": 0.4233,
+      "step": 21125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.174957992859128e-05,
+      "loss": 0.4229,
+      "step": 21126
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.17415395358998e-05,
+      "loss": 0.4382,
+      "step": 21127
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.173349996936319e-05,
+      "loss": 0.4952,
+      "step": 21128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1725461229078834e-05,
+      "loss": 0.5142,
+      "step": 21129
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.171742331514398e-05,
+      "loss": 0.4336,
+      "step": 21130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.170938622765593e-05,
+      "loss": 0.4267,
+      "step": 21131
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1701349966711935e-05,
+      "loss": 0.4341,
+      "step": 21132
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.169331453240925e-05,
+      "loss": 0.3785,
+      "step": 21133
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.168527992484519e-05,
+      "loss": 0.5059,
+      "step": 21134
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1677246144116965e-05,
+      "loss": 0.4252,
+      "step": 21135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.166921319032181e-05,
+      "loss": 0.4291,
+      "step": 21136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.166118106355696e-05,
+      "loss": 0.3873,
+      "step": 21137
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1653149763919585e-05,
+      "loss": 0.4518,
+      "step": 21138
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.164511929150696e-05,
+      "loss": 0.4113,
+      "step": 21139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.163708964641625e-05,
+      "loss": 0.4239,
+      "step": 21140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.162906082874464e-05,
+      "loss": 0.5081,
+      "step": 21141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1621032838589305e-05,
+      "loss": 0.463,
+      "step": 21142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.161300567604738e-05,
+      "loss": 0.3969,
+      "step": 21143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.160497934121609e-05,
+      "loss": 0.4525,
+      "step": 21144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.159695383419253e-05,
+      "loss": 0.5077,
+      "step": 21145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1588929155073855e-05,
+      "loss": 0.4469,
+      "step": 21146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.158090530395718e-05,
+      "loss": 0.4297,
+      "step": 21147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.157288228093957e-05,
+      "loss": 0.5128,
+      "step": 21148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1564860086118243e-05,
+      "loss": 0.4226,
+      "step": 21149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.155683871959022e-05,
+      "loss": 0.4626,
+      "step": 21150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.154881818145261e-05,
+      "loss": 0.4445,
+      "step": 21151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.154079847180247e-05,
+      "loss": 0.4462,
+      "step": 21152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1532779590736885e-05,
+      "loss": 0.4647,
+      "step": 21153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.152476153835286e-05,
+      "loss": 0.5188,
+      "step": 21154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1516744314747505e-05,
+      "loss": 0.423,
+      "step": 21155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.150872792001783e-05,
+      "loss": 0.4471,
+      "step": 21156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1500712354260865e-05,
+      "loss": 0.4343,
+      "step": 21157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.149269761757362e-05,
+      "loss": 0.4405,
+      "step": 21158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.148468371005306e-05,
+      "loss": 0.4374,
+      "step": 21159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.147667063179626e-05,
+      "loss": 0.4128,
+      "step": 21160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1468658382900164e-05,
+      "loss": 0.5019,
+      "step": 21161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.146064696346174e-05,
+      "loss": 0.3608,
+      "step": 21162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.145263637357797e-05,
+      "loss": 0.4692,
+      "step": 21163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1444626613345764e-05,
+      "loss": 0.4034,
+      "step": 21164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.143661768286212e-05,
+      "loss": 0.4002,
+      "step": 21165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.142860958222398e-05,
+      "loss": 0.4422,
+      "step": 21166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.142060231152822e-05,
+      "loss": 0.4092,
+      "step": 21167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1412595870871786e-05,
+      "loss": 0.4605,
+      "step": 21168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1404590260351555e-05,
+      "loss": 0.4072,
+      "step": 21169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.139658548006446e-05,
+      "loss": 0.5221,
+      "step": 21170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1388581530107365e-05,
+      "loss": 0.3598,
+      "step": 21171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.138057841057715e-05,
+      "loss": 0.353,
+      "step": 21172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1372576121570674e-05,
+      "loss": 0.4127,
+      "step": 21173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.136457466318477e-05,
+      "loss": 0.4725,
+      "step": 21174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.135657403551633e-05,
+      "loss": 0.3525,
+      "step": 21175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.134857423866217e-05,
+      "loss": 0.4913,
+      "step": 21176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.13405752727191e-05,
+      "loss": 0.3522,
+      "step": 21177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.133257713778394e-05,
+      "loss": 0.4166,
+      "step": 21178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.132457983395347e-05,
+      "loss": 0.4151,
+      "step": 21179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1316583361324534e-05,
+      "loss": 0.4885,
+      "step": 21180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.13085877199939e-05,
+      "loss": 0.4553,
+      "step": 21181
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.130059291005833e-05,
+      "loss": 0.5159,
+      "step": 21182
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.12925989316146e-05,
+      "loss": 0.4513,
+      "step": 21183
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.128460578475941e-05,
+      "loss": 0.5228,
+      "step": 21184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.12766134695896e-05,
+      "loss": 0.43,
+      "step": 21185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.126862198620184e-05,
+      "loss": 0.4318,
+      "step": 21186
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.126063133469287e-05,
+      "loss": 0.4423,
+      "step": 21187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.12526415151594e-05,
+      "loss": 0.4382,
+      "step": 21188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.124465252769813e-05,
+      "loss": 0.4271,
+      "step": 21189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.123666437240572e-05,
+      "loss": 0.4104,
+      "step": 21190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.122867704937893e-05,
+      "loss": 0.4127,
+      "step": 21191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1220690558714394e-05,
+      "loss": 0.4359,
+      "step": 21192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.121270490050877e-05,
+      "loss": 0.4333,
+      "step": 21193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.120472007485873e-05,
+      "loss": 0.5267,
+      "step": 21194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.119673608186085e-05,
+      "loss": 0.5461,
+      "step": 21195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1188752921611864e-05,
+      "loss": 0.5037,
+      "step": 21196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.118077059420834e-05,
+      "loss": 0.4151,
+      "step": 21197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.117278909974689e-05,
+      "loss": 0.4781,
+      "step": 21198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.116480843832413e-05,
+      "loss": 0.4339,
+      "step": 21199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.115682861003663e-05,
+      "loss": 0.4726,
+      "step": 21200
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 0.4302586317062378,
+      "eval_runtime": 852.6945,
+      "eval_samples_per_second": 28.216,
+      "eval_steps_per_second": 14.108,
+      "step": 21200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1148849614981004e-05,
+      "loss": 0.5743,
+      "step": 21201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.114087145325381e-05,
+      "loss": 0.3352,
+      "step": 21202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.113289412495161e-05,
+      "loss": 0.5194,
+      "step": 21203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1124917630170955e-05,
+      "loss": 0.4636,
+      "step": 21204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.111694196900836e-05,
+      "loss": 0.4056,
+      "step": 21205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.110896714156042e-05,
+      "loss": 0.4247,
+      "step": 21206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.110099314792362e-05,
+      "loss": 0.4644,
+      "step": 21207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.109301998819447e-05,
+      "loss": 0.4282,
+      "step": 21208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.108504766246948e-05,
+      "loss": 0.4618,
+      "step": 21209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.107707617084511e-05,
+      "loss": 0.4488,
+      "step": 21210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.10691055134179e-05,
+      "loss": 0.45,
+      "step": 21211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1061135690284285e-05,
+      "loss": 0.4404,
+      "step": 21212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.105316670154074e-05,
+      "loss": 0.4045,
+      "step": 21213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.104519854728371e-05,
+      "loss": 0.4912,
+      "step": 21214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1037231227609584e-05,
+      "loss": 0.4876,
+      "step": 21215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.102926474261489e-05,
+      "loss": 0.3661,
+      "step": 21216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1021299092396005e-05,
+      "loss": 0.4378,
+      "step": 21217
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1013334277049335e-05,
+      "loss": 0.4271,
+      "step": 21218
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.100537029667129e-05,
+      "loss": 0.4795,
+      "step": 21219
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.099740715135825e-05,
+      "loss": 0.4453,
+      "step": 21220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.098944484120656e-05,
+      "loss": 0.492,
+      "step": 21221
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.098148336631267e-05,
+      "loss": 0.458,
+      "step": 21222
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.097352272677291e-05,
+      "loss": 0.3917,
+      "step": 21223
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.09655629226836e-05,
+      "loss": 0.4913,
+      "step": 21224
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.095760395414112e-05,
+      "loss": 0.4107,
+      "step": 21225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0949645821241745e-05,
+      "loss": 0.4679,
+      "step": 21226
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.094168852408186e-05,
+      "loss": 0.4019,
+      "step": 21227
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.093373206275775e-05,
+      "loss": 0.4413,
+      "step": 21228
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.092577643736571e-05,
+      "loss": 0.4472,
+      "step": 21229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.091782164800202e-05,
+      "loss": 0.5467,
+      "step": 21230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.090986769476295e-05,
+      "loss": 0.4289,
+      "step": 21231
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.090191457774482e-05,
+      "loss": 0.497,
+      "step": 21232
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.089396229704385e-05,
+      "loss": 0.4409,
+      "step": 21233
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.088601085275631e-05,
+      "loss": 0.3964,
+      "step": 21234
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0878060244978424e-05,
+      "loss": 0.38,
+      "step": 21235
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.087011047380642e-05,
+      "loss": 0.4947,
+      "step": 21236
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.086216153933652e-05,
+      "loss": 0.4667,
+      "step": 21237
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0854213441664936e-05,
+      "loss": 0.4147,
+      "step": 21238
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.084626618088786e-05,
+      "loss": 0.3264,
+      "step": 21239
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.083831975710145e-05,
+      "loss": 0.4355,
+      "step": 21240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0830374170401966e-05,
+      "loss": 0.3879,
+      "step": 21241
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.082242942088551e-05,
+      "loss": 0.41,
+      "step": 21242
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.081448550864826e-05,
+      "loss": 0.4886,
+      "step": 21243
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.080654243378637e-05,
+      "loss": 0.4474,
+      "step": 21244
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.079860019639593e-05,
+      "loss": 0.4261,
+      "step": 21245
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.079065879657314e-05,
+      "loss": 0.514,
+      "step": 21246
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0782718234414086e-05,
+      "loss": 0.3688,
+      "step": 21247
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.077477851001488e-05,
+      "loss": 0.3312,
+      "step": 21248
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.076683962347161e-05,
+      "loss": 0.5207,
+      "step": 21249
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.075890157488033e-05,
+      "loss": 0.4164,
+      "step": 21250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.075096436433719e-05,
+      "loss": 0.4449,
+      "step": 21251
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.074302799193821e-05,
+      "loss": 0.5008,
+      "step": 21252
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.073509245777946e-05,
+      "loss": 0.4569,
+      "step": 21253
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.072715776195698e-05,
+      "loss": 0.4284,
+      "step": 21254
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.071922390456682e-05,
+      "loss": 0.4337,
+      "step": 21255
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0711290885704946e-05,
+      "loss": 0.4339,
+      "step": 21256
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.070335870546746e-05,
+      "loss": 0.4906,
+      "step": 21257
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0695427363950337e-05,
+      "loss": 0.3953,
+      "step": 21258
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0687496861249565e-05,
+      "loss": 0.4858,
+      "step": 21259
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0679567197461134e-05,
+      "loss": 0.4914,
+      "step": 21260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0671638372680975e-05,
+      "loss": 0.4625,
+      "step": 21261
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.066371038700513e-05,
+      "loss": 0.5538,
+      "step": 21262
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.065578324052951e-05,
+      "loss": 0.3475,
+      "step": 21263
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0647856933350085e-05,
+      "loss": 0.5698,
+      "step": 21264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0639931465562765e-05,
+      "loss": 0.4911,
+      "step": 21265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.063200683726345e-05,
+      "loss": 0.5133,
+      "step": 21266
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.062408304854811e-05,
+      "loss": 0.4445,
+      "step": 21267
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.061616009951265e-05,
+      "loss": 0.4064,
+      "step": 21268
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.060823799025292e-05,
+      "loss": 0.4465,
+      "step": 21269
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.060031672086483e-05,
+      "loss": 0.5173,
+      "step": 21270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0592396291444214e-05,
+      "loss": 0.4122,
+      "step": 21271
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.058447670208701e-05,
+      "loss": 0.4833,
+      "step": 21272
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0576557952889017e-05,
+      "loss": 0.3625,
+      "step": 21273
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.056864004394611e-05,
+      "loss": 0.4134,
+      "step": 21274
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0560722975354096e-05,
+      "loss": 0.5093,
+      "step": 21275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.055280674720876e-05,
+      "loss": 0.3903,
+      "step": 21276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.054489135960601e-05,
+      "loss": 0.5538,
+      "step": 21277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0536976812641604e-05,
+      "loss": 0.4158,
+      "step": 21278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.052906310641132e-05,
+      "loss": 0.4239,
+      "step": 21279
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.052115024101095e-05,
+      "loss": 0.4267,
+      "step": 21280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0513238216536233e-05,
+      "loss": 0.4803,
+      "step": 21281
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0505327033083008e-05,
+      "loss": 0.4654,
+      "step": 21282
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.049741669074697e-05,
+      "loss": 0.5869,
+      "step": 21283
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.048950718962388e-05,
+      "loss": 0.5108,
+      "step": 21284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0481598529809463e-05,
+      "loss": 0.4646,
+      "step": 21285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0473690711399437e-05,
+      "loss": 0.4591,
+      "step": 21286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0465783734489485e-05,
+      "loss": 0.4784,
+      "step": 21287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0457877599175365e-05,
+      "loss": 0.4369,
+      "step": 21288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0449972305552755e-05,
+      "loss": 0.671,
+      "step": 21289
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0442067853717305e-05,
+      "loss": 0.4074,
+      "step": 21290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0434164243764707e-05,
+      "loss": 0.4242,
+      "step": 21291
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0426261475790597e-05,
+      "loss": 0.4095,
+      "step": 21292
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0418359549890663e-05,
+      "loss": 0.4274,
+      "step": 21293
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.041045846616053e-05,
+      "loss": 0.4494,
+      "step": 21294
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.040255822469582e-05,
+      "loss": 0.4185,
+      "step": 21295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.039465882559215e-05,
+      "loss": 0.4729,
+      "step": 21296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0386760268945113e-05,
+      "loss": 0.4383,
+      "step": 21297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0378862554850353e-05,
+      "loss": 0.4032,
+      "step": 21298
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.037096568340344e-05,
+      "loss": 0.4084,
+      "step": 21299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.036306965469994e-05,
+      "loss": 0.371,
+      "step": 21300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0355174468835436e-05,
+      "loss": 0.4015,
+      "step": 21301
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0347280125905442e-05,
+      "loss": 0.3845,
+      "step": 21302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0339386626005583e-05,
+      "loss": 0.4605,
+      "step": 21303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0331493969231338e-05,
+      "loss": 0.4449,
+      "step": 21304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0323602155678267e-05,
+      "loss": 0.5061,
+      "step": 21305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0315711185441874e-05,
+      "loss": 0.4792,
+      "step": 21306
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0307821058617626e-05,
+      "loss": 0.4625,
+      "step": 21307
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0299931775301094e-05,
+      "loss": 0.4204,
+      "step": 21308
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0292043335587727e-05,
+      "loss": 0.6097,
+      "step": 21309
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0284155739573005e-05,
+      "loss": 0.4062,
+      "step": 21310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0276268987352397e-05,
+      "loss": 0.4245,
+      "step": 21311
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0268383079021313e-05,
+      "loss": 0.4839,
+      "step": 21312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0260498014675277e-05,
+      "loss": 0.4162,
+      "step": 21313
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.025261379440969e-05,
+      "loss": 0.4733,
+      "step": 21314
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0244730418319978e-05,
+      "loss": 0.3811,
+      "step": 21315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0236847886501542e-05,
+      "loss": 0.464,
+      "step": 21316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.022896619904978e-05,
+      "loss": 0.3964,
+      "step": 21317
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0221085356060132e-05,
+      "loss": 0.3993,
+      "step": 21318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.021320535762795e-05,
+      "loss": 0.4151,
+      "step": 21319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0205326203848617e-05,
+      "loss": 0.4891,
+      "step": 21320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.01974478948175e-05,
+      "loss": 0.4586,
+      "step": 21321
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0189570430629942e-05,
+      "loss": 0.5122,
+      "step": 21322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0181693811381263e-05,
+      "loss": 0.3663,
+      "step": 21323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0173818037166855e-05,
+      "loss": 0.4811,
+      "step": 21324
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0165943108082007e-05,
+      "loss": 0.4808,
+      "step": 21325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0158069024222036e-05,
+      "loss": 0.4928,
+      "step": 21326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0150195785682257e-05,
+      "loss": 0.463,
+      "step": 21327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0142323392557905e-05,
+      "loss": 0.4839,
+      "step": 21328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0134451844944344e-05,
+      "loss": 0.411,
+      "step": 21329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0126581142936805e-05,
+      "loss": 0.565,
+      "step": 21330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0118711286630552e-05,
+      "loss": 0.4538,
+      "step": 21331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0110842276120843e-05,
+      "loss": 0.4561,
+      "step": 21332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.010297411150288e-05,
+      "loss": 0.3911,
+      "step": 21333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0095106792871962e-05,
+      "loss": 0.4402,
+      "step": 21334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0087240320323275e-05,
+      "loss": 0.361,
+      "step": 21335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0079374693952034e-05,
+      "loss": 0.3785,
+      "step": 21336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0071509913853435e-05,
+      "loss": 0.4697,
+      "step": 21337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0063645980122634e-05,
+      "loss": 0.4726,
+      "step": 21338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0055782892854877e-05,
+      "loss": 0.453,
+      "step": 21339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0047920652145312e-05,
+      "loss": 0.3708,
+      "step": 21340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.004005925808908e-05,
+      "loss": 0.4648,
+      "step": 21341
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0032198710781346e-05,
+      "loss": 0.4158,
+      "step": 21342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0024339010317203e-05,
+      "loss": 0.5179,
+      "step": 21343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0016480156791848e-05,
+      "loss": 0.4933,
+      "step": 21344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0008622150300382e-05,
+      "loss": 0.4706,
+      "step": 21345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.000076499093789e-05,
+      "loss": 0.4083,
+      "step": 21346
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9992908678799482e-05,
+      "loss": 0.4783,
+      "step": 21347
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9985053213980218e-05,
+      "loss": 0.4259,
+      "step": 21348
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9977198596575228e-05,
+      "loss": 0.4532,
+      "step": 21349
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9969344826679547e-05,
+      "loss": 0.5005,
+      "step": 21350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.996149190438825e-05,
+      "loss": 0.4951,
+      "step": 21351
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9953639829796366e-05,
+      "loss": 0.4946,
+      "step": 21352
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9945788602998938e-05,
+      "loss": 0.4071,
+      "step": 21353
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9937938224090946e-05,
+      "loss": 0.3664,
+      "step": 21354
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9930088693167492e-05,
+      "loss": 0.4788,
+      "step": 21355
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9922240010323545e-05,
+      "loss": 0.4064,
+      "step": 21356
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9914392175654082e-05,
+      "loss": 0.4453,
+      "step": 21357
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.990654518925411e-05,
+      "loss": 0.4764,
+      "step": 21358
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.989869905121856e-05,
+      "loss": 0.4799,
+      "step": 21359
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9890853761642458e-05,
+      "loss": 0.4185,
+      "step": 21360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9883009320620737e-05,
+      "loss": 0.5127,
+      "step": 21361
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9875165728248322e-05,
+      "loss": 0.4145,
+      "step": 21362
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9867322984620173e-05,
+      "loss": 0.4466,
+      "step": 21363
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9859481089831163e-05,
+      "loss": 0.5591,
+      "step": 21364
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.985164004397627e-05,
+      "loss": 0.4643,
+      "step": 21365
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9843799847150378e-05,
+      "loss": 0.4297,
+      "step": 21366
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9835960499448367e-05,
+      "loss": 0.4166,
+      "step": 21367
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.982812200096512e-05,
+      "loss": 0.4091,
+      "step": 21368
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.982028435179548e-05,
+      "loss": 0.427,
+      "step": 21369
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9812447552034395e-05,
+      "loss": 0.4039,
+      "step": 21370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9804611601776644e-05,
+      "loss": 0.4844,
+      "step": 21371
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9796776501117075e-05,
+      "loss": 0.4642,
+      "step": 21372
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.978894225015051e-05,
+      "loss": 0.4181,
+      "step": 21373
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9781108848971806e-05,
+      "loss": 0.5037,
+      "step": 21374
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9773276297675766e-05,
+      "loss": 0.3318,
+      "step": 21375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9765444596357174e-05,
+      "loss": 0.3744,
+      "step": 21376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9757613745110825e-05,
+      "loss": 0.5359,
+      "step": 21377
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.974978374403147e-05,
+      "loss": 0.4526,
+      "step": 21378
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.974195459321394e-05,
+      "loss": 0.4232,
+      "step": 21379
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9734126292752962e-05,
+      "loss": 0.441,
+      "step": 21380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9726298842743283e-05,
+      "loss": 0.5274,
+      "step": 21381
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9718472243279636e-05,
+      "loss": 0.3975,
+      "step": 21382
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.971064649445673e-05,
+      "loss": 0.4787,
+      "step": 21383
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9702821596369336e-05,
+      "loss": 0.4417,
+      "step": 21384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9694997549112135e-05,
+      "loss": 0.4239,
+      "step": 21385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9687174352779822e-05,
+      "loss": 0.4261,
+      "step": 21386
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9679352007467088e-05,
+      "loss": 0.4231,
+      "step": 21387
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9671530513268597e-05,
+      "loss": 0.5336,
+      "step": 21388
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9663709870278987e-05,
+      "loss": 0.4505,
+      "step": 21389
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9655890078592997e-05,
+      "loss": 0.3917,
+      "step": 21390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9648071138305212e-05,
+      "loss": 0.4401,
+      "step": 21391
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.964025304951029e-05,
+      "loss": 0.3357,
+      "step": 21392
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9632435812302838e-05,
+      "loss": 0.4131,
+      "step": 21393
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9624619426777455e-05,
+      "loss": 0.4495,
+      "step": 21394
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.96168038930288e-05,
+      "loss": 0.472,
+      "step": 21395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.960898921115144e-05,
+      "loss": 0.384,
+      "step": 21396
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9601175381239953e-05,
+      "loss": 0.3623,
+      "step": 21397
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9593362403388914e-05,
+      "loss": 0.4459,
+      "step": 21398
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9585550277692854e-05,
+      "loss": 0.5124,
+      "step": 21399
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9577739004246386e-05,
+      "loss": 0.4842,
+      "step": 21400
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 0.427883118391037,
+      "eval_runtime": 846.7395,
+      "eval_samples_per_second": 28.415,
+      "eval_steps_per_second": 14.207,
+      "step": 21400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9569928583144037e-05,
+      "loss": 0.4217,
+      "step": 21401
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9562119014480318e-05,
+      "loss": 0.4747,
+      "step": 21402
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9554310298349764e-05,
+      "loss": 0.4964,
+      "step": 21403
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9546502434846845e-05,
+      "loss": 0.3874,
+      "step": 21404
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9538695424066132e-05,
+      "loss": 0.4541,
+      "step": 21405
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.953088926610209e-05,
+      "loss": 0.3961,
+      "step": 21406
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.952308396104919e-05,
+      "loss": 0.5123,
+      "step": 21407
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9515279509001903e-05,
+      "loss": 0.486,
+      "step": 21408
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9507475910054672e-05,
+      "loss": 0.519,
+      "step": 21409
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9499673164301977e-05,
+      "loss": 0.4375,
+      "step": 21410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.949187127183827e-05,
+      "loss": 0.4245,
+      "step": 21411
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.948407023275793e-05,
+      "loss": 0.3773,
+      "step": 21412
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9476270047155417e-05,
+      "loss": 0.4458,
+      "step": 21413
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9468470715125095e-05,
+      "loss": 0.3952,
+      "step": 21414
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.946067223676141e-05,
+      "loss": 0.4316,
+      "step": 21415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9452874612158744e-05,
+      "loss": 0.4778,
+      "step": 21416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.944507784141145e-05,
+      "loss": 0.3824,
+      "step": 21417
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9437281924613913e-05,
+      "loss": 0.4275,
+      "step": 21418
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9429486861860444e-05,
+      "loss": 0.3744,
+      "step": 21419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9421692653245458e-05,
+      "loss": 0.3862,
+      "step": 21420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9413899298863267e-05,
+      "loss": 0.3649,
+      "step": 21421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9406106798808186e-05,
+      "loss": 0.407,
+      "step": 21422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9398315153174528e-05,
+      "loss": 0.3847,
+      "step": 21423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9390524362056615e-05,
+      "loss": 0.4421,
+      "step": 21424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9382734425548698e-05,
+      "loss": 0.4223,
+      "step": 21425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.937494534374512e-05,
+      "loss": 0.4775,
+      "step": 21426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.936715711674013e-05,
+      "loss": 0.5299,
+      "step": 21427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9359369744627986e-05,
+      "loss": 0.5166,
+      "step": 21428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9351583227502945e-05,
+      "loss": 0.4463,
+      "step": 21429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.934379756545922e-05,
+      "loss": 0.3769,
+      "step": 21430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9336012758591115e-05,
+      "loss": 0.485,
+      "step": 21431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9328228806992797e-05,
+      "loss": 0.4008,
+      "step": 21432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9320445710758493e-05,
+      "loss": 0.4087,
+      "step": 21433
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9312663469982404e-05,
+      "loss": 0.4238,
+      "step": 21434
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.930488208475869e-05,
+      "loss": 0.4395,
+      "step": 21435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9297101555181593e-05,
+      "loss": 0.3395,
+      "step": 21436
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9289321881345254e-05,
+      "loss": 0.3729,
+      "step": 21437
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.928154306334383e-05,
+      "loss": 0.5015,
+      "step": 21438
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9273765101271468e-05,
+      "loss": 0.3645,
+      "step": 21439
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9265987995222288e-05,
+      "loss": 0.4381,
+      "step": 21440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9258211745290464e-05,
+      "loss": 0.3563,
+      "step": 21441
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9250436351570087e-05,
+      "loss": 0.3567,
+      "step": 21442
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9242661814155282e-05,
+      "loss": 0.4646,
+      "step": 21443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9234888133140126e-05,
+      "loss": 0.4684,
+      "step": 21444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.922711530861869e-05,
+      "loss": 0.476,
+      "step": 21445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9219343340685112e-05,
+      "loss": 0.3943,
+      "step": 21446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9211572229433416e-05,
+      "loss": 0.4101,
+      "step": 21447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9203801974957666e-05,
+      "loss": 0.4038,
+      "step": 21448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9196032577351904e-05,
+      "loss": 0.4381,
+      "step": 21449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9188264036710145e-05,
+      "loss": 0.43,
+      "step": 21450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9180496353126476e-05,
+      "loss": 0.4607,
+      "step": 21451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9172729526694863e-05,
+      "loss": 0.4179,
+      "step": 21452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.916496355750933e-05,
+      "loss": 0.4977,
+      "step": 21453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9157198445663858e-05,
+      "loss": 0.522,
+      "step": 21454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9149434191252435e-05,
+      "loss": 0.4652,
+      "step": 21455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9141670794369003e-05,
+      "loss": 0.4683,
+      "step": 21456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9133908255107588e-05,
+      "loss": 0.5218,
+      "step": 21457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9126146573562117e-05,
+      "loss": 0.408,
+      "step": 21458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9118385749826526e-05,
+      "loss": 0.3989,
+      "step": 21459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.911062578399474e-05,
+      "loss": 0.4527,
+      "step": 21460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9102866676160668e-05,
+      "loss": 0.3744,
+      "step": 21461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.909510842641826e-05,
+      "loss": 0.5629,
+      "step": 21462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.908735103486141e-05,
+      "loss": 0.4309,
+      "step": 21463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9079594501583996e-05,
+      "loss": 0.4601,
+      "step": 21464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9071838826679886e-05,
+      "loss": 0.4333,
+      "step": 21465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9064084010242944e-05,
+      "loss": 0.4781,
+      "step": 21466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9056330052367064e-05,
+      "loss": 0.4347,
+      "step": 21467
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.904857695314609e-05,
+      "loss": 0.4521,
+      "step": 21468
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.904082471267384e-05,
+      "loss": 0.5077,
+      "step": 21469
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.903307333104416e-05,
+      "loss": 0.5033,
+      "step": 21470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9025322808350818e-05,
+      "loss": 0.5116,
+      "step": 21471
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9017573144687692e-05,
+      "loss": 0.4241,
+      "step": 21472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.900982434014855e-05,
+      "loss": 0.4064,
+      "step": 21473
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9002076394827172e-05,
+      "loss": 0.4125,
+      "step": 21474
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8994329308817337e-05,
+      "loss": 0.5148,
+      "step": 21475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8986583082212782e-05,
+      "loss": 0.3676,
+      "step": 21476
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.897883771510732e-05,
+      "loss": 0.4882,
+      "step": 21477
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.897109320759468e-05,
+      "loss": 0.5077,
+      "step": 21478
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8963349559768572e-05,
+      "loss": 0.4667,
+      "step": 21479
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.895560677172273e-05,
+      "loss": 0.446,
+      "step": 21480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.894786484355084e-05,
+      "loss": 0.4794,
+      "step": 21481
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8940123775346672e-05,
+      "loss": 0.4112,
+      "step": 21482
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8932383567203868e-05,
+      "loss": 0.4296,
+      "step": 21483
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8924644219216136e-05,
+      "loss": 0.3846,
+      "step": 21484
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.891690573147713e-05,
+      "loss": 0.3802,
+      "step": 21485
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8909168104080485e-05,
+      "loss": 0.3928,
+      "step": 21486
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.890143133711991e-05,
+      "loss": 0.4237,
+      "step": 21487
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8893695430689027e-05,
+      "loss": 0.4192,
+      "step": 21488
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.888596038488146e-05,
+      "loss": 0.4445,
+      "step": 21489
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8878226199790825e-05,
+      "loss": 0.4432,
+      "step": 21490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8870492875510722e-05,
+      "loss": 0.3752,
+      "step": 21491
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8862760412134748e-05,
+      "loss": 0.4636,
+      "step": 21492
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8855028809756522e-05,
+      "loss": 0.523,
+      "step": 21493
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.884729806846962e-05,
+      "loss": 0.437,
+      "step": 21494
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.883956818836758e-05,
+      "loss": 0.4347,
+      "step": 21495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8831839169543996e-05,
+      "loss": 0.4599,
+      "step": 21496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8824111012092348e-05,
+      "loss": 0.4056,
+      "step": 21497
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8816383716106242e-05,
+      "loss": 0.3974,
+      "step": 21498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8808657281679197e-05,
+      "loss": 0.4161,
+      "step": 21499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8800931708904698e-05,
+      "loss": 0.4379,
+      "step": 21500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8793206997876275e-05,
+      "loss": 0.3619,
+      "step": 21501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8785483148687375e-05,
+      "loss": 0.461,
+      "step": 21502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8777760161431545e-05,
+      "loss": 0.484,
+      "step": 21503
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.877003803620224e-05,
+      "loss": 0.4937,
+      "step": 21504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8762316773092945e-05,
+      "loss": 0.3911,
+      "step": 21505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8754596372197017e-05,
+      "loss": 0.4101,
+      "step": 21506
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8746876833608005e-05,
+      "loss": 0.491,
+      "step": 21507
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.87391581574193e-05,
+      "loss": 0.5093,
+      "step": 21508
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8731440343724324e-05,
+      "loss": 0.4497,
+      "step": 21509
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8723723392616485e-05,
+      "loss": 0.4459,
+      "step": 21510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8716007304189176e-05,
+      "loss": 0.4297,
+      "step": 21511
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8708292078535813e-05,
+      "loss": 0.4831,
+      "step": 21512
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8700577715749767e-05,
+      "loss": 0.5406,
+      "step": 21513
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8692864215924408e-05,
+      "loss": 0.3772,
+      "step": 21514
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8685151579153093e-05,
+      "loss": 0.4911,
+      "step": 21515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8677439805529127e-05,
+      "loss": 0.3855,
+      "step": 21516
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.866972889514593e-05,
+      "loss": 0.4235,
+      "step": 21517
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8662018848096784e-05,
+      "loss": 0.4556,
+      "step": 21518
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8654309664475008e-05,
+      "loss": 0.4914,
+      "step": 21519
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8646601344373924e-05,
+      "loss": 0.4075,
+      "step": 21520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8638893887886775e-05,
+      "loss": 0.4115,
+      "step": 21521
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.863118729510692e-05,
+      "loss": 0.4579,
+      "step": 21522
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.862348156612761e-05,
+      "loss": 0.394,
+      "step": 21523
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.861577670104211e-05,
+      "loss": 0.4946,
+      "step": 21524
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8608072699943645e-05,
+      "loss": 0.5316,
+      "step": 21525
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.86003695629255e-05,
+      "loss": 0.47,
+      "step": 21526
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8592667290080854e-05,
+      "loss": 0.4598,
+      "step": 21527
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8584965881503e-05,
+      "loss": 0.3661,
+      "step": 21528
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8577265337285107e-05,
+      "loss": 0.5375,
+      "step": 21529
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8569565657520404e-05,
+      "loss": 0.3869,
+      "step": 21530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8561866842302054e-05,
+      "loss": 0.4842,
+      "step": 21531
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.855416889172322e-05,
+      "loss": 0.5169,
+      "step": 21532
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.854647180587714e-05,
+      "loss": 0.4158,
+      "step": 21533
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8538775584856935e-05,
+      "loss": 0.4883,
+      "step": 21534
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.853108022875576e-05,
+      "loss": 0.4621,
+      "step": 21535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.852338573766675e-05,
+      "loss": 0.4202,
+      "step": 21536
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8515692111683002e-05,
+      "loss": 0.4678,
+      "step": 21537
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8507999350897706e-05,
+      "loss": 0.45,
+      "step": 21538
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8500307455403928e-05,
+      "loss": 0.4217,
+      "step": 21539
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.849261642529478e-05,
+      "loss": 0.4433,
+      "step": 21540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8484926260663324e-05,
+      "loss": 0.3753,
+      "step": 21541
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8477236961602638e-05,
+      "loss": 0.4605,
+      "step": 21542
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8469548528205826e-05,
+      "loss": 0.4609,
+      "step": 21543
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8461860960565913e-05,
+      "loss": 0.441,
+      "step": 21544
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8454174258775957e-05,
+      "loss": 0.4458,
+      "step": 21545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8446488422928985e-05,
+      "loss": 0.4868,
+      "step": 21546
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8438803453118003e-05,
+      "loss": 0.403,
+      "step": 21547
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8431119349436065e-05,
+      "loss": 0.4965,
+      "step": 21548
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8423436111976155e-05,
+      "loss": 0.4779,
+      "step": 21549
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8415753740831263e-05,
+      "loss": 0.4366,
+      "step": 21550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8408072236094384e-05,
+      "loss": 0.4589,
+      "step": 21551
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8400391597858435e-05,
+      "loss": 0.4175,
+      "step": 21552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8392711826216455e-05,
+      "loss": 0.4004,
+      "step": 21553
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8385032921261366e-05,
+      "loss": 0.463,
+      "step": 21554
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8377354883086106e-05,
+      "loss": 0.4703,
+      "step": 21555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8369677711783592e-05,
+      "loss": 0.4887,
+      "step": 21556
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.836200140744677e-05,
+      "loss": 0.4184,
+      "step": 21557
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8354325970168484e-05,
+      "loss": 0.4568,
+      "step": 21558
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8346651400041725e-05,
+      "loss": 0.4057,
+      "step": 21559
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.833897769715933e-05,
+      "loss": 0.3913,
+      "step": 21560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8331304861614184e-05,
+      "loss": 0.4961,
+      "step": 21561
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8323632893499154e-05,
+      "loss": 0.4492,
+      "step": 21562
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8315961792907065e-05,
+      "loss": 0.4778,
+      "step": 21563
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.830829155993082e-05,
+      "loss": 0.4553,
+      "step": 21564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.830062219466324e-05,
+      "loss": 0.4668,
+      "step": 21565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.829295369719712e-05,
+      "loss": 0.3611,
+      "step": 21566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8285286067625305e-05,
+      "loss": 0.4407,
+      "step": 21567
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.827761930604055e-05,
+      "loss": 0.485,
+      "step": 21568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.826995341253572e-05,
+      "loss": 0.3991,
+      "step": 21569
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.826228838720356e-05,
+      "loss": 0.5229,
+      "step": 21570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.825462423013684e-05,
+      "loss": 0.4109,
+      "step": 21571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8246960941428335e-05,
+      "loss": 0.4543,
+      "step": 21572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8239298521170753e-05,
+      "loss": 0.4502,
+      "step": 21573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8231636969456897e-05,
+      "loss": 0.4069,
+      "step": 21574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.822397628637947e-05,
+      "loss": 0.4127,
+      "step": 21575
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.82163164720312e-05,
+      "loss": 0.4078,
+      "step": 21576
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.820865752650478e-05,
+      "loss": 0.4454,
+      "step": 21577
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8200999449892884e-05,
+      "loss": 0.4179,
+      "step": 21578
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8193342242288267e-05,
+      "loss": 0.4734,
+      "step": 21579
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8185685903783578e-05,
+      "loss": 0.4037,
+      "step": 21580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8178030434471482e-05,
+      "loss": 0.4323,
+      "step": 21581
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8170375834444617e-05,
+      "loss": 0.476,
+      "step": 21582
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8162722103795625e-05,
+      "loss": 0.4712,
+      "step": 21583
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8155069242617195e-05,
+      "loss": 0.4707,
+      "step": 21584
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8147417251001908e-05,
+      "loss": 0.4848,
+      "step": 21585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.81397661290424e-05,
+      "loss": 0.4887,
+      "step": 21586
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8132115876831257e-05,
+      "loss": 0.5142,
+      "step": 21587
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8124466494461045e-05,
+      "loss": 0.3464,
+      "step": 21588
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8116817982024423e-05,
+      "loss": 0.4762,
+      "step": 21589
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8109170339613923e-05,
+      "loss": 0.4434,
+      "step": 21590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8101523567322097e-05,
+      "loss": 0.426,
+      "step": 21591
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.809387766524151e-05,
+      "loss": 0.4582,
+      "step": 21592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8086232633464706e-05,
+      "loss": 0.4364,
+      "step": 21593
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8078588472084167e-05,
+      "loss": 0.452,
+      "step": 21594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8070945181192486e-05,
+      "loss": 0.4635,
+      "step": 21595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8063302760882136e-05,
+      "loss": 0.4659,
+      "step": 21596
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8055661211245633e-05,
+      "loss": 0.4194,
+      "step": 21597
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.804802053237544e-05,
+      "loss": 0.3642,
+      "step": 21598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8040380724364025e-05,
+      "loss": 0.4587,
+      "step": 21599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8032741787303906e-05,
+      "loss": 0.3508,
+      "step": 21600
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 0.42593833804130554,
+      "eval_runtime": 851.8703,
+      "eval_samples_per_second": 28.244,
+      "eval_steps_per_second": 14.122,
+      "step": 21600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8025103721287503e-05,
+      "loss": 0.3939,
+      "step": 21601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8017466526407278e-05,
+      "loss": 0.4327,
+      "step": 21602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8009830202755662e-05,
+      "loss": 0.4881,
+      "step": 21603
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8002194750425036e-05,
+      "loss": 0.5002,
+      "step": 21604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.799456016950789e-05,
+      "loss": 0.4513,
+      "step": 21605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7986926460096597e-05,
+      "loss": 0.4034,
+      "step": 21606
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.797929362228353e-05,
+      "loss": 0.4897,
+      "step": 21607
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7971661656161098e-05,
+      "loss": 0.3834,
+      "step": 21608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.796403056182163e-05,
+      "loss": 0.4078,
+      "step": 21609
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.795640033935756e-05,
+      "loss": 0.3976,
+      "step": 21610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7948770988861195e-05,
+      "loss": 0.4723,
+      "step": 21611
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7941142510424877e-05,
+      "loss": 0.3384,
+      "step": 21612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7933514904140933e-05,
+      "loss": 0.4853,
+      "step": 21613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7925888170101665e-05,
+      "loss": 0.4173,
+      "step": 21614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7918262308399434e-05,
+      "loss": 0.4396,
+      "step": 21615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7910637319126522e-05,
+      "loss": 0.457,
+      "step": 21616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7903013202375195e-05,
+      "loss": 0.413,
+      "step": 21617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.789538995823774e-05,
+      "loss": 0.4667,
+      "step": 21618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.788776758680639e-05,
+      "loss": 0.4643,
+      "step": 21619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7880146088173463e-05,
+      "loss": 0.422,
+      "step": 21620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7872525462431186e-05,
+      "loss": 0.4392,
+      "step": 21621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7864905709671774e-05,
+      "loss": 0.4212,
+      "step": 21622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7857286829987473e-05,
+      "loss": 0.519,
+      "step": 21623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7849668823470477e-05,
+      "loss": 0.4425,
+      "step": 21624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7842051690212967e-05,
+      "loss": 0.4597,
+      "step": 21625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7834435430307203e-05,
+      "loss": 0.5067,
+      "step": 21626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7826820043845326e-05,
+      "loss": 0.4306,
+      "step": 21627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7819205530919513e-05,
+      "loss": 0.4388,
+      "step": 21628
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7811591891621926e-05,
+      "loss": 0.4335,
+      "step": 21629
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.780397912604469e-05,
+      "loss": 0.4051,
+      "step": 21630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7796367234279986e-05,
+      "loss": 0.4283,
+      "step": 21631
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.778875621641993e-05,
+      "loss": 0.4403,
+      "step": 21632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7781146072556642e-05,
+      "loss": 0.4089,
+      "step": 21633
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7773536802782218e-05,
+      "loss": 0.4444,
+      "step": 21634
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7765928407188722e-05,
+      "loss": 0.4283,
+      "step": 21635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7758320885868327e-05,
+      "loss": 0.5476,
+      "step": 21636
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7750714238913056e-05,
+      "loss": 0.4531,
+      "step": 21637
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.774310846641499e-05,
+      "loss": 0.4284,
+      "step": 21638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.773550356846617e-05,
+      "loss": 0.4352,
+      "step": 21639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7727899545158655e-05,
+      "loss": 0.4183,
+      "step": 21640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7720296396584457e-05,
+      "loss": 0.4411,
+      "step": 21641
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7712694122835614e-05,
+      "loss": 0.4705,
+      "step": 21642
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7705092724004144e-05,
+      "loss": 0.3876,
+      "step": 21643
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.769749220018202e-05,
+      "loss": 0.4074,
+      "step": 21644
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7689892551461284e-05,
+      "loss": 0.4343,
+      "step": 21645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.768229377793389e-05,
+      "loss": 0.43,
+      "step": 21646
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7674695879691804e-05,
+      "loss": 0.5276,
+      "step": 21647
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.766709885682699e-05,
+      "loss": 0.4639,
+      "step": 21648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.765950270943137e-05,
+      "loss": 0.4844,
+      "step": 21649
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7651907437596937e-05,
+      "loss": 0.4251,
+      "step": 21650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7644313041415603e-05,
+      "loss": 0.5002,
+      "step": 21651
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.763671952097926e-05,
+      "loss": 0.3961,
+      "step": 21652
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.762912687637984e-05,
+      "loss": 0.4699,
+      "step": 21653
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7621535107709206e-05,
+      "loss": 0.5009,
+      "step": 21654
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7613944215059284e-05,
+      "loss": 0.4063,
+      "step": 21655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7606354198521946e-05,
+      "loss": 0.3338,
+      "step": 21656
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7598765058189034e-05,
+      "loss": 0.4346,
+      "step": 21657
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7591176794152408e-05,
+      "loss": 0.421,
+      "step": 21658
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7583589406503916e-05,
+      "loss": 0.4581,
+      "step": 21659
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.757600289533536e-05,
+      "loss": 0.4877,
+      "step": 21660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.756841726073862e-05,
+      "loss": 0.4298,
+      "step": 21661
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.756083250280548e-05,
+      "loss": 0.4755,
+      "step": 21662
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7553248621627736e-05,
+      "loss": 0.4856,
+      "step": 21663
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.754566561729719e-05,
+      "loss": 0.46,
+      "step": 21664
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7538083489905574e-05,
+      "loss": 0.4503,
+      "step": 21665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.753050223954473e-05,
+      "loss": 0.3879,
+      "step": 21666
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7522921866306384e-05,
+      "loss": 0.3841,
+      "step": 21667
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.751534237028227e-05,
+      "loss": 0.419,
+      "step": 21668
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.750776375156415e-05,
+      "loss": 0.4154,
+      "step": 21669
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.75001860102437e-05,
+      "loss": 0.3952,
+      "step": 21670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7492609146412718e-05,
+      "loss": 0.4657,
+      "step": 21671
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7485033160162853e-05,
+      "loss": 0.4428,
+      "step": 21672
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7477458051585815e-05,
+      "loss": 0.4358,
+      "step": 21673
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.746988382077328e-05,
+      "loss": 0.4892,
+      "step": 21674
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.74623104678169e-05,
+      "loss": 0.4059,
+      "step": 21675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.745473799280839e-05,
+      "loss": 0.4132,
+      "step": 21676
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7447166395839385e-05,
+      "loss": 0.4481,
+      "step": 21677
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7439595677001505e-05,
+      "loss": 0.4213,
+      "step": 21678
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.743202583638641e-05,
+      "loss": 0.4794,
+      "step": 21679
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.742445687408567e-05,
+      "loss": 0.3959,
+      "step": 21680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7416888790190954e-05,
+      "loss": 0.4676,
+      "step": 21681
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7409321584793833e-05,
+      "loss": 0.4437,
+      "step": 21682
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7401755257985896e-05,
+      "loss": 0.4413,
+      "step": 21683
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7394189809858727e-05,
+      "loss": 0.3816,
+      "step": 21684
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.738662524050386e-05,
+      "loss": 0.4369,
+      "step": 21685
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.737906155001292e-05,
+      "loss": 0.3732,
+      "step": 21686
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.737149873847741e-05,
+      "loss": 0.4013,
+      "step": 21687
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7363936805988854e-05,
+      "loss": 0.3975,
+      "step": 21688
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.735637575263881e-05,
+      "loss": 0.4674,
+      "step": 21689
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7348815578518762e-05,
+      "loss": 0.3904,
+      "step": 21690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7341256283720196e-05,
+      "loss": 0.5222,
+      "step": 21691
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.733369786833466e-05,
+      "loss": 0.4456,
+      "step": 21692
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.732614033245361e-05,
+      "loss": 0.381,
+      "step": 21693
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.731858367616853e-05,
+      "loss": 0.4723,
+      "step": 21694
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.731102789957085e-05,
+      "loss": 0.5007,
+      "step": 21695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.730347300275201e-05,
+      "loss": 0.3905,
+      "step": 21696
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7295918985803514e-05,
+      "loss": 0.3884,
+      "step": 21697
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7288365848816753e-05,
+      "loss": 0.4037,
+      "step": 21698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7280813591883148e-05,
+      "loss": 0.3626,
+      "step": 21699
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7273262215094098e-05,
+      "loss": 0.4303,
+      "step": 21700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7265711718540975e-05,
+      "loss": 0.4891,
+      "step": 21701
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7258162102315233e-05,
+      "loss": 0.4397,
+      "step": 21702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7250613366508214e-05,
+      "loss": 0.4566,
+      "step": 21703
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7243065511211273e-05,
+      "loss": 0.4813,
+      "step": 21704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.723551853651577e-05,
+      "loss": 0.4184,
+      "step": 21705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7227972442513018e-05,
+      "loss": 0.4288,
+      "step": 21706
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.722042722929442e-05,
+      "loss": 0.4372,
+      "step": 21707
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7212882896951254e-05,
+      "loss": 0.4293,
+      "step": 21708
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.720533944557483e-05,
+      "loss": 0.4033,
+      "step": 21709
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7197796875256466e-05,
+      "loss": 0.4426,
+      "step": 21710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.719025518608741e-05,
+      "loss": 0.5433,
+      "step": 21711
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7182714378159003e-05,
+      "loss": 0.4833,
+      "step": 21712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.717517445156249e-05,
+      "loss": 0.4497,
+      "step": 21713
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7167635406389115e-05,
+      "loss": 0.3476,
+      "step": 21714
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7160097242730142e-05,
+      "loss": 0.4624,
+      "step": 21715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.715255996067676e-05,
+      "loss": 0.4076,
+      "step": 21716
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7145023560320282e-05,
+      "loss": 0.4292,
+      "step": 21717
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.713748804175187e-05,
+      "loss": 0.4617,
+      "step": 21718
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7129953405062723e-05,
+      "loss": 0.3788,
+      "step": 21719
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7122419650344065e-05,
+      "loss": 0.3667,
+      "step": 21720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7114886777687033e-05,
+      "loss": 0.4149,
+      "step": 21721
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7107354787182848e-05,
+      "loss": 0.4669,
+      "step": 21722
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7099823678922664e-05,
+      "loss": 0.38,
+      "step": 21723
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.709229345299762e-05,
+      "loss": 0.4439,
+      "step": 21724
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7084764109498863e-05,
+      "loss": 0.3493,
+      "step": 21725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7077235648517506e-05,
+      "loss": 0.4565,
+      "step": 21726
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7069708070144673e-05,
+      "loss": 0.3901,
+      "step": 21727
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7062181374471495e-05,
+      "loss": 0.4093,
+      "step": 21728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.705465556158906e-05,
+      "loss": 0.431,
+      "step": 21729
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7047130631588456e-05,
+      "loss": 0.5485,
+      "step": 21730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.703960658456075e-05,
+      "loss": 0.389,
+      "step": 21731
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7032083420597e-05,
+      "loss": 0.5509,
+      "step": 21732
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7024561139788285e-05,
+      "loss": 0.4259,
+      "step": 21733
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7017039742225657e-05,
+      "loss": 0.4185,
+      "step": 21734
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7009519228000126e-05,
+      "loss": 0.4182,
+      "step": 21735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.700199959720272e-05,
+      "loss": 0.4457,
+      "step": 21736
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6994480849924432e-05,
+      "loss": 0.4173,
+      "step": 21737
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6986962986256304e-05,
+      "loss": 0.4313,
+      "step": 21738
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.697944600628931e-05,
+      "loss": 0.4158,
+      "step": 21739
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6971929910114446e-05,
+      "loss": 0.4435,
+      "step": 21740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.696441469782265e-05,
+      "loss": 0.376,
+      "step": 21741
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6956900369504868e-05,
+      "loss": 0.4201,
+      "step": 21742
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6949386925252118e-05,
+      "loss": 0.5248,
+      "step": 21743
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6941874365155283e-05,
+      "loss": 0.4247,
+      "step": 21744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6934362689305316e-05,
+      "loss": 0.4068,
+      "step": 21745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6926851897793115e-05,
+      "loss": 0.3816,
+      "step": 21746
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6919341990709556e-05,
+      "loss": 0.4861,
+      "step": 21747
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.691183296814561e-05,
+      "loss": 0.4451,
+      "step": 21748
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6904324830192118e-05,
+      "loss": 0.4463,
+      "step": 21749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6896817576939947e-05,
+      "loss": 0.3913,
+      "step": 21750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6889311208479982e-05,
+      "loss": 0.4384,
+      "step": 21751
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6881805724903018e-05,
+      "loss": 0.5045,
+      "step": 21752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6874301126299982e-05,
+      "loss": 0.438,
+      "step": 21753
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.686679741276167e-05,
+      "loss": 0.422,
+      "step": 21754
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.685929458437888e-05,
+      "loss": 0.4356,
+      "step": 21755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6851792641242445e-05,
+      "loss": 0.4913,
+      "step": 21756
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.684429158344315e-05,
+      "loss": 0.4327,
+      "step": 21757
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.683679141107176e-05,
+      "loss": 0.3768,
+      "step": 21758
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.682929212421911e-05,
+      "loss": 0.466,
+      "step": 21759
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.682179372297593e-05,
+      "loss": 0.4172,
+      "step": 21760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6814296207432986e-05,
+      "loss": 0.4251,
+      "step": 21761
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.680679957768103e-05,
+      "loss": 0.4674,
+      "step": 21762
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.679930383381074e-05,
+      "loss": 0.4148,
+      "step": 21763
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6791808975912925e-05,
+      "loss": 0.3854,
+      "step": 21764
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.678431500407825e-05,
+      "loss": 0.3629,
+      "step": 21765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.677682191839742e-05,
+      "loss": 0.4158,
+      "step": 21766
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6769329718961146e-05,
+      "loss": 0.4089,
+      "step": 21767
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.676183840586005e-05,
+      "loss": 0.4551,
+      "step": 21768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.675434797918488e-05,
+      "loss": 0.5002,
+      "step": 21769
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6746858439026258e-05,
+      "loss": 0.4736,
+      "step": 21770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6739369785474833e-05,
+      "loss": 0.4574,
+      "step": 21771
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.673188201862126e-05,
+      "loss": 0.4163,
+      "step": 21772
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6724395138556102e-05,
+      "loss": 0.4206,
+      "step": 21773
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6716909145370084e-05,
+      "loss": 0.387,
+      "step": 21774
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.670942403915373e-05,
+      "loss": 0.4743,
+      "step": 21775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6701939819997647e-05,
+      "loss": 0.3832,
+      "step": 21776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6694456487992404e-05,
+      "loss": 0.4511,
+      "step": 21777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.668697404322863e-05,
+      "loss": 0.3912,
+      "step": 21778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6679492485796844e-05,
+      "loss": 0.478,
+      "step": 21779
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6672011815787625e-05,
+      "loss": 0.4183,
+      "step": 21780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6664532033291477e-05,
+      "loss": 0.4666,
+      "step": 21781
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6657053138398937e-05,
+      "loss": 0.4019,
+      "step": 21782
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6649575131200567e-05,
+      "loss": 0.4619,
+      "step": 21783
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6642098011786855e-05,
+      "loss": 0.5063,
+      "step": 21784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.663462178024828e-05,
+      "loss": 0.4033,
+      "step": 21785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6627146436675344e-05,
+      "loss": 0.4682,
+      "step": 21786
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.661967198115849e-05,
+      "loss": 0.4227,
+      "step": 21787
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6612198413788246e-05,
+      "loss": 0.4143,
+      "step": 21788
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6604725734655034e-05,
+      "loss": 0.4869,
+      "step": 21789
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.659725394384931e-05,
+      "loss": 0.3544,
+      "step": 21790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.658978304146149e-05,
+      "loss": 0.4194,
+      "step": 21791
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6582313027582005e-05,
+      "loss": 0.5581,
+      "step": 21792
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6574843902301226e-05,
+      "loss": 0.5338,
+      "step": 21793
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.656737566570964e-05,
+      "loss": 0.4168,
+      "step": 21794
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6559908317897586e-05,
+      "loss": 0.4218,
+      "step": 21795
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6552441858955457e-05,
+      "loss": 0.4711,
+      "step": 21796
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.654497628897361e-05,
+      "loss": 0.4602,
+      "step": 21797
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6537511608042376e-05,
+      "loss": 0.4412,
+      "step": 21798
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6530047816252167e-05,
+      "loss": 0.4109,
+      "step": 21799
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6522584913693294e-05,
+      "loss": 0.4452,
+      "step": 21800
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 0.42362895607948303,
+      "eval_runtime": 844.2826,
+      "eval_samples_per_second": 28.498,
+      "eval_steps_per_second": 14.249,
+      "step": 21800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6515122900456067e-05,
+      "loss": 0.4085,
+      "step": 21801
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.650766177663081e-05,
+      "loss": 0.4523,
+      "step": 21802
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6500201542307813e-05,
+      "loss": 0.361,
+      "step": 21803
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6492742197577404e-05,
+      "loss": 0.4339,
+      "step": 21804
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6485283742529843e-05,
+      "loss": 0.4453,
+      "step": 21805
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6477826177255417e-05,
+      "loss": 0.4398,
+      "step": 21806
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6470369501844373e-05,
+      "loss": 0.3775,
+      "step": 21807
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6462913716386927e-05,
+      "loss": 0.4247,
+      "step": 21808
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.64554588209734e-05,
+      "loss": 0.3573,
+      "step": 21809
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.644800481569397e-05,
+      "loss": 0.4952,
+      "step": 21810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6440551700638872e-05,
+      "loss": 0.4372,
+      "step": 21811
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6433099475898306e-05,
+      "loss": 0.4034,
+      "step": 21812
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.642564814156242e-05,
+      "loss": 0.4784,
+      "step": 21813
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6418197697721492e-05,
+      "loss": 0.4751,
+      "step": 21814
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.641074814446566e-05,
+      "loss": 0.4634,
+      "step": 21815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.640329948188507e-05,
+      "loss": 0.4102,
+      "step": 21816
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6395851710069887e-05,
+      "loss": 0.4143,
+      "step": 21817
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6388404829110225e-05,
+      "loss": 0.4007,
+      "step": 21818
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.638095883909628e-05,
+      "loss": 0.4508,
+      "step": 21819
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6373513740118127e-05,
+      "loss": 0.3703,
+      "step": 21820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6366069532265902e-05,
+      "loss": 0.4307,
+      "step": 21821
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.635862621562969e-05,
+      "loss": 0.4486,
+      "step": 21822
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6351183790299537e-05,
+      "loss": 0.5047,
+      "step": 21823
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6343742256365612e-05,
+      "loss": 0.4222,
+      "step": 21824
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6336301613917923e-05,
+      "loss": 0.4337,
+      "step": 21825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.632886186304654e-05,
+      "loss": 0.4538,
+      "step": 21826
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6321423003841506e-05,
+      "loss": 0.388,
+      "step": 21827
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6313985036392864e-05,
+      "loss": 0.4847,
+      "step": 21828
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6306547960790595e-05,
+      "loss": 0.4403,
+      "step": 21829
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6299111777124784e-05,
+      "loss": 0.2851,
+      "step": 21830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6291676485485406e-05,
+      "loss": 0.4016,
+      "step": 21831
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.628424208596243e-05,
+      "loss": 0.384,
+      "step": 21832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6276808578645863e-05,
+      "loss": 0.4096,
+      "step": 21833
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.626937596362563e-05,
+      "loss": 0.3945,
+      "step": 21834
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6261944240991753e-05,
+      "loss": 0.431,
+      "step": 21835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6254513410834157e-05,
+      "loss": 0.4042,
+      "step": 21836
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6247083473242785e-05,
+      "loss": 0.4865,
+      "step": 21837
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6239654428307536e-05,
+      "loss": 0.4332,
+      "step": 21838
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.623222627611833e-05,
+      "loss": 0.5501,
+      "step": 21839
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.622479901676511e-05,
+      "loss": 0.4511,
+      "step": 21840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6217372650337746e-05,
+      "loss": 0.4558,
+      "step": 21841
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.620994717692613e-05,
+      "loss": 0.4259,
+      "step": 21842
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6202522596620127e-05,
+      "loss": 0.4446,
+      "step": 21843
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6195098909509563e-05,
+      "loss": 0.4339,
+      "step": 21844
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6187676115684366e-05,
+      "loss": 0.4485,
+      "step": 21845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6180254215234333e-05,
+      "loss": 0.3962,
+      "step": 21846
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6172833208249303e-05,
+      "loss": 0.3791,
+      "step": 21847
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.616541309481908e-05,
+      "loss": 0.3627,
+      "step": 21848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.615799387503346e-05,
+      "loss": 0.5002,
+      "step": 21849
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6150575548982292e-05,
+      "loss": 0.4027,
+      "step": 21850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.614315811675534e-05,
+      "loss": 0.4157,
+      "step": 21851
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6135741578442364e-05,
+      "loss": 0.4859,
+      "step": 21852
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6128325934133135e-05,
+      "loss": 0.3957,
+      "step": 21853
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6120911183917395e-05,
+      "loss": 0.4367,
+      "step": 21854
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6113497327884917e-05,
+      "loss": 0.5312,
+      "step": 21855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6106084366125428e-05,
+      "loss": 0.5448,
+      "step": 21856
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6098672298728634e-05,
+      "loss": 0.5514,
+      "step": 21857
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.609126112578425e-05,
+      "loss": 0.3575,
+      "step": 21858
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6083850847381975e-05,
+      "loss": 0.3788,
+      "step": 21859
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6076441463611478e-05,
+      "loss": 0.424,
+      "step": 21860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.606903297456249e-05,
+      "loss": 0.4099,
+      "step": 21861
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.606162538032465e-05,
+      "loss": 0.4925,
+      "step": 21862
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6054218680987606e-05,
+      "loss": 0.4877,
+      "step": 21863
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6046812876641013e-05,
+      "loss": 0.442,
+      "step": 21864
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6039407967374475e-05,
+      "loss": 0.4491,
+      "step": 21865
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.603200395327767e-05,
+      "loss": 0.4205,
+      "step": 21866
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.602460083444018e-05,
+      "loss": 0.4434,
+      "step": 21867
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.601719861095163e-05,
+      "loss": 0.4238,
+      "step": 21868
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6009797282901582e-05,
+      "loss": 0.4456,
+      "step": 21869
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6002396850379608e-05,
+      "loss": 0.4183,
+      "step": 21870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5994997313475323e-05,
+      "loss": 0.3838,
+      "step": 21871
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.598759867227827e-05,
+      "loss": 0.4199,
+      "step": 21872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5980200926877985e-05,
+      "loss": 0.3701,
+      "step": 21873
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.597280407736401e-05,
+      "loss": 0.3999,
+      "step": 21874
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.596540812382584e-05,
+      "loss": 0.4614,
+      "step": 21875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5958013066353048e-05,
+      "loss": 0.4067,
+      "step": 21876
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.595061890503512e-05,
+      "loss": 0.4444,
+      "step": 21877
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5943225639961545e-05,
+      "loss": 0.4526,
+      "step": 21878
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.59358332712218e-05,
+      "loss": 0.4436,
+      "step": 21879
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5928441798905334e-05,
+      "loss": 0.4513,
+      "step": 21880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5921051223101666e-05,
+      "loss": 0.4588,
+      "step": 21881
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5913661543900214e-05,
+      "loss": 0.4623,
+      "step": 21882
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5906272761390428e-05,
+      "loss": 0.3692,
+      "step": 21883
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5898884875661722e-05,
+      "loss": 0.4319,
+      "step": 21884
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5891497886803485e-05,
+      "loss": 0.4538,
+      "step": 21885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5884111794905196e-05,
+      "loss": 0.4879,
+      "step": 21886
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5876726600056223e-05,
+      "loss": 0.524,
+      "step": 21887
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5869342302345945e-05,
+      "loss": 0.3784,
+      "step": 21888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5861958901863727e-05,
+      "loss": 0.3816,
+      "step": 21889
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5854576398698916e-05,
+      "loss": 0.4949,
+      "step": 21890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.584719479294092e-05,
+      "loss": 0.4182,
+      "step": 21891
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5839814084679047e-05,
+      "loss": 0.3813,
+      "step": 21892
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5832434274002636e-05,
+      "loss": 0.3951,
+      "step": 21893
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5825055361001005e-05,
+      "loss": 0.4522,
+      "step": 21894
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.581767734576347e-05,
+      "loss": 0.3263,
+      "step": 21895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.581030022837928e-05,
+      "loss": 0.4168,
+      "step": 21896
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5802924008937802e-05,
+      "loss": 0.3728,
+      "step": 21897
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.579554868752827e-05,
+      "loss": 0.4247,
+      "step": 21898
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.578817426423996e-05,
+      "loss": 0.4457,
+      "step": 21899
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5780800739162125e-05,
+      "loss": 0.3835,
+      "step": 21900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.577342811238398e-05,
+      "loss": 0.465,
+      "step": 21901
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5766056383994808e-05,
+      "loss": 0.4139,
+      "step": 21902
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5758685554083804e-05,
+      "loss": 0.4544,
+      "step": 21903
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.575131562274019e-05,
+      "loss": 0.3698,
+      "step": 21904
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5743946590053168e-05,
+      "loss": 0.4791,
+      "step": 21905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.573657845611188e-05,
+      "loss": 0.3773,
+      "step": 21906
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5729211221005588e-05,
+      "loss": 0.5098,
+      "step": 21907
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.572184488482341e-05,
+      "loss": 0.4554,
+      "step": 21908
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.571447944765455e-05,
+      "loss": 0.425,
+      "step": 21909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.570711490958805e-05,
+      "loss": 0.436,
+      "step": 21910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5699751270713136e-05,
+      "loss": 0.4886,
+      "step": 21911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5692388531118906e-05,
+      "loss": 0.3934,
+      "step": 21912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.568502669089449e-05,
+      "loss": 0.3897,
+      "step": 21913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5677665750128964e-05,
+      "loss": 0.397,
+      "step": 21914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5670305708911402e-05,
+      "loss": 0.4966,
+      "step": 21915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.566294656733095e-05,
+      "loss": 0.4144,
+      "step": 21916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.565558832547664e-05,
+      "loss": 0.4662,
+      "step": 21917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.564823098343754e-05,
+      "loss": 0.4805,
+      "step": 21918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5640874541302696e-05,
+      "loss": 0.4235,
+      "step": 21919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.56335189991611e-05,
+      "loss": 0.3762,
+      "step": 21920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5626164357101857e-05,
+      "loss": 0.5094,
+      "step": 21921
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.561881061521395e-05,
+      "loss": 0.4293,
+      "step": 21922
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5611457773586378e-05,
+      "loss": 0.4927,
+      "step": 21923
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5604105832308135e-05,
+      "loss": 0.4146,
+      "step": 21924
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.559675479146818e-05,
+      "loss": 0.3693,
+      "step": 21925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5589404651155545e-05,
+      "loss": 0.4792,
+      "step": 21926
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.558205541145915e-05,
+      "loss": 0.4357,
+      "step": 21927
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.557470707246795e-05,
+      "loss": 0.4727,
+      "step": 21928
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5567359634270905e-05,
+      "loss": 0.4331,
+      "step": 21929
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5560013096956915e-05,
+      "loss": 0.4322,
+      "step": 21930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5552667460614877e-05,
+      "loss": 0.5147,
+      "step": 21931
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5545322725333755e-05,
+      "loss": 0.5406,
+      "step": 21932
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.553797889120243e-05,
+      "loss": 0.4513,
+      "step": 21933
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5530635958309768e-05,
+      "loss": 0.4305,
+      "step": 21934
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.552329392674465e-05,
+      "loss": 0.4053,
+      "step": 21935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.551595279659591e-05,
+      "loss": 0.4239,
+      "step": 21936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5508612567952462e-05,
+      "loss": 0.3918,
+      "step": 21937
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.550127324090312e-05,
+      "loss": 0.4662,
+      "step": 21938
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.54939348155367e-05,
+      "loss": 0.4156,
+      "step": 21939
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5486597291942038e-05,
+      "loss": 0.4051,
+      "step": 21940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5479260670207907e-05,
+      "loss": 0.455,
+      "step": 21941
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5471924950423152e-05,
+      "loss": 0.4577,
+      "step": 21942
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5464590132676557e-05,
+      "loss": 0.5221,
+      "step": 21943
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.545725621705687e-05,
+      "loss": 0.4887,
+      "step": 21944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.544992320365287e-05,
+      "loss": 0.4728,
+      "step": 21945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.544259109255328e-05,
+      "loss": 0.4622,
+      "step": 21946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5435259883846895e-05,
+      "loss": 0.4138,
+      "step": 21947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5427929577622433e-05,
+      "loss": 0.5043,
+      "step": 21948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.54206001739686e-05,
+      "loss": 0.4636,
+      "step": 21949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5413271672974114e-05,
+      "loss": 0.4451,
+      "step": 21950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.540594407472764e-05,
+      "loss": 0.4314,
+      "step": 21951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5398617379317923e-05,
+      "loss": 0.4329,
+      "step": 21952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5391291586833622e-05,
+      "loss": 0.4686,
+      "step": 21953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5383966697363394e-05,
+      "loss": 0.5127,
+      "step": 21954
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.53766427109959e-05,
+      "loss": 0.4511,
+      "step": 21955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5369319627819753e-05,
+      "loss": 0.3649,
+      "step": 21956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5361997447923647e-05,
+      "loss": 0.4619,
+      "step": 21957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5354676171396176e-05,
+      "loss": 0.4246,
+      "step": 21958
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5347355798325945e-05,
+      "loss": 0.5039,
+      "step": 21959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5340036328801554e-05,
+      "loss": 0.4203,
+      "step": 21960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5332717762911607e-05,
+      "loss": 0.4459,
+      "step": 21961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5325400100744645e-05,
+      "loss": 0.437,
+      "step": 21962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.531808334238929e-05,
+      "loss": 0.4539,
+      "step": 21963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5310767487934083e-05,
+      "loss": 0.3854,
+      "step": 21964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5303452537467553e-05,
+      "loss": 0.4793,
+      "step": 21965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5296138491078248e-05,
+      "loss": 0.4064,
+      "step": 21966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5288825348854662e-05,
+      "loss": 0.4557,
+      "step": 21967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.528151311088537e-05,
+      "loss": 0.4264,
+      "step": 21968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5274201777258834e-05,
+      "loss": 0.3926,
+      "step": 21969
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5266891348063547e-05,
+      "loss": 0.3759,
+      "step": 21970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5259581823388e-05,
+      "loss": 0.3934,
+      "step": 21971
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.525227320332062e-05,
+      "loss": 0.4339,
+      "step": 21972
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5244965487949935e-05,
+      "loss": 0.3758,
+      "step": 21973
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5237658677364363e-05,
+      "loss": 0.3854,
+      "step": 21974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.523035277165233e-05,
+      "loss": 0.4598,
+      "step": 21975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5223047770902274e-05,
+      "loss": 0.3649,
+      "step": 21976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5215743675202574e-05,
+      "loss": 0.4839,
+      "step": 21977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5208440484641705e-05,
+      "loss": 0.5148,
+      "step": 21978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5201138199308007e-05,
+      "loss": 0.5598,
+      "step": 21979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5193836819289885e-05,
+      "loss": 0.4989,
+      "step": 21980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5186536344675692e-05,
+      "loss": 0.5521,
+      "step": 21981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.517923677555377e-05,
+      "loss": 0.5153,
+      "step": 21982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5171938112012517e-05,
+      "loss": 0.3986,
+      "step": 21983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5164640354140256e-05,
+      "loss": 0.3471,
+      "step": 21984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.515734350202531e-05,
+      "loss": 0.4387,
+      "step": 21985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5150047555755986e-05,
+      "loss": 0.4259,
+      "step": 21986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5142752515420566e-05,
+      "loss": 0.4659,
+      "step": 21987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5135458381107403e-05,
+      "loss": 0.4791,
+      "step": 21988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5128165152904746e-05,
+      "loss": 0.5044,
+      "step": 21989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5120872830900878e-05,
+      "loss": 0.4026,
+      "step": 21990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5113581415184062e-05,
+      "loss": 0.3717,
+      "step": 21991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.51062909058425e-05,
+      "loss": 0.4482,
+      "step": 21992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.509900130296451e-05,
+      "loss": 0.4197,
+      "step": 21993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5091712606638273e-05,
+      "loss": 0.5122,
+      "step": 21994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5084424816952025e-05,
+      "loss": 0.4339,
+      "step": 21995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5077137933993956e-05,
+      "loss": 0.4435,
+      "step": 21996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5069851957852275e-05,
+      "loss": 0.3883,
+      "step": 21997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5062566888615136e-05,
+      "loss": 0.4619,
+      "step": 21998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5055282726370767e-05,
+      "loss": 0.4838,
+      "step": 21999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5047999471207294e-05,
+      "loss": 0.3565,
+      "step": 22000
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 0.42161640524864197,
+      "eval_runtime": 851.5803,
+      "eval_samples_per_second": 28.253,
+      "eval_steps_per_second": 14.127,
+      "step": 22000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5040717123212874e-05,
+      "loss": 0.4264,
+      "step": 22001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5033435682475658e-05,
+      "loss": 0.4162,
+      "step": 22002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.502615514908373e-05,
+      "loss": 0.4372,
+      "step": 22003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5018875523125285e-05,
+      "loss": 0.4356,
+      "step": 22004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5011596804688386e-05,
+      "loss": 0.4885,
+      "step": 22005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.500431899386114e-05,
+      "loss": 0.3985,
+      "step": 22006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4997042090731624e-05,
+      "loss": 0.4073,
+      "step": 22007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.498976609538789e-05,
+      "loss": 0.407,
+      "step": 22008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4982491007918063e-05,
+      "loss": 0.391,
+      "step": 22009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4975216828410153e-05,
+      "loss": 0.3698,
+      "step": 22010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4967943556952223e-05,
+      "loss": 0.3753,
+      "step": 22011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4960671193632278e-05,
+      "loss": 0.3813,
+      "step": 22012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.495339973853833e-05,
+      "loss": 0.4722,
+      "step": 22013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4946129191758438e-05,
+      "loss": 0.5809,
+      "step": 22014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4938859553380567e-05,
+      "loss": 0.4221,
+      "step": 22015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4931590823492712e-05,
+      "loss": 0.3981,
+      "step": 22016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4924323002182835e-05,
+      "loss": 0.4717,
+      "step": 22017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4917056089538893e-05,
+      "loss": 0.4377,
+      "step": 22018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4909790085648888e-05,
+      "loss": 0.4493,
+      "step": 22019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.490252499060073e-05,
+      "loss": 0.4538,
+      "step": 22020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4895260804482345e-05,
+      "loss": 0.4346,
+      "step": 22021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4887997527381668e-05,
+      "loss": 0.5659,
+      "step": 22022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4880735159386582e-05,
+      "loss": 0.4688,
+      "step": 22023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4873473700585028e-05,
+      "loss": 0.3798,
+      "step": 22024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4866213151064865e-05,
+      "loss": 0.3556,
+      "step": 22025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4858953510913996e-05,
+      "loss": 0.4048,
+      "step": 22026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4851694780220247e-05,
+      "loss": 0.4698,
+      "step": 22027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4844436959071514e-05,
+      "loss": 0.4513,
+      "step": 22028
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4837180047555576e-05,
+      "loss": 0.3548,
+      "step": 22029
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4829924045760355e-05,
+      "loss": 0.5028,
+      "step": 22030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.482266895377362e-05,
+      "loss": 0.4409,
+      "step": 22031
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4815414771683188e-05,
+      "loss": 0.4637,
+      "step": 22032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4808161499576865e-05,
+      "loss": 0.363,
+      "step": 22033
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4800909137542406e-05,
+      "loss": 0.3872,
+      "step": 22034
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.479365768566765e-05,
+      "loss": 0.5007,
+      "step": 22035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4786407144040334e-05,
+      "loss": 0.5181,
+      "step": 22036
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4779157512748207e-05,
+      "loss": 0.4057,
+      "step": 22037
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4771908791879028e-05,
+      "loss": 0.4936,
+      "step": 22038
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4764660981520482e-05,
+      "loss": 0.4568,
+      "step": 22039
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4757414081760366e-05,
+      "loss": 0.4649,
+      "step": 22040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4750168092686355e-05,
+      "loss": 0.4044,
+      "step": 22041
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4742923014386156e-05,
+      "loss": 0.5022,
+      "step": 22042
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4735678846947442e-05,
+      "loss": 0.4262,
+      "step": 22043
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.472843559045792e-05,
+      "loss": 0.4278,
+      "step": 22044
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4721193245005226e-05,
+      "loss": 0.3833,
+      "step": 22045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.471395181067705e-05,
+      "loss": 0.3684,
+      "step": 22046
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4706711287561003e-05,
+      "loss": 0.3342,
+      "step": 22047
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.469947167574471e-05,
+      "loss": 0.3679,
+      "step": 22048
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4692232975315855e-05,
+      "loss": 0.3789,
+      "step": 22049
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4684995186362015e-05,
+      "loss": 0.3741,
+      "step": 22050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.467775830897079e-05,
+      "loss": 0.3736,
+      "step": 22051
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4670522343229775e-05,
+      "loss": 0.4674,
+      "step": 22052
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.466328728922651e-05,
+      "loss": 0.4849,
+      "step": 22053
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.465605314704863e-05,
+      "loss": 0.3497,
+      "step": 22054
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4648819916783673e-05,
+      "loss": 0.4236,
+      "step": 22055
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.464158759851917e-05,
+      "loss": 0.433,
+      "step": 22056
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4634356192342655e-05,
+      "loss": 0.4082,
+      "step": 22057
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4627125698341624e-05,
+      "loss": 0.4797,
+      "step": 22058
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.461989611660367e-05,
+      "loss": 0.4148,
+      "step": 22059
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.461266744721623e-05,
+      "loss": 0.3978,
+      "step": 22060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.460543969026683e-05,
+      "loss": 0.4892,
+      "step": 22061
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4598212845842928e-05,
+      "loss": 0.3714,
+      "step": 22062
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4590986914032e-05,
+      "loss": 0.4184,
+      "step": 22063
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4583761894921463e-05,
+      "loss": 0.396,
+      "step": 22064
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.457653778859884e-05,
+      "loss": 0.4489,
+      "step": 22065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4569314595151526e-05,
+      "loss": 0.4283,
+      "step": 22066
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.456209231466696e-05,
+      "loss": 0.468,
+      "step": 22067
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4554870947232546e-05,
+      "loss": 0.4191,
+      "step": 22068
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4547650492935647e-05,
+      "loss": 0.3786,
+      "step": 22069
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4540430951863725e-05,
+      "loss": 0.4897,
+      "step": 22070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4533212324104127e-05,
+      "loss": 0.3767,
+      "step": 22071
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4525994609744228e-05,
+      "loss": 0.411,
+      "step": 22072
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4518777808871386e-05,
+      "loss": 0.4437,
+      "step": 22073
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.451156192157291e-05,
+      "loss": 0.5283,
+      "step": 22074
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.450434694793621e-05,
+      "loss": 0.3748,
+      "step": 22075
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.449713288804857e-05,
+      "loss": 0.4405,
+      "step": 22076
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4489919741997304e-05,
+      "loss": 0.4412,
+      "step": 22077
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.448270750986972e-05,
+      "loss": 0.4017,
+      "step": 22078
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4475496191753066e-05,
+      "loss": 0.4107,
+      "step": 22079
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.446828578773471e-05,
+      "loss": 0.3741,
+      "step": 22080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4461076297901873e-05,
+      "loss": 0.4093,
+      "step": 22081
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4453867722341816e-05,
+      "loss": 0.4407,
+      "step": 22082
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.444666006114179e-05,
+      "loss": 0.512,
+      "step": 22083
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4439453314389004e-05,
+      "loss": 0.4393,
+      "step": 22084
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4432247482170745e-05,
+      "loss": 0.4522,
+      "step": 22085
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4425042564574184e-05,
+      "loss": 0.3947,
+      "step": 22086
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4417838561686534e-05,
+      "loss": 0.4911,
+      "step": 22087
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4410635473594988e-05,
+      "loss": 0.5019,
+      "step": 22088
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.440343330038669e-05,
+      "loss": 0.475,
+      "step": 22089
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.439623204214889e-05,
+      "loss": 0.4491,
+      "step": 22090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.438903169896869e-05,
+      "loss": 0.4353,
+      "step": 22091
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4381832270933258e-05,
+      "loss": 0.4215,
+      "step": 22092
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4374633758129727e-05,
+      "loss": 0.4916,
+      "step": 22093
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4367436160645218e-05,
+      "loss": 0.4657,
+      "step": 22094
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4360239478566816e-05,
+      "loss": 0.5036,
+      "step": 22095
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4353043711981692e-05,
+      "loss": 0.4112,
+      "step": 22096
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4345848860976907e-05,
+      "loss": 0.3993,
+      "step": 22097
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4338654925639538e-05,
+      "loss": 0.3812,
+      "step": 22098
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.433146190605665e-05,
+      "loss": 0.5865,
+      "step": 22099
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.432426980231529e-05,
+      "loss": 0.4938,
+      "step": 22100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4317078614502553e-05,
+      "loss": 0.4671,
+      "step": 22101
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4309888342705454e-05,
+      "loss": 0.4151,
+      "step": 22102
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4302698987011008e-05,
+      "loss": 0.3522,
+      "step": 22103
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.429551054750624e-05,
+      "loss": 0.4095,
+      "step": 22104
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4288323024278125e-05,
+      "loss": 0.3466,
+      "step": 22105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.428113641741371e-05,
+      "loss": 0.4121,
+      "step": 22106
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.427395072699995e-05,
+      "loss": 0.4268,
+      "step": 22107
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4266765953123814e-05,
+      "loss": 0.4381,
+      "step": 22108
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.425958209587226e-05,
+      "loss": 0.3716,
+      "step": 22109
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4252399155332218e-05,
+      "loss": 0.4967,
+      "step": 22110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.424521713159067e-05,
+      "loss": 0.4699,
+      "step": 22111
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4238036024734523e-05,
+      "loss": 0.4267,
+      "step": 22112
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4230855834850685e-05,
+      "loss": 0.4,
+      "step": 22113
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4223676562026067e-05,
+      "loss": 0.4446,
+      "step": 22114
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4216498206347528e-05,
+      "loss": 0.4812,
+      "step": 22115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.420932076790201e-05,
+      "loss": 0.4247,
+      "step": 22116
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4202144246776358e-05,
+      "loss": 0.4761,
+      "step": 22117
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4194968643057426e-05,
+      "loss": 0.4254,
+      "step": 22118
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4187793956832073e-05,
+      "loss": 0.4195,
+      "step": 22119
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4180620188187087e-05,
+      "loss": 0.4402,
+      "step": 22120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4173447337209376e-05,
+      "loss": 0.396,
+      "step": 22121
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4166275403985718e-05,
+      "loss": 0.486,
+      "step": 22122
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4159104388602905e-05,
+      "loss": 0.4094,
+      "step": 22123
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.415193429114775e-05,
+      "loss": 0.3709,
+      "step": 22124
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.414476511170699e-05,
+      "loss": 0.4892,
+      "step": 22125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4137596850367472e-05,
+      "loss": 0.5367,
+      "step": 22126
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.41304295072159e-05,
+      "loss": 0.4108,
+      "step": 22127
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4123263082339055e-05,
+      "loss": 0.3904,
+      "step": 22128
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.411609757582365e-05,
+      "loss": 0.4021,
+      "step": 22129
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4108932987756416e-05,
+      "loss": 0.3623,
+      "step": 22130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4101769318224043e-05,
+      "loss": 0.3954,
+      "step": 22131
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4094606567313294e-05,
+      "loss": 0.3863,
+      "step": 22132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.408744473511083e-05,
+      "loss": 0.456,
+      "step": 22133
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4080283821703342e-05,
+      "loss": 0.3682,
+      "step": 22134
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.40731238271775e-05,
+      "loss": 0.4688,
+      "step": 22135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4065964751619918e-05,
+      "loss": 0.4985,
+      "step": 22136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4058806595117323e-05,
+      "loss": 0.4821,
+      "step": 22137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.40516493577563e-05,
+      "loss": 0.4676,
+      "step": 22138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4044493039623504e-05,
+      "loss": 0.5045,
+      "step": 22139
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4037337640805523e-05,
+      "loss": 0.3881,
+      "step": 22140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4030183161388954e-05,
+      "loss": 0.3749,
+      "step": 22141
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4023029601460447e-05,
+      "loss": 0.3852,
+      "step": 22142
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4015876961106533e-05,
+      "loss": 0.3942,
+      "step": 22143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.40087252404138e-05,
+      "loss": 0.5396,
+      "step": 22144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.400157443946881e-05,
+      "loss": 0.3807,
+      "step": 22145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3994424558358075e-05,
+      "loss": 0.5005,
+      "step": 22146
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.398727559716819e-05,
+      "loss": 0.4683,
+      "step": 22147
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3980127555985655e-05,
+      "loss": 0.4022,
+      "step": 22148
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3972980434896986e-05,
+      "loss": 0.4083,
+      "step": 22149
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3965834233988694e-05,
+      "loss": 0.4935,
+      "step": 22150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.395868895334722e-05,
+      "loss": 0.386,
+      "step": 22151
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3951544593059127e-05,
+      "loss": 0.4004,
+      "step": 22152
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.394440115321085e-05,
+      "loss": 0.4662,
+      "step": 22153
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.393725863388885e-05,
+      "loss": 0.4467,
+      "step": 22154
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3930117035179555e-05,
+      "loss": 0.4044,
+      "step": 22155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.39229763571694e-05,
+      "loss": 0.458,
+      "step": 22156
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3915836599944862e-05,
+      "loss": 0.4773,
+      "step": 22157
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3908697763592323e-05,
+      "loss": 0.4432,
+      "step": 22158
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3901559848198186e-05,
+      "loss": 0.4166,
+      "step": 22159
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.389442285384885e-05,
+      "loss": 0.3615,
+      "step": 22160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.388728678063068e-05,
+      "loss": 0.4162,
+      "step": 22161
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3880151628630043e-05,
+      "loss": 0.4124,
+      "step": 22162
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3873017397933327e-05,
+      "loss": 0.4135,
+      "step": 22163
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3865884088626878e-05,
+      "loss": 0.4079,
+      "step": 22164
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.385875170079702e-05,
+      "loss": 0.3887,
+      "step": 22165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.385162023453007e-05,
+      "loss": 0.4224,
+      "step": 22166
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3844489689912332e-05,
+      "loss": 0.4329,
+      "step": 22167
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3837360067030156e-05,
+      "loss": 0.4061,
+      "step": 22168
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3830231365969813e-05,
+      "loss": 0.5168,
+      "step": 22169
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3823103586817562e-05,
+      "loss": 0.4133,
+      "step": 22170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3815976729659705e-05,
+      "loss": 0.4313,
+      "step": 22171
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.380885079458245e-05,
+      "loss": 0.4136,
+      "step": 22172
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3801725781672112e-05,
+      "loss": 0.4174,
+      "step": 22173
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3794601691014896e-05,
+      "loss": 0.4067,
+      "step": 22174
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3787478522697027e-05,
+      "loss": 0.3292,
+      "step": 22175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3780356276804705e-05,
+      "loss": 0.3952,
+      "step": 22176
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.377323495342413e-05,
+      "loss": 0.4986,
+      "step": 22177
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3766114552641572e-05,
+      "loss": 0.5435,
+      "step": 22178
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.375899507454311e-05,
+      "loss": 0.3396,
+      "step": 22179
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3751876519214954e-05,
+      "loss": 0.5047,
+      "step": 22180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3744758886743246e-05,
+      "loss": 0.4255,
+      "step": 22181
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.373764217721416e-05,
+      "loss": 0.4888,
+      "step": 22182
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3730526390713825e-05,
+      "loss": 0.4012,
+      "step": 22183
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3723411527328365e-05,
+      "loss": 0.4043,
+      "step": 22184
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3716297587143886e-05,
+      "loss": 0.4493,
+      "step": 22185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.370918457024648e-05,
+      "loss": 0.4301,
+      "step": 22186
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.370207247672226e-05,
+      "loss": 0.5508,
+      "step": 22187
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3694961306657314e-05,
+      "loss": 0.4464,
+      "step": 22188
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.368785106013769e-05,
+      "loss": 0.5049,
+      "step": 22189
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3680741737249445e-05,
+      "loss": 0.4949,
+      "step": 22190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3673633338078605e-05,
+      "loss": 0.4452,
+      "step": 22191
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3666525862711262e-05,
+      "loss": 0.5788,
+      "step": 22192
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.365941931123341e-05,
+      "loss": 0.4394,
+      "step": 22193
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3652313683731066e-05,
+      "loss": 0.4384,
+      "step": 22194
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3645208980290224e-05,
+      "loss": 0.4009,
+      "step": 22195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3638105200996887e-05,
+      "loss": 0.4528,
+      "step": 22196
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.363100234593698e-05,
+      "loss": 0.4748,
+      "step": 22197
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3623900415196553e-05,
+      "loss": 0.4725,
+      "step": 22198
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3616799408861534e-05,
+      "loss": 0.5237,
+      "step": 22199
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3609699327017855e-05,
+      "loss": 0.4634,
+      "step": 22200
+    },
+    {
+      "epoch": 0.78,
+      "eval_loss": 0.41962045431137085,
+      "eval_runtime": 854.3091,
+      "eval_samples_per_second": 28.163,
+      "eval_steps_per_second": 14.082,
+      "step": 22200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3602600169751454e-05,
+      "loss": 0.499,
+      "step": 22201
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3595501937148234e-05,
+      "loss": 0.5354,
+      "step": 22202
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.358840462929416e-05,
+      "loss": 0.3844,
+      "step": 22203
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3581308246275103e-05,
+      "loss": 0.4055,
+      "step": 22204
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3574212788176953e-05,
+      "loss": 0.4829,
+      "step": 22205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3567118255085584e-05,
+      "loss": 0.4392,
+      "step": 22206
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3560024647086853e-05,
+      "loss": 0.4588,
+      "step": 22207
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3552931964266655e-05,
+      "loss": 0.4606,
+      "step": 22208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3545840206710812e-05,
+      "loss": 0.3849,
+      "step": 22209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3538749374505165e-05,
+      "loss": 0.408,
+      "step": 22210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3531659467735522e-05,
+      "loss": 0.4827,
+      "step": 22211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3524570486487684e-05,
+      "loss": 0.3984,
+      "step": 22212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3517482430847493e-05,
+      "loss": 0.4443,
+      "step": 22213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.351039530090071e-05,
+      "loss": 0.4833,
+      "step": 22214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3503309096733118e-05,
+      "loss": 0.3936,
+      "step": 22215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3496223818430495e-05,
+      "loss": 0.4337,
+      "step": 22216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3489139466078546e-05,
+      "loss": 0.4286,
+      "step": 22217
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.348205603976309e-05,
+      "loss": 0.4322,
+      "step": 22218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.347497353956982e-05,
+      "loss": 0.4182,
+      "step": 22219
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3467891965584466e-05,
+      "loss": 0.4279,
+      "step": 22220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.346081131789274e-05,
+      "loss": 0.4837,
+      "step": 22221
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3453731596580297e-05,
+      "loss": 0.42,
+      "step": 22222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3446652801732892e-05,
+      "loss": 0.5234,
+      "step": 22223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.343957493343618e-05,
+      "loss": 0.3547,
+      "step": 22224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3432497991775826e-05,
+      "loss": 0.4437,
+      "step": 22225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3425421976837482e-05,
+      "loss": 0.5307,
+      "step": 22226
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3418346888706756e-05,
+      "loss": 0.4008,
+      "step": 22227
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3411272727469335e-05,
+      "loss": 0.4463,
+      "step": 22228
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.340419949321083e-05,
+      "loss": 0.4336,
+      "step": 22229
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.339712718601683e-05,
+      "loss": 0.4214,
+      "step": 22230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3390055805972956e-05,
+      "loss": 0.4351,
+      "step": 22231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.338298535316478e-05,
+      "loss": 0.3954,
+      "step": 22232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.337591582767784e-05,
+      "loss": 0.3942,
+      "step": 22233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3368847229597778e-05,
+      "loss": 0.4393,
+      "step": 22234
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.336177955901011e-05,
+      "loss": 0.4501,
+      "step": 22235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.335471281600037e-05,
+      "loss": 0.4167,
+      "step": 22236
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3347647000654104e-05,
+      "loss": 0.461,
+      "step": 22237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3340582113056807e-05,
+      "loss": 0.4093,
+      "step": 22238
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.333351815329402e-05,
+      "loss": 0.4151,
+      "step": 22239
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.332645512145124e-05,
+      "loss": 0.4234,
+      "step": 22240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.331939301761393e-05,
+      "loss": 0.4148,
+      "step": 22241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3312331841867586e-05,
+      "loss": 0.4139,
+      "step": 22242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3305271594297627e-05,
+      "loss": 0.4019,
+      "step": 22243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.329821227498956e-05,
+      "loss": 0.4504,
+      "step": 22244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3291153884028815e-05,
+      "loss": 0.4142,
+      "step": 22245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3284096421500822e-05,
+      "loss": 0.4376,
+      "step": 22246
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3277039887490982e-05,
+      "loss": 0.4707,
+      "step": 22247
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3269984282084677e-05,
+      "loss": 0.4469,
+      "step": 22248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.326292960536738e-05,
+      "loss": 0.399,
+      "step": 22249
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3255875857424416e-05,
+      "loss": 0.3353,
+      "step": 22250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3248823038341193e-05,
+      "loss": 0.4629,
+      "step": 22251
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.324177114820305e-05,
+      "loss": 0.5102,
+      "step": 22252
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3234720187095315e-05,
+      "loss": 0.4105,
+      "step": 22253
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3227670155103387e-05,
+      "loss": 0.4109,
+      "step": 22254
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3220621052312575e-05,
+      "loss": 0.5209,
+      "step": 22255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3213572878808186e-05,
+      "loss": 0.3655,
+      "step": 22256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3206525634675523e-05,
+      "loss": 0.4279,
+      "step": 22257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3199479319999862e-05,
+      "loss": 0.3569,
+      "step": 22258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3192433934866542e-05,
+      "loss": 0.5243,
+      "step": 22259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3185389479360807e-05,
+      "loss": 0.3896,
+      "step": 22260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3178345953567914e-05,
+      "loss": 0.4094,
+      "step": 22261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3171303357573117e-05,
+      "loss": 0.396,
+      "step": 22262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3164261691461653e-05,
+      "loss": 0.3652,
+      "step": 22263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3157220955318727e-05,
+      "loss": 0.4656,
+      "step": 22264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.31501811492296e-05,
+      "loss": 0.4026,
+      "step": 22265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.314314227327946e-05,
+      "loss": 0.4636,
+      "step": 22266
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3136104327553497e-05,
+      "loss": 0.4531,
+      "step": 22267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3129067312136887e-05,
+      "loss": 0.3622,
+      "step": 22268
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3122031227114793e-05,
+      "loss": 0.4342,
+      "step": 22269
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3114996072572404e-05,
+      "loss": 0.4354,
+      "step": 22270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3107961848594862e-05,
+      "loss": 0.3673,
+      "step": 22271
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.31009285552673e-05,
+      "loss": 0.4199,
+      "step": 22272
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3093896192674837e-05,
+      "loss": 0.419,
+      "step": 22273
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.308686476090257e-05,
+      "loss": 0.436,
+      "step": 22274
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3079834260035647e-05,
+      "loss": 0.3701,
+      "step": 22275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3072804690159143e-05,
+      "loss": 0.4096,
+      "step": 22276
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3065776051358135e-05,
+      "loss": 0.3866,
+      "step": 22277
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3058748343717694e-05,
+      "loss": 0.4411,
+      "step": 22278
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3051721567322846e-05,
+      "loss": 0.4037,
+      "step": 22279
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3044695722258702e-05,
+      "loss": 0.5081,
+      "step": 22280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3037670808610258e-05,
+      "loss": 0.3629,
+      "step": 22281
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3030646826462554e-05,
+      "loss": 0.437,
+      "step": 22282
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3023623775900605e-05,
+      "loss": 0.4246,
+      "step": 22283
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.301660165700936e-05,
+      "loss": 0.4197,
+      "step": 22284
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.300958046987389e-05,
+      "loss": 0.3846,
+      "step": 22285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.300256021457915e-05,
+      "loss": 0.4265,
+      "step": 22286
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.299554089121009e-05,
+      "loss": 0.4511,
+      "step": 22287
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.298852249985167e-05,
+      "loss": 0.407,
+      "step": 22288
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2981505040588823e-05,
+      "loss": 0.4702,
+      "step": 22289
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.297448851350653e-05,
+      "loss": 0.4683,
+      "step": 22290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.296747291868968e-05,
+      "loss": 0.4268,
+      "step": 22291
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2960458256223204e-05,
+      "loss": 0.5102,
+      "step": 22292
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.295344452619198e-05,
+      "loss": 0.3682,
+      "step": 22293
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2946431728680885e-05,
+      "loss": 0.4128,
+      "step": 22294
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2939419863774846e-05,
+      "loss": 0.4513,
+      "step": 22295
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2932408931558703e-05,
+      "loss": 0.5332,
+      "step": 22296
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2925398932117316e-05,
+      "loss": 0.4267,
+      "step": 22297
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2918389865535515e-05,
+      "loss": 0.5086,
+      "step": 22298
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.291138173189816e-05,
+      "loss": 0.3898,
+      "step": 22299
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2904374531290018e-05,
+      "loss": 0.4645,
+      "step": 22300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.289736826379597e-05,
+      "loss": 0.4968,
+      "step": 22301
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2890362929500786e-05,
+      "loss": 0.4365,
+      "step": 22302
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.288335852848925e-05,
+      "loss": 0.4039,
+      "step": 22303
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2876355060846144e-05,
+      "loss": 0.4539,
+      "step": 22304
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2869352526656196e-05,
+      "loss": 0.5118,
+      "step": 22305
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2862350926004227e-05,
+      "loss": 0.3915,
+      "step": 22306
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.285535025897495e-05,
+      "loss": 0.4335,
+      "step": 22307
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2848350525653085e-05,
+      "loss": 0.4084,
+      "step": 22308
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2841351726123362e-05,
+      "loss": 0.4034,
+      "step": 22309
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.283435386047046e-05,
+      "loss": 0.4257,
+      "step": 22310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.282735692877912e-05,
+      "loss": 0.3951,
+      "step": 22311
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2820360931134032e-05,
+      "loss": 0.4541,
+      "step": 22312
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.281336586761986e-05,
+      "loss": 0.3949,
+      "step": 22313
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2806371738321218e-05,
+      "loss": 0.47,
+      "step": 22314
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2799378543322823e-05,
+      "loss": 0.3951,
+      "step": 22315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2792386282709288e-05,
+      "loss": 0.4419,
+      "step": 22316
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2785394956565255e-05,
+      "loss": 0.4049,
+      "step": 22317
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.277840456497533e-05,
+      "loss": 0.4289,
+      "step": 22318
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2771415108024085e-05,
+      "loss": 0.3592,
+      "step": 22319
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2764426585796195e-05,
+      "loss": 0.3941,
+      "step": 22320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.27574389983762e-05,
+      "loss": 0.425,
+      "step": 22321
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2750452345848682e-05,
+      "loss": 0.4108,
+      "step": 22322
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2743466628298195e-05,
+      "loss": 0.4866,
+      "step": 22323
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2736481845809265e-05,
+      "loss": 0.449,
+      "step": 22324
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.272949799846649e-05,
+      "loss": 0.376,
+      "step": 22325
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2722515086354358e-05,
+      "loss": 0.5186,
+      "step": 22326
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2715533109557406e-05,
+      "loss": 0.3937,
+      "step": 22327
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2708552068160115e-05,
+      "loss": 0.4248,
+      "step": 22328
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.270157196224697e-05,
+      "loss": 0.4795,
+      "step": 22329
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2694592791902503e-05,
+      "loss": 0.4718,
+      "step": 22330
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2687614557211156e-05,
+      "loss": 0.4822,
+      "step": 22331
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2680637258257387e-05,
+      "loss": 0.3718,
+      "step": 22332
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.267366089512565e-05,
+      "loss": 0.456,
+      "step": 22333
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2666685467900383e-05,
+      "loss": 0.4299,
+      "step": 22334
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.265971097666597e-05,
+      "loss": 0.5339,
+      "step": 22335
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2652737421506897e-05,
+      "loss": 0.4481,
+      "step": 22336
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.264576480250754e-05,
+      "loss": 0.4205,
+      "step": 22337
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.263879311975229e-05,
+      "loss": 0.4669,
+      "step": 22338
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2631822373325505e-05,
+      "loss": 0.424,
+      "step": 22339
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.262485256331156e-05,
+      "loss": 0.4077,
+      "step": 22340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.261788368979485e-05,
+      "loss": 0.4191,
+      "step": 22341
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2610915752859684e-05,
+      "loss": 0.4427,
+      "step": 22342
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.260394875259042e-05,
+      "loss": 0.3855,
+      "step": 22343
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2596982689071377e-05,
+      "loss": 0.4304,
+      "step": 22344
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2590017562386833e-05,
+      "loss": 0.3848,
+      "step": 22345
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2583053372621142e-05,
+      "loss": 0.4706,
+      "step": 22346
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2576090119858572e-05,
+      "loss": 0.4204,
+      "step": 22347
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2569127804183398e-05,
+      "loss": 0.4155,
+      "step": 22348
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2562166425679886e-05,
+      "loss": 0.4229,
+      "step": 22349
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.255520598443228e-05,
+      "loss": 0.4543,
+      "step": 22350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.254824648052486e-05,
+      "loss": 0.4774,
+      "step": 22351
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.254128791404184e-05,
+      "loss": 0.4971,
+      "step": 22352
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2534330285067438e-05,
+      "loss": 0.386,
+      "step": 22353
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2527373593685874e-05,
+      "loss": 0.4037,
+      "step": 22354
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2520417839981312e-05,
+      "loss": 0.4759,
+      "step": 22355
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2513463024037985e-05,
+      "loss": 0.389,
+      "step": 22356
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.250650914594007e-05,
+      "loss": 0.4549,
+      "step": 22357
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.24995562057717e-05,
+      "loss": 0.5069,
+      "step": 22358
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.249260420361705e-05,
+      "loss": 0.4006,
+      "step": 22359
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2485653139560227e-05,
+      "loss": 0.4277,
+      "step": 22360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.247870301368542e-05,
+      "loss": 0.4188,
+      "step": 22361
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2471753826076724e-05,
+      "loss": 0.4061,
+      "step": 22362
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.246480557681824e-05,
+      "loss": 0.3873,
+      "step": 22363
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.245785826599407e-05,
+      "loss": 0.4402,
+      "step": 22364
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2450911893688297e-05,
+      "loss": 0.4898,
+      "step": 22365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2443966459984966e-05,
+      "loss": 0.4342,
+      "step": 22366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2437021964968197e-05,
+      "loss": 0.4052,
+      "step": 22367
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.243007840872202e-05,
+      "loss": 0.3787,
+      "step": 22368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2423135791330463e-05,
+      "loss": 0.3909,
+      "step": 22369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2416194112877563e-05,
+      "loss": 0.4373,
+      "step": 22370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2409253373447313e-05,
+      "loss": 0.4499,
+      "step": 22371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.240231357312377e-05,
+      "loss": 0.378,
+      "step": 22372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2395374711990892e-05,
+      "loss": 0.4356,
+      "step": 22373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.238843679013267e-05,
+      "loss": 0.3592,
+      "step": 22374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2381499807633087e-05,
+      "loss": 0.3894,
+      "step": 22375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.237456376457606e-05,
+      "loss": 0.4623,
+      "step": 22376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2367628661045602e-05,
+      "loss": 0.4108,
+      "step": 22377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2360694497125623e-05,
+      "loss": 0.4088,
+      "step": 22378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2353761272900052e-05,
+      "loss": 0.4932,
+      "step": 22379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2346828988452796e-05,
+      "loss": 0.4093,
+      "step": 22380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2339897643867725e-05,
+      "loss": 0.3873,
+      "step": 22381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2332967239228807e-05,
+      "loss": 0.5007,
+      "step": 22382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2326037774619878e-05,
+      "loss": 0.4356,
+      "step": 22383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2319109250124813e-05,
+      "loss": 0.5157,
+      "step": 22384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.231218166582748e-05,
+      "loss": 0.4044,
+      "step": 22385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2305255021811676e-05,
+      "loss": 0.4692,
+      "step": 22386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2298329318161315e-05,
+      "loss": 0.4475,
+      "step": 22387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2291404554960184e-05,
+      "loss": 0.4288,
+      "step": 22388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.22844807322921e-05,
+      "loss": 0.4407,
+      "step": 22389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2277557850240848e-05,
+      "loss": 0.4429,
+      "step": 22390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2270635908890214e-05,
+      "loss": 0.3914,
+      "step": 22391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2263714908324017e-05,
+      "loss": 0.512,
+      "step": 22392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2256794848625996e-05,
+      "loss": 0.427,
+      "step": 22393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.224987572987991e-05,
+      "loss": 0.5323,
+      "step": 22394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.224295755216951e-05,
+      "loss": 0.4876,
+      "step": 22395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2236040315578487e-05,
+      "loss": 0.4121,
+      "step": 22396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.222912402019063e-05,
+      "loss": 0.443,
+      "step": 22397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2222208666089617e-05,
+      "loss": 0.3308,
+      "step": 22398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2215294253359143e-05,
+      "loss": 0.3775,
+      "step": 22399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2208380782082906e-05,
+      "loss": 0.3925,
+      "step": 22400
+    },
+    {
+      "epoch": 0.78,
+      "eval_loss": 0.41787704825401306,
+      "eval_runtime": 846.078,
+      "eval_samples_per_second": 28.437,
+      "eval_steps_per_second": 14.219,
+      "step": 22400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2201468252344583e-05,
+      "loss": 0.4442,
+      "step": 22401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2194556664227793e-05,
+      "loss": 0.4474,
+      "step": 22402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2187646017816254e-05,
+      "loss": 0.5167,
+      "step": 22403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2180736313193584e-05,
+      "loss": 0.3826,
+      "step": 22404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2173827550443417e-05,
+      "loss": 0.4519,
+      "step": 22405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.216691972964936e-05,
+      "loss": 0.5189,
+      "step": 22406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2160012850895008e-05,
+      "loss": 0.4629,
+      "step": 22407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2153106914263998e-05,
+      "loss": 0.4608,
+      "step": 22408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2146201919839892e-05,
+      "loss": 0.4911,
+      "step": 22409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2139297867706265e-05,
+      "loss": 0.4332,
+      "step": 22410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2132394757946672e-05,
+      "loss": 0.4254,
+      "step": 22411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2125492590644658e-05,
+      "loss": 0.4626,
+      "step": 22412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.211859136588379e-05,
+      "loss": 0.4034,
+      "step": 22413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2111691083747587e-05,
+      "loss": 0.421,
+      "step": 22414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2104791744319554e-05,
+      "loss": 0.5044,
+      "step": 22415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2097893347683198e-05,
+      "loss": 0.4154,
+      "step": 22416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2090995893921996e-05,
+      "loss": 0.4324,
+      "step": 22417
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.208409938311947e-05,
+      "loss": 0.4472,
+      "step": 22418
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2077203815359073e-05,
+      "loss": 0.3626,
+      "step": 22419
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.207030919072426e-05,
+      "loss": 0.42,
+      "step": 22420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.206341550929849e-05,
+      "loss": 0.4229,
+      "step": 22421
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2056522771165156e-05,
+      "loss": 0.3854,
+      "step": 22422
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.204963097640774e-05,
+      "loss": 0.4081,
+      "step": 22423
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.204274012510964e-05,
+      "loss": 0.4157,
+      "step": 22424
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2035850217354238e-05,
+      "loss": 0.5074,
+      "step": 22425
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2028961253224957e-05,
+      "loss": 0.4348,
+      "step": 22426
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2022073232805118e-05,
+      "loss": 0.4597,
+      "step": 22427
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2015186156178158e-05,
+      "loss": 0.3788,
+      "step": 22428
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2008300023427397e-05,
+      "loss": 0.3502,
+      "step": 22429
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2001414834636204e-05,
+      "loss": 0.4456,
+      "step": 22430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1994530589887886e-05,
+      "loss": 0.4221,
+      "step": 22431
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1987647289265768e-05,
+      "loss": 0.5028,
+      "step": 22432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.198076493285315e-05,
+      "loss": 0.3418,
+      "step": 22433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1973883520733372e-05,
+      "loss": 0.4208,
+      "step": 22434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.196700305298971e-05,
+      "loss": 0.4948,
+      "step": 22435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1960123529705423e-05,
+      "loss": 0.4194,
+      "step": 22436
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.195324495096378e-05,
+      "loss": 0.4587,
+      "step": 22437
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1946367316848016e-05,
+      "loss": 0.4601,
+      "step": 22438
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1939490627441416e-05,
+      "loss": 0.3776,
+      "step": 22439
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1932614882827197e-05,
+      "loss": 0.4632,
+      "step": 22440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1925740083088564e-05,
+      "loss": 0.369,
+      "step": 22441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1918866228308744e-05,
+      "loss": 0.4532,
+      "step": 22442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1911993318570888e-05,
+      "loss": 0.5223,
+      "step": 22443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1905121353958235e-05,
+      "loss": 0.437,
+      "step": 22444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1898250334553948e-05,
+      "loss": 0.3655,
+      "step": 22445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1891380260441185e-05,
+      "loss": 0.3594,
+      "step": 22446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.188451113170309e-05,
+      "loss": 0.5024,
+      "step": 22447
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1877642948422795e-05,
+      "loss": 0.3749,
+      "step": 22448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1870775710683444e-05,
+      "loss": 0.4532,
+      "step": 22449
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.186390941856815e-05,
+      "loss": 0.5046,
+      "step": 22450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1857044072160005e-05,
+      "loss": 0.4619,
+      "step": 22451
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1850179671542092e-05,
+      "loss": 0.3687,
+      "step": 22452
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.184331621679754e-05,
+      "loss": 0.3934,
+      "step": 22453
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1836453708009397e-05,
+      "loss": 0.4007,
+      "step": 22454
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1829592145260712e-05,
+      "loss": 0.3468,
+      "step": 22455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1822731528634545e-05,
+      "loss": 0.3922,
+      "step": 22456
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1815871858213887e-05,
+      "loss": 0.3766,
+      "step": 22457
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1809013134081846e-05,
+      "loss": 0.4873,
+      "step": 22458
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1802155356321373e-05,
+      "loss": 0.3926,
+      "step": 22459
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1795298525015506e-05,
+      "loss": 0.4071,
+      "step": 22460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.178844264024721e-05,
+      "loss": 0.3686,
+      "step": 22461
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.178158770209945e-05,
+      "loss": 0.4554,
+      "step": 22462
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.177473371065524e-05,
+      "loss": 0.4888,
+      "step": 22463
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1767880665997508e-05,
+      "loss": 0.482,
+      "step": 22464
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1761028568209207e-05,
+      "loss": 0.4363,
+      "step": 22465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.175417741737327e-05,
+      "loss": 0.3961,
+      "step": 22466
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.174732721357261e-05,
+      "loss": 0.4572,
+      "step": 22467
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1740477956890115e-05,
+      "loss": 0.4578,
+      "step": 22468
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1733629647408736e-05,
+      "loss": 0.4062,
+      "step": 22469
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1726782285211346e-05,
+      "loss": 0.457,
+      "step": 22470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.17199358703808e-05,
+      "loss": 0.5065,
+      "step": 22471
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.171309040299998e-05,
+      "loss": 0.4255,
+      "step": 22472
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1706245883151698e-05,
+      "loss": 0.3735,
+      "step": 22473
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.169940231091886e-05,
+      "loss": 0.4642,
+      "step": 22474
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1692559686384263e-05,
+      "loss": 0.373,
+      "step": 22475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1685718009630728e-05,
+      "loss": 0.4291,
+      "step": 22476
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.167887728074106e-05,
+      "loss": 0.4296,
+      "step": 22477
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1672037499798024e-05,
+      "loss": 0.4461,
+      "step": 22478
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1665198666884477e-05,
+      "loss": 0.3867,
+      "step": 22479
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.165836078208314e-05,
+      "loss": 0.4336,
+      "step": 22480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1651523845476796e-05,
+      "loss": 0.3939,
+      "step": 22481
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.164468785714817e-05,
+      "loss": 0.4019,
+      "step": 22482
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1637852817180003e-05,
+      "loss": 0.3935,
+      "step": 22483
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.163101872565505e-05,
+      "loss": 0.433,
+      "step": 22484
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.162418558265602e-05,
+      "loss": 0.4331,
+      "step": 22485
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1617353388265594e-05,
+      "loss": 0.4636,
+      "step": 22486
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.161052214256648e-05,
+      "loss": 0.4905,
+      "step": 22487
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1603691845641326e-05,
+      "loss": 0.4467,
+      "step": 22488
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.159686249757287e-05,
+      "loss": 0.4278,
+      "step": 22489
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1590034098443722e-05,
+      "loss": 0.4304,
+      "step": 22490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.158320664833654e-05,
+      "loss": 0.3458,
+      "step": 22491
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1576380147333952e-05,
+      "loss": 0.4225,
+      "step": 22492
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.156955459551856e-05,
+      "loss": 0.4727,
+      "step": 22493
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.156272999297303e-05,
+      "loss": 0.4163,
+      "step": 22494
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.155590633977993e-05,
+      "loss": 0.4297,
+      "step": 22495
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1549083636021848e-05,
+      "loss": 0.388,
+      "step": 22496
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1542261881781378e-05,
+      "loss": 0.4405,
+      "step": 22497
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1535441077141048e-05,
+      "loss": 0.4409,
+      "step": 22498
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1528621222183454e-05,
+      "loss": 0.3936,
+      "step": 22499
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1521802316991137e-05,
+      "loss": 0.4139,
+      "step": 22500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1514984361646605e-05,
+      "loss": 0.4346,
+      "step": 22501
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.150816735623239e-05,
+      "loss": 0.4297,
+      "step": 22502
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1501351300830995e-05,
+      "loss": 0.5668,
+      "step": 22503
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1494536195524905e-05,
+      "loss": 0.4194,
+      "step": 22504
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1487722040396642e-05,
+      "loss": 0.4379,
+      "step": 22505
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1480908835528656e-05,
+      "loss": 0.476,
+      "step": 22506
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1474096581003413e-05,
+      "loss": 0.379,
+      "step": 22507
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.146728527690337e-05,
+      "loss": 0.4514,
+      "step": 22508
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1460474923310925e-05,
+      "loss": 0.4332,
+      "step": 22509
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1453665520308574e-05,
+      "loss": 0.4375,
+      "step": 22510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.14468570679787e-05,
+      "loss": 0.4192,
+      "step": 22511
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1440049566403697e-05,
+      "loss": 0.4364,
+      "step": 22512
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1433243015665984e-05,
+      "loss": 0.428,
+      "step": 22513
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.14264374158479e-05,
+      "loss": 0.4454,
+      "step": 22514
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.141963276703186e-05,
+      "loss": 0.457,
+      "step": 22515
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1412829069300223e-05,
+      "loss": 0.3689,
+      "step": 22516
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1406026322735307e-05,
+      "loss": 0.4675,
+      "step": 22517
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.139922452741947e-05,
+      "loss": 0.4268,
+      "step": 22518
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1392423683435003e-05,
+      "loss": 0.4884,
+      "step": 22519
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.138562379086426e-05,
+      "loss": 0.4149,
+      "step": 22520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1378824849789536e-05,
+      "loss": 0.5279,
+      "step": 22521
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1372026860293115e-05,
+      "loss": 0.4355,
+      "step": 22522
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1365229822457255e-05,
+      "loss": 0.3751,
+      "step": 22523
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1358433736364226e-05,
+      "loss": 0.5773,
+      "step": 22524
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1351638602096313e-05,
+      "loss": 0.4661,
+      "step": 22525
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1344844419735755e-05,
+      "loss": 0.4083,
+      "step": 22526
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.133805118936477e-05,
+      "loss": 0.4225,
+      "step": 22527
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1331258911065577e-05,
+      "loss": 0.4229,
+      "step": 22528
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.132446758492036e-05,
+      "loss": 0.4429,
+      "step": 22529
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.131767721101138e-05,
+      "loss": 0.4233,
+      "step": 22530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1310887789420786e-05,
+      "loss": 0.4211,
+      "step": 22531
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1304099320230752e-05,
+      "loss": 0.4129,
+      "step": 22532
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1297311803523457e-05,
+      "loss": 0.4577,
+      "step": 22533
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1290525239381032e-05,
+      "loss": 0.4141,
+      "step": 22534
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1283739627885602e-05,
+      "loss": 0.5678,
+      "step": 22535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1276954969119334e-05,
+      "loss": 0.3993,
+      "step": 22536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.127017126316434e-05,
+      "loss": 0.4576,
+      "step": 22537
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1263388510102724e-05,
+      "loss": 0.3604,
+      "step": 22538
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1256606710016558e-05,
+      "loss": 0.418,
+      "step": 22539
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1249825862987917e-05,
+      "loss": 0.4364,
+      "step": 22540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.124304596909892e-05,
+      "loss": 0.333,
+      "step": 22541
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.12362670284316e-05,
+      "loss": 0.3877,
+      "step": 22542
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1229489041067997e-05,
+      "loss": 0.3564,
+      "step": 22543
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1222712007090163e-05,
+      "loss": 0.4396,
+      "step": 22544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.121593592658008e-05,
+      "loss": 0.4831,
+      "step": 22545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.120916079961983e-05,
+      "loss": 0.4022,
+      "step": 22546
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.120238662629137e-05,
+      "loss": 0.354,
+      "step": 22547
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1195613406676706e-05,
+      "loss": 0.3935,
+      "step": 22548
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1188841140857806e-05,
+      "loss": 0.3907,
+      "step": 22549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1182069828916617e-05,
+      "loss": 0.4617,
+      "step": 22550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.117529947093515e-05,
+      "loss": 0.4122,
+      "step": 22551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1168530066995306e-05,
+      "loss": 0.5126,
+      "step": 22552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1161761617179042e-05,
+      "loss": 0.333,
+      "step": 22553
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1154994121568263e-05,
+      "loss": 0.3843,
+      "step": 22554
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.114822758024485e-05,
+      "loss": 0.4183,
+      "step": 22555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1141461993290757e-05,
+      "loss": 0.3759,
+      "step": 22556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1134697360787847e-05,
+      "loss": 0.3916,
+      "step": 22557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.112793368281799e-05,
+      "loss": 0.4475,
+      "step": 22558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1121170959463054e-05,
+      "loss": 0.4161,
+      "step": 22559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.111440919080486e-05,
+      "loss": 0.407,
+      "step": 22560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1107648376925317e-05,
+      "loss": 0.4386,
+      "step": 22561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1100888517906194e-05,
+      "loss": 0.3626,
+      "step": 22562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1094129613829328e-05,
+      "loss": 0.4358,
+      "step": 22563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1087371664776534e-05,
+      "loss": 0.4247,
+      "step": 22564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.108061467082958e-05,
+      "loss": 0.4324,
+      "step": 22565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1073858632070243e-05,
+      "loss": 0.4155,
+      "step": 22566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1067103548580348e-05,
+      "loss": 0.4473,
+      "step": 22567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1060349420441605e-05,
+      "loss": 0.5209,
+      "step": 22568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1053596247735775e-05,
+      "loss": 0.4245,
+      "step": 22569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1046844030544598e-05,
+      "loss": 0.539,
+      "step": 22570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.104009276894977e-05,
+      "loss": 0.4465,
+      "step": 22571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1033342463033057e-05,
+      "loss": 0.3368,
+      "step": 22572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1026593112876136e-05,
+      "loss": 0.3509,
+      "step": 22573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1019844718560677e-05,
+      "loss": 0.4731,
+      "step": 22574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.101309728016838e-05,
+      "loss": 0.4135,
+      "step": 22575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1006350797780884e-05,
+      "loss": 0.4715,
+      "step": 22576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0999605271479882e-05,
+      "loss": 0.3945,
+      "step": 22577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0992860701347005e-05,
+      "loss": 0.4214,
+      "step": 22578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0986117087463886e-05,
+      "loss": 0.348,
+      "step": 22579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0979374429912135e-05,
+      "loss": 0.4918,
+      "step": 22580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0972632728773335e-05,
+      "loss": 0.4411,
+      "step": 22581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.096589198412917e-05,
+      "loss": 0.4166,
+      "step": 22582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0959152196061137e-05,
+      "loss": 0.4634,
+      "step": 22583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0952413364650837e-05,
+      "loss": 0.4032,
+      "step": 22584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0945675489979822e-05,
+      "loss": 0.3993,
+      "step": 22585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0938938572129672e-05,
+      "loss": 0.4,
+      "step": 22586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0932202611181917e-05,
+      "loss": 0.4017,
+      "step": 22587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0925467607218087e-05,
+      "loss": 0.4135,
+      "step": 22588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.091873356031968e-05,
+      "loss": 0.3832,
+      "step": 22589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0912000470568184e-05,
+      "loss": 0.4005,
+      "step": 22590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0905268338045147e-05,
+      "loss": 0.6068,
+      "step": 22591
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0898537162832022e-05,
+      "loss": 0.5253,
+      "step": 22592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0891806945010273e-05,
+      "loss": 0.4368,
+      "step": 22593
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.088507768466137e-05,
+      "loss": 0.4258,
+      "step": 22594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0878349381866726e-05,
+      "loss": 0.4406,
+      "step": 22595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0871622036707815e-05,
+      "loss": 0.487,
+      "step": 22596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.086489564926607e-05,
+      "loss": 0.498,
+      "step": 22597
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0858170219622864e-05,
+      "loss": 0.3902,
+      "step": 22598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.085144574785962e-05,
+      "loss": 0.4508,
+      "step": 22599
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.084472223405772e-05,
+      "loss": 0.4086,
+      "step": 22600
+    },
+    {
+      "epoch": 0.79,
+      "eval_loss": 0.4164082109928131,
+      "eval_runtime": 846.4984,
+      "eval_samples_per_second": 28.423,
+      "eval_steps_per_second": 14.211,
+      "step": 22600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0837999678298516e-05,
+      "loss": 0.4985,
+      "step": 22601
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.083127808066342e-05,
+      "loss": 0.3977,
+      "step": 22602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0824557441233773e-05,
+      "loss": 0.3765,
+      "step": 22603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0817837760090897e-05,
+      "loss": 0.5418,
+      "step": 22604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.081111903731614e-05,
+      "loss": 0.4676,
+      "step": 22605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0804401272990794e-05,
+      "loss": 0.4841,
+      "step": 22606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.07976844671962e-05,
+      "loss": 0.4428,
+      "step": 22607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.079096862001365e-05,
+      "loss": 0.4415,
+      "step": 22608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0784253731524418e-05,
+      "loss": 0.4101,
+      "step": 22609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0777539801809776e-05,
+      "loss": 0.5084,
+      "step": 22610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0770826830950962e-05,
+      "loss": 0.4516,
+      "step": 22611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0764114819029267e-05,
+      "loss": 0.3358,
+      "step": 22612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0757403766125926e-05,
+      "loss": 0.4892,
+      "step": 22613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.075069367232214e-05,
+      "loss": 0.484,
+      "step": 22614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0743984537699147e-05,
+      "loss": 0.4137,
+      "step": 22615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.07372763623381e-05,
+      "loss": 0.4003,
+      "step": 22616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0730569146320255e-05,
+      "loss": 0.4099,
+      "step": 22617
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0723862889726765e-05,
+      "loss": 0.3606,
+      "step": 22618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0717157592638793e-05,
+      "loss": 0.3972,
+      "step": 22619
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0710453255137507e-05,
+      "loss": 0.4016,
+      "step": 22620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.070374987730401e-05,
+      "loss": 0.4712,
+      "step": 22621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.069704745921951e-05,
+      "loss": 0.5385,
+      "step": 22622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.069034600096508e-05,
+      "loss": 0.4107,
+      "step": 22623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.068364550262183e-05,
+      "loss": 0.4692,
+      "step": 22624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0676945964270877e-05,
+      "loss": 0.3709,
+      "step": 22625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.067024738599327e-05,
+      "loss": 0.4122,
+      "step": 22626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0663549767870126e-05,
+      "loss": 0.4938,
+      "step": 22627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.065685310998251e-05,
+      "loss": 0.4045,
+      "step": 22628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.065015741241144e-05,
+      "loss": 0.3454,
+      "step": 22629
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0643462675237992e-05,
+      "loss": 0.4793,
+      "step": 22630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0636768898543146e-05,
+      "loss": 0.4224,
+      "step": 22631
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0630076082407966e-05,
+      "loss": 0.4458,
+      "step": 22632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0623384226913455e-05,
+      "loss": 0.4949,
+      "step": 22633
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0616693332140578e-05,
+      "loss": 0.3914,
+      "step": 22634
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.061000339817034e-05,
+      "loss": 0.4042,
+      "step": 22635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.060331442508371e-05,
+      "loss": 0.3648,
+      "step": 22636
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0596626412961606e-05,
+      "loss": 0.4156,
+      "step": 22637
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0589939361885036e-05,
+      "loss": 0.4317,
+      "step": 22638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.058325327193491e-05,
+      "loss": 0.4255,
+      "step": 22639
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0576568143192154e-05,
+      "loss": 0.5055,
+      "step": 22640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.056988397573767e-05,
+      "loss": 0.4501,
+      "step": 22641
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0563200769652347e-05,
+      "loss": 0.4407,
+      "step": 22642
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.055651852501712e-05,
+      "loss": 0.4612,
+      "step": 22643
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.054983724191284e-05,
+      "loss": 0.3994,
+      "step": 22644
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.054315692042037e-05,
+      "loss": 0.3569,
+      "step": 22645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.053647756062056e-05,
+      "loss": 0.3927,
+      "step": 22646
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0529799162594244e-05,
+      "loss": 0.3701,
+      "step": 22647
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.052312172642229e-05,
+      "loss": 0.4793,
+      "step": 22648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.051644525218549e-05,
+      "loss": 0.357,
+      "step": 22649
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.050976973996467e-05,
+      "loss": 0.4045,
+      "step": 22650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0503095189840605e-05,
+      "loss": 0.4427,
+      "step": 22651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0496421601894057e-05,
+      "loss": 0.4358,
+      "step": 22652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0489748976205857e-05,
+      "loss": 0.4625,
+      "step": 22653
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0483077312856736e-05,
+      "loss": 0.4342,
+      "step": 22654
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0476406611927456e-05,
+      "loss": 0.4948,
+      "step": 22655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0469736873498736e-05,
+      "loss": 0.3607,
+      "step": 22656
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0463068097651284e-05,
+      "loss": 0.3941,
+      "step": 22657
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.045640028446586e-05,
+      "loss": 0.4289,
+      "step": 22658
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0449733434023154e-05,
+      "loss": 0.4081,
+      "step": 22659
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0443067546403862e-05,
+      "loss": 0.4895,
+      "step": 22660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0436402621688632e-05,
+      "loss": 0.4178,
+      "step": 22661
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.042973865995813e-05,
+      "loss": 0.4576,
+      "step": 22662
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0423075661293068e-05,
+      "loss": 0.4117,
+      "step": 22663
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0416413625774056e-05,
+      "loss": 0.5385,
+      "step": 22664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.040975255348172e-05,
+      "loss": 0.411,
+      "step": 22665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0403092444496696e-05,
+      "loss": 0.4404,
+      "step": 22666
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0396433298899574e-05,
+      "loss": 0.4764,
+      "step": 22667
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.038977511677095e-05,
+      "loss": 0.4951,
+      "step": 22668
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0383117898191438e-05,
+      "loss": 0.4001,
+      "step": 22669
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0376461643241608e-05,
+      "loss": 0.4091,
+      "step": 22670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0369806352002006e-05,
+      "loss": 0.4869,
+      "step": 22671
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0363152024553202e-05,
+      "loss": 0.3999,
+      "step": 22672
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0356498660975688e-05,
+      "loss": 0.3791,
+      "step": 22673
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0349846261350046e-05,
+      "loss": 0.5385,
+      "step": 22674
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0343194825756783e-05,
+      "loss": 0.5096,
+      "step": 22675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.03365443542764e-05,
+      "loss": 0.4568,
+      "step": 22676
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0329894846989383e-05,
+      "loss": 0.4755,
+      "step": 22677
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0323246303976186e-05,
+      "loss": 0.4381,
+      "step": 22678
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0316598725317327e-05,
+      "loss": 0.5051,
+      "step": 22679
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0309952111093256e-05,
+      "loss": 0.443,
+      "step": 22680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.03033064613844e-05,
+      "loss": 0.4538,
+      "step": 22681
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0296661776271196e-05,
+      "loss": 0.4327,
+      "step": 22682
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.029001805583406e-05,
+      "loss": 0.4935,
+      "step": 22683
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0283375300153428e-05,
+      "loss": 0.3983,
+      "step": 22684
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0276733509309685e-05,
+      "loss": 0.4702,
+      "step": 22685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0270092683383223e-05,
+      "loss": 0.4777,
+      "step": 22686
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.026345282245442e-05,
+      "loss": 0.442,
+      "step": 22687
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0256813926603613e-05,
+      "loss": 0.4272,
+      "step": 22688
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.025017599591119e-05,
+      "loss": 0.3632,
+      "step": 22689
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0243539030457492e-05,
+      "loss": 0.3549,
+      "step": 22690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.023690303032283e-05,
+      "loss": 0.4121,
+      "step": 22691
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0230267995587525e-05,
+      "loss": 0.3797,
+      "step": 22692
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0223633926331864e-05,
+      "loss": 0.4722,
+      "step": 22693
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.021700082263619e-05,
+      "loss": 0.4882,
+      "step": 22694
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.021036868458076e-05,
+      "loss": 0.4394,
+      "step": 22695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0203737512245833e-05,
+      "loss": 0.4285,
+      "step": 22696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0197107305711693e-05,
+      "loss": 0.3928,
+      "step": 22697
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0190478065058536e-05,
+      "loss": 0.429,
+      "step": 22698
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0183849790366672e-05,
+      "loss": 0.3627,
+      "step": 22699
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0177222481716284e-05,
+      "loss": 0.4097,
+      "step": 22700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0170596139187593e-05,
+      "loss": 0.3652,
+      "step": 22701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0163970762860783e-05,
+      "loss": 0.4187,
+      "step": 22702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0157346352816063e-05,
+      "loss": 0.3882,
+      "step": 22703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0150722909133578e-05,
+      "loss": 0.3984,
+      "step": 22704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0144100431893554e-05,
+      "loss": 0.4084,
+      "step": 22705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.01374789211761e-05,
+      "loss": 0.4066,
+      "step": 22706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0130858377061378e-05,
+      "loss": 0.378,
+      "step": 22707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.01242387996295e-05,
+      "loss": 0.4367,
+      "step": 22708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0117620188960572e-05,
+      "loss": 0.4127,
+      "step": 22709
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0111002545134748e-05,
+      "loss": 0.4593,
+      "step": 22710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.01043858682321e-05,
+      "loss": 0.4408,
+      "step": 22711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0097770158332708e-05,
+      "loss": 0.4131,
+      "step": 22712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.009115541551665e-05,
+      "loss": 0.4324,
+      "step": 22713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.008454163986396e-05,
+      "loss": 0.3882,
+      "step": 22714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.007792883145474e-05,
+      "loss": 0.4827,
+      "step": 22715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0071316990368993e-05,
+      "loss": 0.3804,
+      "step": 22716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0064706116686772e-05,
+      "loss": 0.409,
+      "step": 22717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.005809621048802e-05,
+      "loss": 0.421,
+      "step": 22718
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.00514872718528e-05,
+      "loss": 0.4498,
+      "step": 22719
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0044879300861096e-05,
+      "loss": 0.4615,
+      "step": 22720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0038272297592887e-05,
+      "loss": 0.4573,
+      "step": 22721
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0031666262128125e-05,
+      "loss": 0.3615,
+      "step": 22722
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0025061194546747e-05,
+      "loss": 0.4361,
+      "step": 22723
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0018457094928745e-05,
+      "loss": 0.414,
+      "step": 22724
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0011853963354022e-05,
+      "loss": 0.3843,
+      "step": 22725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0005251799902515e-05,
+      "loss": 0.4398,
+      "step": 22726
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9998650604654112e-05,
+      "loss": 0.499,
+      "step": 22727
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9992050377688686e-05,
+      "loss": 0.4006,
+      "step": 22728
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9985451119086184e-05,
+      "loss": 0.4615,
+      "step": 22729
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9978852828926454e-05,
+      "loss": 0.4337,
+      "step": 22730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.997225550728934e-05,
+      "loss": 0.4303,
+      "step": 22731
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9965659154254714e-05,
+      "loss": 0.4352,
+      "step": 22732
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9959063769902363e-05,
+      "loss": 0.5225,
+      "step": 22733
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9952469354312185e-05,
+      "loss": 0.4925,
+      "step": 22734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.994587590756397e-05,
+      "loss": 0.4151,
+      "step": 22735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.993928342973751e-05,
+      "loss": 0.4504,
+      "step": 22736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9932691920912593e-05,
+      "loss": 0.4007,
+      "step": 22737
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9926101381169005e-05,
+      "loss": 0.4571,
+      "step": 22738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.991951181058649e-05,
+      "loss": 0.4072,
+      "step": 22739
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.991292320924485e-05,
+      "loss": 0.3655,
+      "step": 22740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.990633557722381e-05,
+      "loss": 0.483,
+      "step": 22741
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.98997489146031e-05,
+      "loss": 0.4571,
+      "step": 22742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.989316322146243e-05,
+      "loss": 0.3684,
+      "step": 22743
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.988657849788149e-05,
+      "loss": 0.3896,
+      "step": 22744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9879994743940035e-05,
+      "loss": 0.4297,
+      "step": 22745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9873411959717723e-05,
+      "loss": 0.517,
+      "step": 22746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9866830145294214e-05,
+      "loss": 0.4312,
+      "step": 22747
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.986024930074919e-05,
+      "loss": 0.4488,
+      "step": 22748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9853669426162257e-05,
+      "loss": 0.3663,
+      "step": 22749
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9847090521613122e-05,
+      "loss": 0.4329,
+      "step": 22750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.984051258718137e-05,
+      "loss": 0.4035,
+      "step": 22751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.983393562294663e-05,
+      "loss": 0.4454,
+      "step": 22752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9827359628988497e-05,
+      "loss": 0.427,
+      "step": 22753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9820784605386534e-05,
+      "loss": 0.3718,
+      "step": 22754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.981421055222038e-05,
+      "loss": 0.4555,
+      "step": 22755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9807637469569575e-05,
+      "loss": 0.3428,
+      "step": 22756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.980106535751367e-05,
+      "loss": 0.4886,
+      "step": 22757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.979449421613222e-05,
+      "loss": 0.3935,
+      "step": 22758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9787924045504714e-05,
+      "loss": 0.4561,
+      "step": 22759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9781354845710743e-05,
+      "loss": 0.5358,
+      "step": 22760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.977478661682979e-05,
+      "loss": 0.4848,
+      "step": 22761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.976821935894133e-05,
+      "loss": 0.4178,
+      "step": 22762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9761653072124875e-05,
+      "loss": 0.443,
+      "step": 22763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9755087756459854e-05,
+      "loss": 0.4128,
+      "step": 22764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.97485234120258e-05,
+      "loss": 0.4392,
+      "step": 22765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9741960038902118e-05,
+      "loss": 0.4885,
+      "step": 22766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9735397637168262e-05,
+      "loss": 0.3929,
+      "step": 22767
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9728836206903656e-05,
+      "loss": 0.3624,
+      "step": 22768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9722275748187703e-05,
+      "loss": 0.3844,
+      "step": 22769
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9715716261099792e-05,
+      "loss": 0.3932,
+      "step": 22770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9709157745719364e-05,
+      "loss": 0.4324,
+      "step": 22771
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9702600202125775e-05,
+      "loss": 0.408,
+      "step": 22772
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9696043630398386e-05,
+      "loss": 0.4797,
+      "step": 22773
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9689488030616564e-05,
+      "loss": 0.5194,
+      "step": 22774
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9682933402859615e-05,
+      "loss": 0.3921,
+      "step": 22775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9676379747206943e-05,
+      "loss": 0.3916,
+      "step": 22776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9669827063737813e-05,
+      "loss": 0.4163,
+      "step": 22777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9663275352531562e-05,
+      "loss": 0.4755,
+      "step": 22778
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9656724613667476e-05,
+      "loss": 0.4295,
+      "step": 22779
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9650174847224823e-05,
+      "loss": 0.3764,
+      "step": 22780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9643626053282916e-05,
+      "loss": 0.3971,
+      "step": 22781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9637078231921e-05,
+      "loss": 0.4334,
+      "step": 22782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.963053138321832e-05,
+      "loss": 0.4326,
+      "step": 22783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9623985507254127e-05,
+      "loss": 0.4382,
+      "step": 22784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9617440604107607e-05,
+      "loss": 0.5176,
+      "step": 22785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9610896673858025e-05,
+      "loss": 0.4315,
+      "step": 22786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9604353716584577e-05,
+      "loss": 0.3925,
+      "step": 22787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9597811732366432e-05,
+      "loss": 0.4367,
+      "step": 22788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9591270721282796e-05,
+      "loss": 0.4859,
+      "step": 22789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.958473068341279e-05,
+      "loss": 0.4987,
+      "step": 22790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9578191618835628e-05,
+      "loss": 0.4789,
+      "step": 22791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.957165352763043e-05,
+      "loss": 0.4676,
+      "step": 22792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9565116409876327e-05,
+      "loss": 0.4377,
+      "step": 22793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9558580265652448e-05,
+      "loss": 0.4284,
+      "step": 22794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9552045095037864e-05,
+      "loss": 0.4568,
+      "step": 22795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.954551089811173e-05,
+      "loss": 0.3299,
+      "step": 22796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9538977674953106e-05,
+      "loss": 0.4149,
+      "step": 22797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9532445425641065e-05,
+      "loss": 0.4206,
+      "step": 22798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9525914150254663e-05,
+      "loss": 0.5136,
+      "step": 22799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.951938384887294e-05,
+      "loss": 0.4149,
+      "step": 22800
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 0.4144943356513977,
+      "eval_runtime": 843.9802,
+      "eval_samples_per_second": 28.508,
+      "eval_steps_per_second": 14.254,
+      "step": 22800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9512854521574976e-05,
+      "loss": 0.3957,
+      "step": 22801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.950632616843976e-05,
+      "loss": 0.3805,
+      "step": 22802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.949979878954633e-05,
+      "loss": 0.3927,
+      "step": 22803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.949327238497367e-05,
+      "loss": 0.4135,
+      "step": 22804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9486746954800793e-05,
+      "loss": 0.3867,
+      "step": 22805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9480222499106626e-05,
+      "loss": 0.3901,
+      "step": 22806
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9473699017970204e-05,
+      "loss": 0.4269,
+      "step": 22807
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9467176511470464e-05,
+      "loss": 0.4409,
+      "step": 22808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9460654979686322e-05,
+      "loss": 0.4076,
+      "step": 22809
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9454134422696746e-05,
+      "loss": 0.5018,
+      "step": 22810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9447614840580608e-05,
+      "loss": 0.4054,
+      "step": 22811
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9441096233416868e-05,
+      "loss": 0.4175,
+      "step": 22812
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9434578601284405e-05,
+      "loss": 0.3839,
+      "step": 22813
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.94280619442621e-05,
+      "loss": 0.4261,
+      "step": 22814
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9421546262428835e-05,
+      "loss": 0.4466,
+      "step": 22815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9415031555863428e-05,
+      "loss": 0.423,
+      "step": 22816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.94085178246448e-05,
+      "loss": 0.3793,
+      "step": 22817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.940200506885176e-05,
+      "loss": 0.3898,
+      "step": 22818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9395493288563117e-05,
+      "loss": 0.3535,
+      "step": 22819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.93889824838577e-05,
+      "loss": 0.516,
+      "step": 22820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9382472654814276e-05,
+      "loss": 0.4305,
+      "step": 22821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9375963801511697e-05,
+      "loss": 0.4071,
+      "step": 22822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.936945592402871e-05,
+      "loss": 0.3625,
+      "step": 22823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9362949022444087e-05,
+      "loss": 0.4666,
+      "step": 22824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9356443096836585e-05,
+      "loss": 0.3597,
+      "step": 22825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9349938147284918e-05,
+      "loss": 0.4592,
+      "step": 22826
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.934343417386786e-05,
+      "loss": 0.3881,
+      "step": 22827
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.933693117666412e-05,
+      "loss": 0.4273,
+      "step": 22828
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9330429155752394e-05,
+      "loss": 0.4921,
+      "step": 22829
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9323928111211377e-05,
+      "loss": 0.4464,
+      "step": 22830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9317428043119746e-05,
+      "loss": 0.4221,
+      "step": 22831
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9310928951556205e-05,
+      "loss": 0.4191,
+      "step": 22832
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9304430836599395e-05,
+      "loss": 0.4176,
+      "step": 22833
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9297933698327976e-05,
+      "loss": 0.4053,
+      "step": 22834
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.929143753682057e-05,
+      "loss": 0.3656,
+      "step": 22835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9284942352155812e-05,
+      "loss": 0.453,
+      "step": 22836
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9278448144412276e-05,
+      "loss": 0.4728,
+      "step": 22837
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.927195491366863e-05,
+      "loss": 0.3904,
+      "step": 22838
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.926546266000343e-05,
+      "loss": 0.4644,
+      "step": 22839
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.925897138349526e-05,
+      "loss": 0.328,
+      "step": 22840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9252481084222685e-05,
+      "loss": 0.3219,
+      "step": 22841
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9245991762264225e-05,
+      "loss": 0.4578,
+      "step": 22842
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9239503417698478e-05,
+      "loss": 0.3887,
+      "step": 22843
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9233016050603958e-05,
+      "loss": 0.4405,
+      "step": 22844
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.922652966105918e-05,
+      "loss": 0.5243,
+      "step": 22845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.922004424914263e-05,
+      "loss": 0.3824,
+      "step": 22846
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9213559814932814e-05,
+      "loss": 0.4739,
+      "step": 22847
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9207076358508235e-05,
+      "loss": 0.4294,
+      "step": 22848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9200593879947358e-05,
+      "loss": 0.415,
+      "step": 22849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9194112379328634e-05,
+      "loss": 0.4641,
+      "step": 22850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9187631856730516e-05,
+      "loss": 0.4815,
+      "step": 22851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9181152312231422e-05,
+      "loss": 0.4466,
+      "step": 22852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.917467374590981e-05,
+      "loss": 0.451,
+      "step": 22853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9168196157844054e-05,
+      "loss": 0.4627,
+      "step": 22854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.916171954811259e-05,
+      "loss": 0.4531,
+      "step": 22855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.915524391679375e-05,
+      "loss": 0.4664,
+      "step": 22856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9148769263965994e-05,
+      "loss": 0.408,
+      "step": 22857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9142295589707625e-05,
+      "loss": 0.4713,
+      "step": 22858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9135822894097023e-05,
+      "loss": 0.4633,
+      "step": 22859
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912935117721252e-05,
+      "loss": 0.4228,
+      "step": 22860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912288043913242e-05,
+      "loss": 0.397,
+      "step": 22861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911641067993509e-05,
+      "loss": 0.4838,
+      "step": 22862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9109941899698813e-05,
+      "loss": 0.4922,
+      "step": 22863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9103474098501885e-05,
+      "loss": 0.4182,
+      "step": 22864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9097007276422585e-05,
+      "loss": 0.5062,
+      "step": 22865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9090541433539155e-05,
+      "loss": 0.4578,
+      "step": 22866
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.908407656992991e-05,
+      "loss": 0.3798,
+      "step": 22867
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9077612685673064e-05,
+      "loss": 0.4802,
+      "step": 22868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.907114978084685e-05,
+      "loss": 0.4963,
+      "step": 22869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9064687855529506e-05,
+      "loss": 0.3617,
+      "step": 22870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9058226909799238e-05,
+      "loss": 0.4366,
+      "step": 22871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9051766943734206e-05,
+      "loss": 0.4941,
+      "step": 22872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.904530795741265e-05,
+      "loss": 0.4319,
+      "step": 22873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9038849950912728e-05,
+      "loss": 0.4227,
+      "step": 22874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9032392924312614e-05,
+      "loss": 0.589,
+      "step": 22875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9025936877690433e-05,
+      "loss": 0.3849,
+      "step": 22876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9019481811124318e-05,
+      "loss": 0.3788,
+      "step": 22877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9013027724692444e-05,
+      "loss": 0.4325,
+      "step": 22878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9006574618472905e-05,
+      "loss": 0.4445,
+      "step": 22879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9000122492543792e-05,
+      "loss": 0.4599,
+      "step": 22880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.899367134698321e-05,
+      "loss": 0.4765,
+      "step": 22881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.898722118186922e-05,
+      "loss": 0.4398,
+      "step": 22882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8980771997279913e-05,
+      "loss": 0.4454,
+      "step": 22883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8974323793293347e-05,
+      "loss": 0.4004,
+      "step": 22884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8967876569987565e-05,
+      "loss": 0.3905,
+      "step": 22885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8961430327440576e-05,
+      "loss": 0.3585,
+      "step": 22886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8954985065730403e-05,
+      "loss": 0.4481,
+      "step": 22887
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8948540784935098e-05,
+      "loss": 0.4,
+      "step": 22888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8942097485132626e-05,
+      "loss": 0.4707,
+      "step": 22889
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8935655166400977e-05,
+      "loss": 0.4688,
+      "step": 22890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8929213828818128e-05,
+      "loss": 0.502,
+      "step": 22891
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.892277347246201e-05,
+      "loss": 0.4964,
+      "step": 22892
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8916334097410625e-05,
+      "loss": 0.4172,
+      "step": 22893
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8909895703741888e-05,
+      "loss": 0.4602,
+      "step": 22894
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8903458291533715e-05,
+      "loss": 0.4303,
+      "step": 22895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.889702186086404e-05,
+      "loss": 0.4348,
+      "step": 22896
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.889058641181072e-05,
+      "loss": 0.4652,
+      "step": 22897
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.88841519444517e-05,
+      "loss": 0.4742,
+      "step": 22898
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8877718458864846e-05,
+      "loss": 0.4053,
+      "step": 22899
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8871285955128005e-05,
+      "loss": 0.4051,
+      "step": 22900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8864854433319058e-05,
+      "loss": 0.4509,
+      "step": 22901
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8858423893515786e-05,
+      "loss": 0.4464,
+      "step": 22902
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8851994335796107e-05,
+      "loss": 0.427,
+      "step": 22903
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8845565760237792e-05,
+      "loss": 0.4203,
+      "step": 22904
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.883913816691867e-05,
+      "loss": 0.3931,
+      "step": 22905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8832711555916506e-05,
+      "loss": 0.4831,
+      "step": 22906
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8826285927309108e-05,
+      "loss": 0.4115,
+      "step": 22907
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8819861281174212e-05,
+      "loss": 0.406,
+      "step": 22908
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8813437617589624e-05,
+      "loss": 0.4973,
+      "step": 22909
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.880701493663307e-05,
+      "loss": 0.4436,
+      "step": 22910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8800593238382304e-05,
+      "loss": 0.476,
+      "step": 22911
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.879417252291502e-05,
+      "loss": 0.3906,
+      "step": 22912
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.878775279030892e-05,
+      "loss": 0.4354,
+      "step": 22913
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.878133404064176e-05,
+      "loss": 0.4232,
+      "step": 22914
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8774916273991193e-05,
+      "loss": 0.517,
+      "step": 22915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8768499490434898e-05,
+      "loss": 0.4306,
+      "step": 22916
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8762083690050537e-05,
+      "loss": 0.3701,
+      "step": 22917
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8755668872915743e-05,
+      "loss": 0.4079,
+      "step": 22918
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8749255039108216e-05,
+      "loss": 0.4431,
+      "step": 22919
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8742842188705535e-05,
+      "loss": 0.3791,
+      "step": 22920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8736430321785338e-05,
+      "loss": 0.4177,
+      "step": 22921
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8730019438425216e-05,
+      "loss": 0.4434,
+      "step": 22922
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8723609538702745e-05,
+      "loss": 0.4131,
+      "step": 22923
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.871720062269555e-05,
+      "loss": 0.3584,
+      "step": 22924
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8710792690481183e-05,
+      "loss": 0.4299,
+      "step": 22925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.87043857421372e-05,
+      "loss": 0.4147,
+      "step": 22926
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8697979777741148e-05,
+      "loss": 0.4798,
+      "step": 22927
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8691574797370526e-05,
+      "loss": 0.4613,
+      "step": 22928
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8685170801102914e-05,
+      "loss": 0.3866,
+      "step": 22929
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8678767789015793e-05,
+      "loss": 0.4191,
+      "step": 22930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.867236576118666e-05,
+      "loss": 0.4069,
+      "step": 22931
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8665964717693007e-05,
+      "loss": 0.3915,
+      "step": 22932
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8659564658612272e-05,
+      "loss": 0.4737,
+      "step": 22933
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.865316558402198e-05,
+      "loss": 0.4184,
+      "step": 22934
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.864676749399955e-05,
+      "loss": 0.3344,
+      "step": 22935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.864037038862242e-05,
+      "loss": 0.4123,
+      "step": 22936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8633974267968014e-05,
+      "loss": 0.4566,
+      "step": 22937
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.862757913211375e-05,
+      "loss": 0.383,
+      "step": 22938
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8621184981137006e-05,
+      "loss": 0.4399,
+      "step": 22939
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8614791815115217e-05,
+      "loss": 0.4221,
+      "step": 22940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.860839963412575e-05,
+      "loss": 0.4439,
+      "step": 22941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8602008438245954e-05,
+      "loss": 0.5027,
+      "step": 22942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.859561822755319e-05,
+      "loss": 0.4261,
+      "step": 22943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8589229002124786e-05,
+      "loss": 0.4229,
+      "step": 22944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.858284076203811e-05,
+      "loss": 0.4439,
+      "step": 22945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.857645350737046e-05,
+      "loss": 0.4743,
+      "step": 22946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.857006723819914e-05,
+      "loss": 0.4968,
+      "step": 22947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8563681954601454e-05,
+      "loss": 0.3862,
+      "step": 22948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.855729765665466e-05,
+      "loss": 0.4591,
+      "step": 22949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8550914344436065e-05,
+      "loss": 0.368,
+      "step": 22950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.854453201802292e-05,
+      "loss": 0.4635,
+      "step": 22951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.853815067749246e-05,
+      "loss": 0.5125,
+      "step": 22952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.853177032292194e-05,
+      "loss": 0.3873,
+      "step": 22953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8525390954388522e-05,
+      "loss": 0.4592,
+      "step": 22954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8519012571969507e-05,
+      "loss": 0.4167,
+      "step": 22955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8512635175742042e-05,
+      "loss": 0.4036,
+      "step": 22956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8506258765783336e-05,
+      "loss": 0.3822,
+      "step": 22957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8499883342170542e-05,
+      "loss": 0.3596,
+      "step": 22958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8493508904980816e-05,
+      "loss": 0.4605,
+      "step": 22959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.848713545429136e-05,
+      "loss": 0.3703,
+      "step": 22960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.848076299017928e-05,
+      "loss": 0.417,
+      "step": 22961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.847439151272171e-05,
+      "loss": 0.3961,
+      "step": 22962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8468021021995763e-05,
+      "loss": 0.347,
+      "step": 22963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8461651518078516e-05,
+      "loss": 0.4144,
+      "step": 22964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.845528300104712e-05,
+      "loss": 0.3914,
+      "step": 22965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8448915470978623e-05,
+      "loss": 0.4442,
+      "step": 22966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.84425489279501e-05,
+      "loss": 0.4326,
+      "step": 22967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8436183372038597e-05,
+      "loss": 0.4024,
+      "step": 22968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8429818803321165e-05,
+      "loss": 0.443,
+      "step": 22969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.842345522187481e-05,
+      "loss": 0.5004,
+      "step": 22970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.84170926277766e-05,
+      "loss": 0.4191,
+      "step": 22971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8410731021103522e-05,
+      "loss": 0.4378,
+      "step": 22972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.840437040193258e-05,
+      "loss": 0.4091,
+      "step": 22973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8398010770340744e-05,
+      "loss": 0.3858,
+      "step": 22974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.839165212640498e-05,
+      "loss": 0.4046,
+      "step": 22975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8385294470202275e-05,
+      "loss": 0.4193,
+      "step": 22976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.837893780180957e-05,
+      "loss": 0.3473,
+      "step": 22977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8372582121303793e-05,
+      "loss": 0.4865,
+      "step": 22978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8366227428761885e-05,
+      "loss": 0.5113,
+      "step": 22979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8359873724260712e-05,
+      "loss": 0.4207,
+      "step": 22980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8353521007877237e-05,
+      "loss": 0.4033,
+      "step": 22981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8347169279688326e-05,
+      "loss": 0.3748,
+      "step": 22982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8340818539770843e-05,
+      "loss": 0.3912,
+      "step": 22983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8334468788201675e-05,
+      "loss": 0.4528,
+      "step": 22984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8328120025057626e-05,
+      "loss": 0.4111,
+      "step": 22985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8321772250415602e-05,
+      "loss": 0.412,
+      "step": 22986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8315425464352433e-05,
+      "loss": 0.4833,
+      "step": 22987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8309079666944883e-05,
+      "loss": 0.4274,
+      "step": 22988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8302734858269754e-05,
+      "loss": 0.4098,
+      "step": 22989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8296391038403893e-05,
+      "loss": 0.4893,
+      "step": 22990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.829004820742406e-05,
+      "loss": 0.3907,
+      "step": 22991
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8283706365407004e-05,
+      "loss": 0.4791,
+      "step": 22992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8277365512429512e-05,
+      "loss": 0.3899,
+      "step": 22993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8271025648568286e-05,
+      "loss": 0.4277,
+      "step": 22994
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.826468677390012e-05,
+      "loss": 0.4442,
+      "step": 22995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8258348888501697e-05,
+      "loss": 0.4163,
+      "step": 22996
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8252011992449735e-05,
+      "loss": 0.4156,
+      "step": 22997
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.824567608582094e-05,
+      "loss": 0.4143,
+      "step": 22998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8239341168691947e-05,
+      "loss": 0.3941,
+      "step": 22999
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.823300724113951e-05,
+      "loss": 0.3856,
+      "step": 23000
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 0.41282618045806885,
+      "eval_runtime": 846.2056,
+      "eval_samples_per_second": 28.433,
+      "eval_steps_per_second": 14.216,
+      "step": 23000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8226674303240243e-05,
+      "loss": 0.4699,
+      "step": 23001
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8220342355070808e-05,
+      "loss": 0.4653,
+      "step": 23002
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8214011396707842e-05,
+      "loss": 0.3773,
+      "step": 23003
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8207681428227962e-05,
+      "loss": 0.4806,
+      "step": 23004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8201352449707766e-05,
+      "loss": 0.4781,
+      "step": 23005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8195024461223897e-05,
+      "loss": 0.4553,
+      "step": 23006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8188697462852934e-05,
+      "loss": 0.3764,
+      "step": 23007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.818237145467143e-05,
+      "loss": 0.4341,
+      "step": 23008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8176046436755988e-05,
+      "loss": 0.3815,
+      "step": 23009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8169722409183097e-05,
+      "loss": 0.4418,
+      "step": 23010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8163399372029376e-05,
+      "loss": 0.4675,
+      "step": 23011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8157077325371318e-05,
+      "loss": 0.4832,
+      "step": 23012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.815075626928544e-05,
+      "loss": 0.3953,
+      "step": 23013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8144436203848257e-05,
+      "loss": 0.4478,
+      "step": 23014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8138117129136232e-05,
+      "loss": 0.3853,
+      "step": 23015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8131799045225893e-05,
+      "loss": 0.4831,
+      "step": 23016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.812548195219369e-05,
+      "loss": 0.3785,
+      "step": 23017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.811916585011608e-05,
+      "loss": 0.5717,
+      "step": 23018
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.811285073906952e-05,
+      "loss": 0.5406,
+      "step": 23019
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.81065366191304e-05,
+      "loss": 0.5322,
+      "step": 23020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.810022349037519e-05,
+      "loss": 0.3224,
+      "step": 23021
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.80939113528803e-05,
+      "loss": 0.4249,
+      "step": 23022
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.80876002067221e-05,
+      "loss": 0.3848,
+      "step": 23023
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8081290051976996e-05,
+      "loss": 0.5075,
+      "step": 23024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8074980888721328e-05,
+      "loss": 0.4348,
+      "step": 23025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8068672717031508e-05,
+      "loss": 0.3691,
+      "step": 23026
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.806236553698386e-05,
+      "loss": 0.3624,
+      "step": 23027
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.805605934865473e-05,
+      "loss": 0.4334,
+      "step": 23028
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8049754152120435e-05,
+      "loss": 0.4237,
+      "step": 23029
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.804344994745727e-05,
+      "loss": 0.392,
+      "step": 23030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8037146734741572e-05,
+      "loss": 0.4076,
+      "step": 23031
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.803084451404963e-05,
+      "loss": 0.3595,
+      "step": 23032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8024543285457707e-05,
+      "loss": 0.5786,
+      "step": 23033
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.801824304904206e-05,
+      "loss": 0.4478,
+      "step": 23034
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.801194380487894e-05,
+      "loss": 0.4225,
+      "step": 23035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8005645553044626e-05,
+      "loss": 0.3735,
+      "step": 23036
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7999348293615325e-05,
+      "loss": 0.4793,
+      "step": 23037
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7993052026667258e-05,
+      "loss": 0.4135,
+      "step": 23038
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7986756752276623e-05,
+      "loss": 0.5275,
+      "step": 23039
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.798046247051962e-05,
+      "loss": 0.3539,
+      "step": 23040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.797416918147241e-05,
+      "loss": 0.3841,
+      "step": 23041
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7967876885211198e-05,
+      "loss": 0.3676,
+      "step": 23042
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7961585581812125e-05,
+      "loss": 0.4717,
+      "step": 23043
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7955295271351345e-05,
+      "loss": 0.3557,
+      "step": 23044
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7949005953904975e-05,
+      "loss": 0.3952,
+      "step": 23045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7942717629549134e-05,
+      "loss": 0.421,
+      "step": 23046
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.793643029835995e-05,
+      "loss": 0.4646,
+      "step": 23047
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7930143960413526e-05,
+      "loss": 0.4294,
+      "step": 23048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7923858615785937e-05,
+      "loss": 0.434,
+      "step": 23049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.791757426455325e-05,
+      "loss": 0.3911,
+      "step": 23050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.791129090679151e-05,
+      "loss": 0.4891,
+      "step": 23051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7905008542576818e-05,
+      "loss": 0.3848,
+      "step": 23052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7898727171985174e-05,
+      "loss": 0.4926,
+      "step": 23053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.789244679509262e-05,
+      "loss": 0.3553,
+      "step": 23054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7886167411975153e-05,
+      "loss": 0.3878,
+      "step": 23055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.787988902270876e-05,
+      "loss": 0.4116,
+      "step": 23056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.787361162736948e-05,
+      "loss": 0.4585,
+      "step": 23057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7867335226033266e-05,
+      "loss": 0.4483,
+      "step": 23058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7861059818776083e-05,
+      "loss": 0.4211,
+      "step": 23059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.785478540567387e-05,
+      "loss": 0.4708,
+      "step": 23060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7848511986802562e-05,
+      "loss": 0.3868,
+      "step": 23061
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.784223956223814e-05,
+      "loss": 0.4668,
+      "step": 23062
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.783596813205648e-05,
+      "loss": 0.3989,
+      "step": 23063
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7829697696333492e-05,
+      "loss": 0.4834,
+      "step": 23064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.782342825514508e-05,
+      "loss": 0.4252,
+      "step": 23065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7817159808567084e-05,
+      "loss": 0.4225,
+      "step": 23066
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7810892356675434e-05,
+      "loss": 0.4536,
+      "step": 23067
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.780462589954597e-05,
+      "loss": 0.4636,
+      "step": 23068
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7798360437254514e-05,
+      "loss": 0.4222,
+      "step": 23069
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.779209596987692e-05,
+      "loss": 0.5081,
+      "step": 23070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7785832497488995e-05,
+      "loss": 0.4086,
+      "step": 23071
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7779570020166536e-05,
+      "loss": 0.4484,
+      "step": 23072
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7773308537985377e-05,
+      "loss": 0.3776,
+      "step": 23073
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7767048051021283e-05,
+      "loss": 0.3221,
+      "step": 23074
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7760788559350026e-05,
+      "loss": 0.4361,
+      "step": 23075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.775453006304737e-05,
+      "loss": 0.3597,
+      "step": 23076
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.774827256218904e-05,
+      "loss": 0.4437,
+      "step": 23077
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7742016056850806e-05,
+      "loss": 0.4134,
+      "step": 23078
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7735760547108394e-05,
+      "loss": 0.4408,
+      "step": 23079
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.77295060330375e-05,
+      "loss": 0.4006,
+      "step": 23080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.772325251471383e-05,
+      "loss": 0.3889,
+      "step": 23081
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7716999992213036e-05,
+      "loss": 0.3752,
+      "step": 23082
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.771074846561086e-05,
+      "loss": 0.4585,
+      "step": 23083
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7704497934982934e-05,
+      "loss": 0.373,
+      "step": 23084
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7698248400404906e-05,
+      "loss": 0.4961,
+      "step": 23085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7691999861952423e-05,
+      "loss": 0.3434,
+      "step": 23086
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7685752319701088e-05,
+      "loss": 0.4716,
+      "step": 23087
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.767950577372657e-05,
+      "loss": 0.56,
+      "step": 23088
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7673260224104436e-05,
+      "loss": 0.5054,
+      "step": 23089
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.766701567091029e-05,
+      "loss": 0.3933,
+      "step": 23090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7660772114219705e-05,
+      "loss": 0.4542,
+      "step": 23091
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7654529554108233e-05,
+      "loss": 0.4717,
+      "step": 23092
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.764828799065147e-05,
+      "loss": 0.4161,
+      "step": 23093
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7642047423924933e-05,
+      "loss": 0.4635,
+      "step": 23094
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.763580785400416e-05,
+      "loss": 0.4111,
+      "step": 23095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7629569280964664e-05,
+      "loss": 0.4701,
+      "step": 23096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.762333170488194e-05,
+      "loss": 0.4373,
+      "step": 23097
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.761709512583152e-05,
+      "loss": 0.4638,
+      "step": 23098
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.761085954388887e-05,
+      "loss": 0.3767,
+      "step": 23099
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7604624959129467e-05,
+      "loss": 0.4422,
+      "step": 23100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7598391371628753e-05,
+      "loss": 0.42,
+      "step": 23101
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7592158781462164e-05,
+      "loss": 0.3031,
+      "step": 23102
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7585927188705176e-05,
+      "loss": 0.3632,
+      "step": 23103
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7579696593433203e-05,
+      "loss": 0.3841,
+      "step": 23104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7573466995721633e-05,
+      "loss": 0.3668,
+      "step": 23105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7567238395645892e-05,
+      "loss": 0.4345,
+      "step": 23106
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7561010793281342e-05,
+      "loss": 0.4241,
+      "step": 23107
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7554784188703345e-05,
+      "loss": 0.3945,
+      "step": 23108
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7548558581987307e-05,
+      "loss": 0.431,
+      "step": 23109
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7542333973208557e-05,
+      "loss": 0.4355,
+      "step": 23110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7536110362442437e-05,
+      "loss": 0.4593,
+      "step": 23111
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7529887749764272e-05,
+      "loss": 0.3819,
+      "step": 23112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7523666135249362e-05,
+      "loss": 0.4237,
+      "step": 23113
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.751744551897303e-05,
+      "loss": 0.4231,
+      "step": 23114
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7511225901010563e-05,
+      "loss": 0.408,
+      "step": 23115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7505007281437236e-05,
+      "loss": 0.4385,
+      "step": 23116
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7498789660328318e-05,
+      "loss": 0.4412,
+      "step": 23117
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.749257303775903e-05,
+      "loss": 0.4715,
+      "step": 23118
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7486357413804666e-05,
+      "loss": 0.4338,
+      "step": 23119
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7480142788540434e-05,
+      "loss": 0.4055,
+      "step": 23120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.747392916204158e-05,
+      "loss": 0.4526,
+      "step": 23121
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7467716534383226e-05,
+      "loss": 0.409,
+      "step": 23122
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7461504905640634e-05,
+      "loss": 0.4754,
+      "step": 23123
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.745529427588899e-05,
+      "loss": 0.4338,
+      "step": 23124
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.744908464520343e-05,
+      "loss": 0.369,
+      "step": 23125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7442876013659128e-05,
+      "loss": 0.4689,
+      "step": 23126
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.74366683813312e-05,
+      "loss": 0.4133,
+      "step": 23127
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7430461748294833e-05,
+      "loss": 0.436,
+      "step": 23128
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7424256114625115e-05,
+      "loss": 0.4438,
+      "step": 23129
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.741805148039717e-05,
+      "loss": 0.5293,
+      "step": 23130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.741184784568608e-05,
+      "loss": 0.4711,
+      "step": 23131
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7405645210566913e-05,
+      "loss": 0.4555,
+      "step": 23132
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.739944357511478e-05,
+      "loss": 0.404,
+      "step": 23133
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.739324293940473e-05,
+      "loss": 0.4101,
+      "step": 23134
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7387043303511806e-05,
+      "loss": 0.4579,
+      "step": 23135
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.738084466751104e-05,
+      "loss": 0.4738,
+      "step": 23136
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.737464703147744e-05,
+      "loss": 0.4793,
+      "step": 23137
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7368450395486058e-05,
+      "loss": 0.3826,
+      "step": 23138
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7362254759611885e-05,
+      "loss": 0.5142,
+      "step": 23139
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7356060123929884e-05,
+      "loss": 0.4118,
+      "step": 23140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7349866488515053e-05,
+      "loss": 0.4464,
+      "step": 23141
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7343673853442345e-05,
+      "loss": 0.4695,
+      "step": 23142
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7337482218786695e-05,
+      "loss": 0.3896,
+      "step": 23143
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.733129158462308e-05,
+      "loss": 0.4536,
+      "step": 23144
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.732510195102641e-05,
+      "loss": 0.4065,
+      "step": 23145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7318913318071605e-05,
+      "loss": 0.5154,
+      "step": 23146
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7312725685833554e-05,
+      "loss": 0.4146,
+      "step": 23147
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.730653905438714e-05,
+      "loss": 0.4944,
+      "step": 23148
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.730035342380727e-05,
+      "loss": 0.4196,
+      "step": 23149
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7294168794168808e-05,
+      "loss": 0.5027,
+      "step": 23150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7287985165546593e-05,
+      "loss": 0.4215,
+      "step": 23151
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.728180253801548e-05,
+      "loss": 0.3876,
+      "step": 23152
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.727562091165027e-05,
+      "loss": 0.4191,
+      "step": 23153
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.726944028652583e-05,
+      "loss": 0.4055,
+      "step": 23154
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.726326066271693e-05,
+      "loss": 0.3677,
+      "step": 23155
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.725708204029839e-05,
+      "loss": 0.4826,
+      "step": 23156
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.725090441934496e-05,
+      "loss": 0.4618,
+      "step": 23157
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7244727799931415e-05,
+      "loss": 0.4161,
+      "step": 23158
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7238552182132538e-05,
+      "loss": 0.4176,
+      "step": 23159
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7232377566023074e-05,
+      "loss": 0.3722,
+      "step": 23160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.722620395167773e-05,
+      "loss": 0.4729,
+      "step": 23161
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.722003133917125e-05,
+      "loss": 0.3868,
+      "step": 23162
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.721385972857831e-05,
+      "loss": 0.4793,
+      "step": 23163
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7207689119973646e-05,
+      "loss": 0.4264,
+      "step": 23164
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.720151951343193e-05,
+      "loss": 0.41,
+      "step": 23165
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.719535090902783e-05,
+      "loss": 0.41,
+      "step": 23166
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.718918330683602e-05,
+      "loss": 0.4566,
+      "step": 23167
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.718301670693111e-05,
+      "loss": 0.5274,
+      "step": 23168
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.717685110938778e-05,
+      "loss": 0.3862,
+      "step": 23169
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.717068651428064e-05,
+      "loss": 0.4424,
+      "step": 23170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716452292168431e-05,
+      "loss": 0.3809,
+      "step": 23171
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7158360331673374e-05,
+      "loss": 0.4567,
+      "step": 23172
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7152198744322433e-05,
+      "loss": 0.4108,
+      "step": 23173
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7146038159706025e-05,
+      "loss": 0.3992,
+      "step": 23174
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713987857789877e-05,
+      "loss": 0.3648,
+      "step": 23175
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7133719998975195e-05,
+      "loss": 0.4051,
+      "step": 23176
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7127562423009834e-05,
+      "loss": 0.4151,
+      "step": 23177
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712140585007722e-05,
+      "loss": 0.4201,
+      "step": 23178
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711525028025185e-05,
+      "loss": 0.3668,
+      "step": 23179
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7109095713608268e-05,
+      "loss": 0.3141,
+      "step": 23180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7102942150220934e-05,
+      "loss": 0.4441,
+      "step": 23181
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709678959016433e-05,
+      "loss": 0.4586,
+      "step": 23182
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709063803351294e-05,
+      "loss": 0.4845,
+      "step": 23183
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7084487480341173e-05,
+      "loss": 0.407,
+      "step": 23184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.707833793072352e-05,
+      "loss": 0.443,
+      "step": 23185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7072189384734395e-05,
+      "loss": 0.4656,
+      "step": 23186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.706604184244822e-05,
+      "loss": 0.4132,
+      "step": 23187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7059895303939388e-05,
+      "loss": 0.4263,
+      "step": 23188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7053749769282278e-05,
+      "loss": 0.3895,
+      "step": 23189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.704760523855132e-05,
+      "loss": 0.4796,
+      "step": 23190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.704146171182086e-05,
+      "loss": 0.3438,
+      "step": 23191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7035319189165244e-05,
+      "loss": 0.3906,
+      "step": 23192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7029177670658824e-05,
+      "loss": 0.4123,
+      "step": 23193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.702303715637591e-05,
+      "loss": 0.4184,
+      "step": 23194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7016897646390874e-05,
+      "loss": 0.3506,
+      "step": 23195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7010759140778e-05,
+      "loss": 0.4005,
+      "step": 23196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7004621639611574e-05,
+      "loss": 0.3815,
+      "step": 23197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6998485142965882e-05,
+      "loss": 0.4438,
+      "step": 23198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6992349650915185e-05,
+      "loss": 0.4761,
+      "step": 23199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.698621516353378e-05,
+      "loss": 0.4053,
+      "step": 23200
+    },
+    {
+      "epoch": 0.81,
+      "eval_loss": 0.4113035500049591,
+      "eval_runtime": 853.1364,
+      "eval_samples_per_second": 28.202,
+      "eval_steps_per_second": 14.101,
+      "step": 23200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.69800816808959e-05,
+      "loss": 0.4819,
+      "step": 23201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6973949203075767e-05,
+      "loss": 0.3742,
+      "step": 23202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.696781773014762e-05,
+      "loss": 0.4119,
+      "step": 23203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.696168726218562e-05,
+      "loss": 0.3209,
+      "step": 23204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6955557799264043e-05,
+      "loss": 0.375,
+      "step": 23205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6949429341457047e-05,
+      "loss": 0.4142,
+      "step": 23206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6943301888838804e-05,
+      "loss": 0.4568,
+      "step": 23207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6937175441483455e-05,
+      "loss": 0.349,
+      "step": 23208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6931049999465188e-05,
+      "loss": 0.4525,
+      "step": 23209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6924925562858086e-05,
+      "loss": 0.4023,
+      "step": 23210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6918802131736344e-05,
+      "loss": 0.4305,
+      "step": 23211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.691267970617404e-05,
+      "loss": 0.4525,
+      "step": 23212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6906558286245278e-05,
+      "loss": 0.4246,
+      "step": 23213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6900437872024146e-05,
+      "loss": 0.3774,
+      "step": 23214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.68943184635847e-05,
+      "loss": 0.5068,
+      "step": 23215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.688820006100106e-05,
+      "loss": 0.4558,
+      "step": 23216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6882082664347242e-05,
+      "loss": 0.478,
+      "step": 23217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6875966273697297e-05,
+      "loss": 0.3658,
+      "step": 23218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.686985088912524e-05,
+      "loss": 0.3578,
+      "step": 23219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6863736510705086e-05,
+      "loss": 0.4063,
+      "step": 23220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.685762313851087e-05,
+      "loss": 0.4353,
+      "step": 23221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6851510772616563e-05,
+      "loss": 0.4648,
+      "step": 23222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6845399413096153e-05,
+      "loss": 0.4149,
+      "step": 23223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.683928906002359e-05,
+      "loss": 0.5241,
+      "step": 23224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6833179713472825e-05,
+      "loss": 0.5278,
+      "step": 23225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6827071373517844e-05,
+      "loss": 0.4163,
+      "step": 23226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6820964040232556e-05,
+      "loss": 0.4158,
+      "step": 23227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.681485771369087e-05,
+      "loss": 0.4008,
+      "step": 23228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.68087523939667e-05,
+      "loss": 0.3987,
+      "step": 23229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.680264808113391e-05,
+      "loss": 0.4051,
+      "step": 23230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6796544775266442e-05,
+      "loss": 0.469,
+      "step": 23231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.679044247643814e-05,
+      "loss": 0.4267,
+      "step": 23232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6784341184722853e-05,
+      "loss": 0.4091,
+      "step": 23233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6778240900194443e-05,
+      "loss": 0.4005,
+      "step": 23234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6772141622926695e-05,
+      "loss": 0.4306,
+      "step": 23235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.67660433529935e-05,
+      "loss": 0.5063,
+      "step": 23236
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6759946090468636e-05,
+      "loss": 0.3505,
+      "step": 23237
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6753849835425905e-05,
+      "loss": 0.375,
+      "step": 23238
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.674775458793909e-05,
+      "loss": 0.4981,
+      "step": 23239
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6741660348081955e-05,
+      "loss": 0.3557,
+      "step": 23240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6735567115928242e-05,
+      "loss": 0.362,
+      "step": 23241
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6729474891551754e-05,
+      "loss": 0.426,
+      "step": 23242
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6723383675026194e-05,
+      "loss": 0.5215,
+      "step": 23243
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6717293466425287e-05,
+      "loss": 0.5077,
+      "step": 23244
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6711204265822755e-05,
+      "loss": 0.4098,
+      "step": 23245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.670511607329226e-05,
+      "loss": 0.3917,
+      "step": 23246
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6699028888907553e-05,
+      "loss": 0.3377,
+      "step": 23247
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6692942712742266e-05,
+      "loss": 0.4433,
+      "step": 23248
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.668685754487008e-05,
+      "loss": 0.466,
+      "step": 23249
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6680773385364633e-05,
+      "loss": 0.4474,
+      "step": 23250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.667469023429955e-05,
+      "loss": 0.5032,
+      "step": 23251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6668608091748495e-05,
+      "loss": 0.3516,
+      "step": 23252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6662526957785063e-05,
+      "loss": 0.3942,
+      "step": 23253
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6656446832482864e-05,
+      "loss": 0.4609,
+      "step": 23254
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.665036771591547e-05,
+      "loss": 0.4393,
+      "step": 23255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6644289608156472e-05,
+      "loss": 0.4167,
+      "step": 23256
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6638212509279426e-05,
+      "loss": 0.4086,
+      "step": 23257
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6632136419357902e-05,
+      "loss": 0.4411,
+      "step": 23258
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6626061338465428e-05,
+      "loss": 0.3321,
+      "step": 23259
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6619987266675497e-05,
+      "loss": 0.4421,
+      "step": 23260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6613914204061698e-05,
+      "loss": 0.3926,
+      "step": 23261
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6607842150697507e-05,
+      "loss": 0.4632,
+      "step": 23262
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6601771106656395e-05,
+      "loss": 0.4428,
+      "step": 23263
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6595701072011872e-05,
+      "loss": 0.3902,
+      "step": 23264
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6589632046837356e-05,
+      "loss": 0.5317,
+      "step": 23265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6583564031206357e-05,
+      "loss": 0.3773,
+      "step": 23266
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.65774970251923e-05,
+      "loss": 0.4619,
+      "step": 23267
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.657143102886862e-05,
+      "loss": 0.3584,
+      "step": 23268
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.656536604230873e-05,
+      "loss": 0.5068,
+      "step": 23269
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6559302065586003e-05,
+      "loss": 0.4289,
+      "step": 23270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6553239098773897e-05,
+      "loss": 0.4469,
+      "step": 23271
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6547177141945758e-05,
+      "loss": 0.4271,
+      "step": 23272
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6541116195174965e-05,
+      "loss": 0.4224,
+      "step": 23273
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6535056258534876e-05,
+      "loss": 0.3503,
+      "step": 23274
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.652899733209883e-05,
+      "loss": 0.4453,
+      "step": 23275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.652293941594014e-05,
+      "loss": 0.3787,
+      "step": 23276
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.651688251013217e-05,
+      "loss": 0.4148,
+      "step": 23277
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.651082661474821e-05,
+      "loss": 0.3917,
+      "step": 23278
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6504771729861557e-05,
+      "loss": 0.5023,
+      "step": 23279
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6498717855545488e-05,
+      "loss": 0.4017,
+      "step": 23280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6492664991873264e-05,
+      "loss": 0.3935,
+      "step": 23281
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6486613138918172e-05,
+      "loss": 0.4202,
+      "step": 23282
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6480562296753465e-05,
+      "loss": 0.3968,
+      "step": 23283
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6474512465452352e-05,
+      "loss": 0.4231,
+      "step": 23284
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6468463645088074e-05,
+      "loss": 0.4507,
+      "step": 23285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6462415835733803e-05,
+      "loss": 0.3914,
+      "step": 23286
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6456369037462792e-05,
+      "loss": 0.4247,
+      "step": 23287
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.645032325034821e-05,
+      "loss": 0.5446,
+      "step": 23288
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6444278474463216e-05,
+      "loss": 0.4034,
+      "step": 23289
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6438234709880984e-05,
+      "loss": 0.4396,
+      "step": 23290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6432191956674647e-05,
+      "loss": 0.4708,
+      "step": 23291
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.642615021491737e-05,
+      "loss": 0.4451,
+      "step": 23292
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6420109484682268e-05,
+      "loss": 0.4165,
+      "step": 23293
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.641406976604245e-05,
+      "loss": 0.3654,
+      "step": 23294
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6408031059071015e-05,
+      "loss": 0.4157,
+      "step": 23295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6401993363841038e-05,
+      "loss": 0.5003,
+      "step": 23296
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.639595668042563e-05,
+      "loss": 0.4324,
+      "step": 23297
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6389921008897835e-05,
+      "loss": 0.4379,
+      "step": 23298
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.63838863493307e-05,
+      "loss": 0.4068,
+      "step": 23299
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.637785270179728e-05,
+      "loss": 0.3854,
+      "step": 23300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.637182006637057e-05,
+      "loss": 0.4591,
+      "step": 23301
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6365788443123632e-05,
+      "loss": 0.3821,
+      "step": 23302
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.635975783212943e-05,
+      "loss": 0.4051,
+      "step": 23303
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6353728233460985e-05,
+      "loss": 0.4276,
+      "step": 23304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.634769964719125e-05,
+      "loss": 0.3739,
+      "step": 23305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.634167207339318e-05,
+      "loss": 0.3645,
+      "step": 23306
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6335645512139762e-05,
+      "loss": 0.3864,
+      "step": 23307
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.632961996350394e-05,
+      "loss": 0.3993,
+      "step": 23308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6323595427558624e-05,
+      "loss": 0.4305,
+      "step": 23309
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6317571904376727e-05,
+      "loss": 0.4321,
+      "step": 23310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6311549394031167e-05,
+      "loss": 0.3449,
+      "step": 23311
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.630552789659481e-05,
+      "loss": 0.4322,
+      "step": 23312
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6299507412140568e-05,
+      "loss": 0.4543,
+      "step": 23313
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6293487940741304e-05,
+      "loss": 0.3846,
+      "step": 23314
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6287469482469873e-05,
+      "loss": 0.5142,
+      "step": 23315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6281452037399102e-05,
+      "loss": 0.4482,
+      "step": 23316
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6275435605601817e-05,
+      "loss": 0.4573,
+      "step": 23317
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6269420187150885e-05,
+      "loss": 0.4329,
+      "step": 23318
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6263405782119067e-05,
+      "loss": 0.4004,
+      "step": 23319
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6257392390579173e-05,
+      "loss": 0.3925,
+      "step": 23320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.625138001260399e-05,
+      "loss": 0.3622,
+      "step": 23321
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6245368648266245e-05,
+      "loss": 0.534,
+      "step": 23322
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6239358297638763e-05,
+      "loss": 0.5085,
+      "step": 23323
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.623334896079426e-05,
+      "loss": 0.4321,
+      "step": 23324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6227340637805454e-05,
+      "loss": 0.4375,
+      "step": 23325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.622133332874509e-05,
+      "loss": 0.5231,
+      "step": 23326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6215327033685834e-05,
+      "loss": 0.4161,
+      "step": 23327
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6209321752700445e-05,
+      "loss": 0.4269,
+      "step": 23328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.620331748586157e-05,
+      "loss": 0.3896,
+      "step": 23329
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.619731423324189e-05,
+      "loss": 0.4108,
+      "step": 23330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.619131199491405e-05,
+      "loss": 0.4004,
+      "step": 23331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.618531077095069e-05,
+      "loss": 0.3578,
+      "step": 23332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6179310561424486e-05,
+      "loss": 0.4291,
+      "step": 23333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6173311366408028e-05,
+      "loss": 0.4484,
+      "step": 23334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6167313185973943e-05,
+      "loss": 0.5071,
+      "step": 23335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6161316020194818e-05,
+      "loss": 0.3566,
+      "step": 23336
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6155319869143214e-05,
+      "loss": 0.4977,
+      "step": 23337
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6149324732891768e-05,
+      "loss": 0.3972,
+      "step": 23338
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6143330611512997e-05,
+      "loss": 0.3589,
+      "step": 23339
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.613733750507945e-05,
+      "loss": 0.4268,
+      "step": 23340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.613134541366368e-05,
+      "loss": 0.4806,
+      "step": 23341
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6125354337338204e-05,
+      "loss": 0.4627,
+      "step": 23342
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6119364276175497e-05,
+      "loss": 0.4539,
+      "step": 23343
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6113375230248128e-05,
+      "loss": 0.4495,
+      "step": 23344
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6107387199628554e-05,
+      "loss": 0.3871,
+      "step": 23345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6101400184389238e-05,
+      "loss": 0.4866,
+      "step": 23346
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6095414184602652e-05,
+      "loss": 0.4161,
+      "step": 23347
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.608942920034122e-05,
+      "loss": 0.4425,
+      "step": 23348
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6083445231677442e-05,
+      "loss": 0.4631,
+      "step": 23349
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6077462278683698e-05,
+      "loss": 0.4144,
+      "step": 23350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.607148034143242e-05,
+      "loss": 0.4597,
+      "step": 23351
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6065499419996e-05,
+      "loss": 0.453,
+      "step": 23352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.60595195144468e-05,
+      "loss": 0.4078,
+      "step": 23353
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6053540624857243e-05,
+      "loss": 0.4117,
+      "step": 23354
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6047562751299682e-05,
+      "loss": 0.4036,
+      "step": 23355
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6041585893846467e-05,
+      "loss": 0.4536,
+      "step": 23356
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6035610052569938e-05,
+      "loss": 0.3826,
+      "step": 23357
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6029635227542383e-05,
+      "loss": 0.4501,
+      "step": 23358
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6023661418836177e-05,
+      "loss": 0.4729,
+      "step": 23359
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6017688626523608e-05,
+      "loss": 0.4033,
+      "step": 23360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.601171685067695e-05,
+      "loss": 0.4177,
+      "step": 23361
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.600574609136849e-05,
+      "loss": 0.3551,
+      "step": 23362
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.599977634867047e-05,
+      "loss": 0.483,
+      "step": 23363
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.59938076226552e-05,
+      "loss": 0.4004,
+      "step": 23364
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5987839913394875e-05,
+      "loss": 0.4576,
+      "step": 23365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.598187322096174e-05,
+      "loss": 0.3458,
+      "step": 23366
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5975907545428015e-05,
+      "loss": 0.4008,
+      "step": 23367
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5969942886865863e-05,
+      "loss": 0.4022,
+      "step": 23368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.596397924534755e-05,
+      "loss": 0.4017,
+      "step": 23369
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5958016620945216e-05,
+      "loss": 0.3419,
+      "step": 23370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5952055013731027e-05,
+      "loss": 0.43,
+      "step": 23371
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5946094423777136e-05,
+      "loss": 0.4224,
+      "step": 23372
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5940134851155697e-05,
+      "loss": 0.4541,
+      "step": 23373
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.593417629593881e-05,
+      "loss": 0.3177,
+      "step": 23374
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5928218758198643e-05,
+      "loss": 0.4499,
+      "step": 23375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.592226223800728e-05,
+      "loss": 0.4849,
+      "step": 23376
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5916306735436816e-05,
+      "loss": 0.4677,
+      "step": 23377
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5910352250559323e-05,
+      "loss": 0.4765,
+      "step": 23378
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5904398783446862e-05,
+      "loss": 0.4641,
+      "step": 23379
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5898446334171522e-05,
+      "loss": 0.405,
+      "step": 23380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.589249490280533e-05,
+      "loss": 0.4059,
+      "step": 23381
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5886544489420317e-05,
+      "loss": 0.3799,
+      "step": 23382
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5880595094088502e-05,
+      "loss": 0.4701,
+      "step": 23383
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.587464671688187e-05,
+      "loss": 0.4075,
+      "step": 23384
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5868699357872464e-05,
+      "loss": 0.3672,
+      "step": 23385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.586275301713225e-05,
+      "loss": 0.4263,
+      "step": 23386
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.585680769473319e-05,
+      "loss": 0.4339,
+      "step": 23387
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.585086339074724e-05,
+      "loss": 0.4103,
+      "step": 23388
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.584492010524633e-05,
+      "loss": 0.4166,
+      "step": 23389
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5838977838302425e-05,
+      "loss": 0.4131,
+      "step": 23390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.583303658998747e-05,
+      "loss": 0.4051,
+      "step": 23391
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5827096360373317e-05,
+      "loss": 0.442,
+      "step": 23392
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5821157149531862e-05,
+      "loss": 0.3794,
+      "step": 23393
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.581521895753503e-05,
+      "loss": 0.3908,
+      "step": 23394
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5809281784454667e-05,
+      "loss": 0.4071,
+      "step": 23395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5803345630362653e-05,
+      "loss": 0.3537,
+      "step": 23396
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.579741049533081e-05,
+      "loss": 0.5029,
+      "step": 23397
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.579147637943097e-05,
+      "loss": 0.4027,
+      "step": 23398
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5785543282734994e-05,
+      "loss": 0.3946,
+      "step": 23399
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.577961120531467e-05,
+      "loss": 0.47,
+      "step": 23400
+    },
+    {
+      "epoch": 0.82,
+      "eval_loss": 0.4098736345767975,
+      "eval_runtime": 846.1331,
+      "eval_samples_per_second": 28.435,
+      "eval_steps_per_second": 14.218,
+      "step": 23400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5773680147241787e-05,
+      "loss": 0.3759,
+      "step": 23401
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.576775010858814e-05,
+      "loss": 0.3994,
+      "step": 23402
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.576182108942549e-05,
+      "loss": 0.334,
+      "step": 23403
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5755893089825625e-05,
+      "loss": 0.4254,
+      "step": 23404
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5749966109860272e-05,
+      "loss": 0.5122,
+      "step": 23405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.574404014960117e-05,
+      "loss": 0.4383,
+      "step": 23406
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5738115209120053e-05,
+      "loss": 0.3647,
+      "step": 23407
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5732191288488617e-05,
+      "loss": 0.4144,
+      "step": 23408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5726268387778552e-05,
+      "loss": 0.3747,
+      "step": 23409
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5720346507061588e-05,
+      "loss": 0.409,
+      "step": 23410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.571442564640937e-05,
+      "loss": 0.3833,
+      "step": 23411
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5708505805893557e-05,
+      "loss": 0.4099,
+      "step": 23412
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5702586985585822e-05,
+      "loss": 0.5119,
+      "step": 23413
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5696669185557743e-05,
+      "loss": 0.4299,
+      "step": 23414
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5690752405881015e-05,
+      "loss": 0.3781,
+      "step": 23415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5684836646627233e-05,
+      "loss": 0.5144,
+      "step": 23416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5678921907867982e-05,
+      "loss": 0.4629,
+      "step": 23417
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5673008189674854e-05,
+      "loss": 0.3525,
+      "step": 23418
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.566709549211941e-05,
+      "loss": 0.4923,
+      "step": 23419
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.566118381527325e-05,
+      "loss": 0.4727,
+      "step": 23420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5655273159207905e-05,
+      "loss": 0.4423,
+      "step": 23421
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.564936352399491e-05,
+      "loss": 0.4676,
+      "step": 23422
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.56434549097058e-05,
+      "loss": 0.6162,
+      "step": 23423
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.563754731641206e-05,
+      "loss": 0.4426,
+      "step": 23424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.563164074418524e-05,
+      "loss": 0.5182,
+      "step": 23425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.56257351930968e-05,
+      "loss": 0.4847,
+      "step": 23426
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5619830663218227e-05,
+      "loss": 0.365,
+      "step": 23427
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.561392715462098e-05,
+      "loss": 0.4087,
+      "step": 23428
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5608024667376476e-05,
+      "loss": 0.4206,
+      "step": 23429
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.560212320155623e-05,
+      "loss": 0.4156,
+      "step": 23430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5596222757231617e-05,
+      "loss": 0.4279,
+      "step": 23431
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5590323334474065e-05,
+      "loss": 0.4379,
+      "step": 23432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.558442493335498e-05,
+      "loss": 0.4574,
+      "step": 23433
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.557852755394573e-05,
+      "loss": 0.446,
+      "step": 23434
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5572631196317723e-05,
+      "loss": 0.4486,
+      "step": 23435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5566735860542327e-05,
+      "loss": 0.3567,
+      "step": 23436
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5560841546690874e-05,
+      "loss": 0.3882,
+      "step": 23437
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.555494825483471e-05,
+      "loss": 0.3929,
+      "step": 23438
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.554905598504515e-05,
+      "loss": 0.3973,
+      "step": 23439
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5543164737393533e-05,
+      "loss": 0.3727,
+      "step": 23440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5537274511951162e-05,
+      "loss": 0.4122,
+      "step": 23441
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5531385308789324e-05,
+      "loss": 0.4785,
+      "step": 23442
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5525497127979307e-05,
+      "loss": 0.4409,
+      "step": 23443
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.551960996959235e-05,
+      "loss": 0.5164,
+      "step": 23444
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.55137238336997e-05,
+      "loss": 0.4204,
+      "step": 23445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.550783872037266e-05,
+      "loss": 0.498,
+      "step": 23446
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5501954629682415e-05,
+      "loss": 0.4468,
+      "step": 23447
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5496071561700188e-05,
+      "loss": 0.4251,
+      "step": 23448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5490189516497177e-05,
+      "loss": 0.3708,
+      "step": 23449
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5484308494144573e-05,
+      "loss": 0.3976,
+      "step": 23450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.547842849471358e-05,
+      "loss": 0.3852,
+      "step": 23451
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.547254951827535e-05,
+      "loss": 0.3638,
+      "step": 23452
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5466671564901037e-05,
+      "loss": 0.4625,
+      "step": 23453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.546079463466179e-05,
+      "loss": 0.4845,
+      "step": 23454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5454918727628708e-05,
+      "loss": 0.4639,
+      "step": 23455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5449043843872967e-05,
+      "loss": 0.4544,
+      "step": 23456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.544316998346563e-05,
+      "loss": 0.3933,
+      "step": 23457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5437297146477815e-05,
+      "loss": 0.406,
+      "step": 23458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5431425332980586e-05,
+      "loss": 0.4348,
+      "step": 23459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5425554543044997e-05,
+      "loss": 0.4007,
+      "step": 23460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.541968477674214e-05,
+      "loss": 0.4377,
+      "step": 23461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5413816034143036e-05,
+      "loss": 0.3306,
+      "step": 23462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.540794831531873e-05,
+      "loss": 0.3789,
+      "step": 23463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5402081620340226e-05,
+      "loss": 0.3281,
+      "step": 23464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5396215949278526e-05,
+      "loss": 0.4963,
+      "step": 23465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5390351302204642e-05,
+      "loss": 0.4687,
+      "step": 23466
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5384487679189563e-05,
+      "loss": 0.3828,
+      "step": 23467
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5378625080304243e-05,
+      "loss": 0.3987,
+      "step": 23468
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5372763505619636e-05,
+      "loss": 0.3749,
+      "step": 23469
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5366902955206663e-05,
+      "loss": 0.4675,
+      "step": 23470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5361043429136322e-05,
+      "loss": 0.4393,
+      "step": 23471
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5355184927479483e-05,
+      "loss": 0.4628,
+      "step": 23472
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.534932745030707e-05,
+      "loss": 0.417,
+      "step": 23473
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.534347099768998e-05,
+      "loss": 0.3981,
+      "step": 23474
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.533761556969908e-05,
+      "loss": 0.4022,
+      "step": 23475
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5331761166405224e-05,
+      "loss": 0.5177,
+      "step": 23476
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5325907787879325e-05,
+      "loss": 0.3454,
+      "step": 23477
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5320055434192205e-05,
+      "loss": 0.3471,
+      "step": 23478
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5314204105414674e-05,
+      "loss": 0.4117,
+      "step": 23479
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.530835380161758e-05,
+      "loss": 0.4091,
+      "step": 23480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5302504522871696e-05,
+      "loss": 0.345,
+      "step": 23481
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5296656269247866e-05,
+      "loss": 0.4508,
+      "step": 23482
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5290809040816845e-05,
+      "loss": 0.3925,
+      "step": 23483
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5284962837649407e-05,
+      "loss": 0.3902,
+      "step": 23484
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5279117659816322e-05,
+      "loss": 0.3873,
+      "step": 23485
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5273273507388298e-05,
+      "loss": 0.3725,
+      "step": 23486
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.526743038043612e-05,
+      "loss": 0.438,
+      "step": 23487
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.526158827903048e-05,
+      "loss": 0.4755,
+      "step": 23488
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5255747203242099e-05,
+      "loss": 0.4417,
+      "step": 23489
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5249907153141673e-05,
+      "loss": 0.4564,
+      "step": 23490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5244068128799848e-05,
+      "loss": 0.4304,
+      "step": 23491
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5238230130287357e-05,
+      "loss": 0.4602,
+      "step": 23492
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5232393157674828e-05,
+      "loss": 0.4405,
+      "step": 23493
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.522655721103291e-05,
+      "loss": 0.4344,
+      "step": 23494
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5220722290432243e-05,
+      "loss": 0.4298,
+      "step": 23495
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5214888395943416e-05,
+      "loss": 0.4508,
+      "step": 23496
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5209055527637084e-05,
+      "loss": 0.4243,
+      "step": 23497
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5203223685583834e-05,
+      "loss": 0.4957,
+      "step": 23498
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5197392869854244e-05,
+      "loss": 0.411,
+      "step": 23499
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5191563080518878e-05,
+      "loss": 0.3957,
+      "step": 23500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5185734317648293e-05,
+      "loss": 0.3933,
+      "step": 23501
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5179906581313064e-05,
+      "loss": 0.3125,
+      "step": 23502
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5174079871583701e-05,
+      "loss": 0.3992,
+      "step": 23503
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.516825418853074e-05,
+      "loss": 0.3985,
+      "step": 23504
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5162429532224687e-05,
+      "loss": 0.3631,
+      "step": 23505
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5156605902736021e-05,
+      "loss": 0.393,
+      "step": 23506
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5150783300135263e-05,
+      "loss": 0.5156,
+      "step": 23507
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5144961724492857e-05,
+      "loss": 0.4645,
+      "step": 23508
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.513914117587929e-05,
+      "loss": 0.4526,
+      "step": 23509
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5133321654364985e-05,
+      "loss": 0.4049,
+      "step": 23510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5127503160020396e-05,
+      "loss": 0.4748,
+      "step": 23511
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5121685692915899e-05,
+      "loss": 0.4292,
+      "step": 23512
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5115869253121973e-05,
+      "loss": 0.353,
+      "step": 23513
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5110053840708993e-05,
+      "loss": 0.4216,
+      "step": 23514
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5104239455747326e-05,
+      "loss": 0.4771,
+      "step": 23515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5098426098307373e-05,
+      "loss": 0.3614,
+      "step": 23516
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5092613768459441e-05,
+      "loss": 0.4662,
+      "step": 23517
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5086802466273941e-05,
+      "loss": 0.4265,
+      "step": 23518
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5080992191821197e-05,
+      "loss": 0.4082,
+      "step": 23519
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5075182945171518e-05,
+      "loss": 0.4489,
+      "step": 23520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5069374726395214e-05,
+      "loss": 0.4199,
+      "step": 23521
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5063567535562562e-05,
+      "loss": 0.4726,
+      "step": 23522
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5057761372743894e-05,
+      "loss": 0.4545,
+      "step": 23523
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5051956238009469e-05,
+      "loss": 0.3922,
+      "step": 23524
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.504615213142957e-05,
+      "loss": 0.5574,
+      "step": 23525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5040349053074376e-05,
+      "loss": 0.4519,
+      "step": 23526
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5034547003014177e-05,
+      "loss": 0.3893,
+      "step": 23527
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5028745981319203e-05,
+      "loss": 0.4235,
+      "step": 23528
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5022945988059633e-05,
+      "loss": 0.3933,
+      "step": 23529
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.50171470233057e-05,
+      "loss": 0.5547,
+      "step": 23530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5011349087127546e-05,
+      "loss": 0.3986,
+      "step": 23531
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5005552179595394e-05,
+      "loss": 0.4107,
+      "step": 23532
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4999756300779378e-05,
+      "loss": 0.5282,
+      "step": 23533
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4993961450749661e-05,
+      "loss": 0.4076,
+      "step": 23534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4988167629576377e-05,
+      "loss": 0.4156,
+      "step": 23535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4982374837329627e-05,
+      "loss": 0.4165,
+      "step": 23536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4976583074079564e-05,
+      "loss": 0.3789,
+      "step": 23537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4970792339896267e-05,
+      "loss": 0.3866,
+      "step": 23538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4965002634849812e-05,
+      "loss": 0.4794,
+      "step": 23539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.49592139590103e-05,
+      "loss": 0.4375,
+      "step": 23540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.495342631244775e-05,
+      "loss": 0.4336,
+      "step": 23541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4947639695232264e-05,
+      "loss": 0.381,
+      "step": 23542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.494185410743385e-05,
+      "loss": 0.5332,
+      "step": 23543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4936069549122545e-05,
+      "loss": 0.4238,
+      "step": 23544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4930286020368356e-05,
+      "loss": 0.3444,
+      "step": 23545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4924503521241273e-05,
+      "loss": 0.3826,
+      "step": 23546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4918722051811274e-05,
+      "loss": 0.3505,
+      "step": 23547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.491294161214838e-05,
+      "loss": 0.4287,
+      "step": 23548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4907162202322521e-05,
+      "loss": 0.4356,
+      "step": 23549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4901383822403658e-05,
+      "loss": 0.43,
+      "step": 23550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.489560647246171e-05,
+      "loss": 0.4077,
+      "step": 23551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4889830152566609e-05,
+      "loss": 0.4813,
+      "step": 23552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4884054862788276e-05,
+      "loss": 0.4137,
+      "step": 23553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4878280603196626e-05,
+      "loss": 0.4033,
+      "step": 23554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4872507373861521e-05,
+      "loss": 0.2943,
+      "step": 23555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4866735174852852e-05,
+      "loss": 0.4083,
+      "step": 23556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4860964006240441e-05,
+      "loss": 0.3697,
+      "step": 23557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4855193868094209e-05,
+      "loss": 0.4764,
+      "step": 23558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4849424760483954e-05,
+      "loss": 0.4079,
+      "step": 23559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.48436566834795e-05,
+      "loss": 0.4828,
+      "step": 23560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4837889637150659e-05,
+      "loss": 0.3639,
+      "step": 23561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4832123621567218e-05,
+      "loss": 0.4861,
+      "step": 23562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.482635863679901e-05,
+      "loss": 0.3396,
+      "step": 23563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.482059468291579e-05,
+      "loss": 0.5079,
+      "step": 23564
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4814831759987302e-05,
+      "loss": 0.5681,
+      "step": 23565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4809069868083326e-05,
+      "loss": 0.4796,
+      "step": 23566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4803309007273547e-05,
+      "loss": 0.3949,
+      "step": 23567
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4797549177627757e-05,
+      "loss": 0.3775,
+      "step": 23568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4791790379215641e-05,
+      "loss": 0.4279,
+      "step": 23569
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4786032612106904e-05,
+      "loss": 0.4343,
+      "step": 23570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4780275876371219e-05,
+      "loss": 0.3669,
+      "step": 23571
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4774520172078254e-05,
+      "loss": 0.4534,
+      "step": 23572
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.476876549929771e-05,
+      "loss": 0.4217,
+      "step": 23573
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4763011858099218e-05,
+      "loss": 0.5028,
+      "step": 23574
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4757259248552413e-05,
+      "loss": 0.4272,
+      "step": 23575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4751507670726927e-05,
+      "loss": 0.4464,
+      "step": 23576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4745757124692372e-05,
+      "loss": 0.4629,
+      "step": 23577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4740007610518325e-05,
+      "loss": 0.4257,
+      "step": 23578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4734259128274409e-05,
+      "loss": 0.4181,
+      "step": 23579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.47285116780302e-05,
+      "loss": 0.357,
+      "step": 23580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.472276525985523e-05,
+      "loss": 0.5176,
+      "step": 23581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4717019873819082e-05,
+      "loss": 0.4095,
+      "step": 23582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4711275519991252e-05,
+      "loss": 0.3834,
+      "step": 23583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4705532198441318e-05,
+      "loss": 0.462,
+      "step": 23584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.469978990923877e-05,
+      "loss": 0.4345,
+      "step": 23585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4694048652453108e-05,
+      "loss": 0.4105,
+      "step": 23586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4688308428153818e-05,
+      "loss": 0.3301,
+      "step": 23587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4682569236410359e-05,
+      "loss": 0.4872,
+      "step": 23588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.467683107729224e-05,
+      "loss": 0.4112,
+      "step": 23589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4671093950868875e-05,
+      "loss": 0.5175,
+      "step": 23590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4665357857209715e-05,
+      "loss": 0.5291,
+      "step": 23591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4659622796384187e-05,
+      "loss": 0.5422,
+      "step": 23592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4653888768461688e-05,
+      "loss": 0.4725,
+      "step": 23593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4648155773511641e-05,
+      "loss": 0.4118,
+      "step": 23594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4642423811603423e-05,
+      "loss": 0.4158,
+      "step": 23595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4636692882806425e-05,
+      "loss": 0.4961,
+      "step": 23596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4630962987189989e-05,
+      "loss": 0.3968,
+      "step": 23597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.462523412482345e-05,
+      "loss": 0.4265,
+      "step": 23598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4619506295776198e-05,
+      "loss": 0.4286,
+      "step": 23599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.461377950011753e-05,
+      "loss": 0.3918,
+      "step": 23600
+    },
+    {
+      "epoch": 0.83,
+      "eval_loss": 0.4085943102836609,
+      "eval_runtime": 853.7343,
+      "eval_samples_per_second": 28.182,
+      "eval_steps_per_second": 14.091,
+      "step": 23600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.460805373791676e-05,
+      "loss": 0.4109,
+      "step": 23601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4602329009243188e-05,
+      "loss": 0.4076,
+      "step": 23602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.459660531416609e-05,
+      "loss": 0.345,
+      "step": 23603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4590882652754767e-05,
+      "loss": 0.4603,
+      "step": 23604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4585161025078476e-05,
+      "loss": 0.4883,
+      "step": 23605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.457944043120646e-05,
+      "loss": 0.5209,
+      "step": 23606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4573720871207963e-05,
+      "loss": 0.4024,
+      "step": 23607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4568002345152188e-05,
+      "loss": 0.4018,
+      "step": 23608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4562284853108388e-05,
+      "loss": 0.463,
+      "step": 23609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4556568395145742e-05,
+      "loss": 0.4387,
+      "step": 23610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.455085297133344e-05,
+      "loss": 0.3721,
+      "step": 23611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4545138581740657e-05,
+      "loss": 0.3754,
+      "step": 23612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4539425226436554e-05,
+      "loss": 0.3868,
+      "step": 23613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.453371290549026e-05,
+      "loss": 0.3835,
+      "step": 23614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4528001618970966e-05,
+      "loss": 0.4054,
+      "step": 23615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4522291366947772e-05,
+      "loss": 0.3807,
+      "step": 23616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4516582149489789e-05,
+      "loss": 0.452,
+      "step": 23617
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4510873966666105e-05,
+      "loss": 0.5086,
+      "step": 23618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4505166818545801e-05,
+      "loss": 0.3964,
+      "step": 23619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4499460705197998e-05,
+      "loss": 0.3827,
+      "step": 23620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.449375562669173e-05,
+      "loss": 0.3614,
+      "step": 23621
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4488051583096051e-05,
+      "loss": 0.4243,
+      "step": 23622
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4482348574479998e-05,
+      "loss": 0.3809,
+      "step": 23623
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.447664660091258e-05,
+      "loss": 0.3859,
+      "step": 23624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4470945662462832e-05,
+      "loss": 0.3698,
+      "step": 23625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4465245759199764e-05,
+      "loss": 0.4149,
+      "step": 23626
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4459546891192344e-05,
+      "loss": 0.4337,
+      "step": 23627
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4453849058509549e-05,
+      "loss": 0.3789,
+      "step": 23628
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4448152261220316e-05,
+      "loss": 0.4862,
+      "step": 23629
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4442456499393654e-05,
+      "loss": 0.4828,
+      "step": 23630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4436761773098472e-05,
+      "loss": 0.3302,
+      "step": 23631
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4431068082403687e-05,
+      "loss": 0.4006,
+      "step": 23632
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4425375427378218e-05,
+      "loss": 0.3805,
+      "step": 23633
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4419683808090945e-05,
+      "loss": 0.4533,
+      "step": 23634
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.44139932246108e-05,
+      "loss": 0.4269,
+      "step": 23635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.440830367700663e-05,
+      "loss": 0.505,
+      "step": 23636
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.44026151653473e-05,
+      "loss": 0.4861,
+      "step": 23637
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4396927689701666e-05,
+      "loss": 0.426,
+      "step": 23638
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4391241250138532e-05,
+      "loss": 0.4285,
+      "step": 23639
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4385555846726784e-05,
+      "loss": 0.4425,
+      "step": 23640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4379871479535189e-05,
+      "loss": 0.3803,
+      "step": 23641
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4374188148632572e-05,
+      "loss": 0.4424,
+      "step": 23642
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.43685058540877e-05,
+      "loss": 0.4563,
+      "step": 23643
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.436282459596936e-05,
+      "loss": 0.3825,
+      "step": 23644
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4357144374346288e-05,
+      "loss": 0.4005,
+      "step": 23645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4351465189287272e-05,
+      "loss": 0.4515,
+      "step": 23646
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4345787040861036e-05,
+      "loss": 0.3603,
+      "step": 23647
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4340109929136291e-05,
+      "loss": 0.4116,
+      "step": 23648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4334433854181773e-05,
+      "loss": 0.3975,
+      "step": 23649
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4328758816066135e-05,
+      "loss": 0.4479,
+      "step": 23650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4323084814858123e-05,
+      "loss": 0.3671,
+      "step": 23651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4317411850626382e-05,
+      "loss": 0.4711,
+      "step": 23652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4311739923439582e-05,
+      "loss": 0.4601,
+      "step": 23653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4306069033366365e-05,
+      "loss": 0.4394,
+      "step": 23654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4300399180475343e-05,
+      "loss": 0.4018,
+      "step": 23655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4294730364835186e-05,
+      "loss": 0.4125,
+      "step": 23656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.428906258651449e-05,
+      "loss": 0.4827,
+      "step": 23657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4283395845581849e-05,
+      "loss": 0.4239,
+      "step": 23658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.427773014210585e-05,
+      "loss": 0.4178,
+      "step": 23659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4272065476155071e-05,
+      "loss": 0.3898,
+      "step": 23660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.426640184779806e-05,
+      "loss": 0.3635,
+      "step": 23661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4260739257103384e-05,
+      "loss": 0.419,
+      "step": 23662
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4255077704139574e-05,
+      "loss": 0.5038,
+      "step": 23663
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4249417188975123e-05,
+      "loss": 0.3318,
+      "step": 23664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4243757711678585e-05,
+      "loss": 0.4076,
+      "step": 23665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.423809927231845e-05,
+      "loss": 0.4615,
+      "step": 23666
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4232441870963198e-05,
+      "loss": 0.4495,
+      "step": 23667
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4226785507681296e-05,
+      "loss": 0.4461,
+      "step": 23668
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4221130182541188e-05,
+      "loss": 0.4334,
+      "step": 23669
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4215475895611374e-05,
+      "loss": 0.4299,
+      "step": 23670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4209822646960258e-05,
+      "loss": 0.4634,
+      "step": 23671
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4204170436656262e-05,
+      "loss": 0.461,
+      "step": 23672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4198519264767795e-05,
+      "loss": 0.4414,
+      "step": 23673
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.419286913136324e-05,
+      "loss": 0.4641,
+      "step": 23674
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4187220036511028e-05,
+      "loss": 0.3774,
+      "step": 23675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4181571980279506e-05,
+      "loss": 0.3763,
+      "step": 23676
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4175924962737031e-05,
+      "loss": 0.4044,
+      "step": 23677
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4170278983951956e-05,
+      "loss": 0.4067,
+      "step": 23678
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4164634043992619e-05,
+      "loss": 0.4147,
+      "step": 23679
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.415899014292732e-05,
+      "loss": 0.4023,
+      "step": 23680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4153347280824391e-05,
+      "loss": 0.4204,
+      "step": 23681
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4147705457752136e-05,
+      "loss": 0.4056,
+      "step": 23682
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4142064673778832e-05,
+      "loss": 0.3912,
+      "step": 23683
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4136424928972748e-05,
+      "loss": 0.4243,
+      "step": 23684
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4130786223402114e-05,
+      "loss": 0.3629,
+      "step": 23685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4125148557135238e-05,
+      "loss": 0.3739,
+      "step": 23686
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4119511930240325e-05,
+      "loss": 0.369,
+      "step": 23687
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4113876342785592e-05,
+      "loss": 0.4478,
+      "step": 23688
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.410824179483925e-05,
+      "loss": 0.4777,
+      "step": 23689
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4102608286469477e-05,
+      "loss": 0.3685,
+      "step": 23690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4096975817744507e-05,
+      "loss": 0.4359,
+      "step": 23691
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4091344388732475e-05,
+      "loss": 0.3716,
+      "step": 23692
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4085713999501548e-05,
+      "loss": 0.4785,
+      "step": 23693
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4080084650119885e-05,
+      "loss": 0.378,
+      "step": 23694
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4074456340655572e-05,
+      "loss": 0.4452,
+      "step": 23695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4068829071176793e-05,
+      "loss": 0.4651,
+      "step": 23696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4063202841751632e-05,
+      "loss": 0.3169,
+      "step": 23697
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4057577652448172e-05,
+      "loss": 0.5233,
+      "step": 23698
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4051953503334514e-05,
+      "loss": 0.4632,
+      "step": 23699
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4046330394478702e-05,
+      "loss": 0.4712,
+      "step": 23700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4040708325948837e-05,
+      "loss": 0.4266,
+      "step": 23701
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4035087297812944e-05,
+      "loss": 0.4348,
+      "step": 23702
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4029467310139043e-05,
+      "loss": 0.4569,
+      "step": 23703
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4023848362995174e-05,
+      "loss": 0.4786,
+      "step": 23704
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4018230456449322e-05,
+      "loss": 0.4481,
+      "step": 23705
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4012613590569512e-05,
+      "loss": 0.3838,
+      "step": 23706
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4006997765423713e-05,
+      "loss": 0.3587,
+      "step": 23707
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.400138298107989e-05,
+      "loss": 0.3572,
+      "step": 23708
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3995769237606004e-05,
+      "loss": 0.403,
+      "step": 23709
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3990156535069988e-05,
+      "loss": 0.3398,
+      "step": 23710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3984544873539807e-05,
+      "loss": 0.422,
+      "step": 23711
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3978934253083354e-05,
+      "loss": 0.4827,
+      "step": 23712
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.397332467376855e-05,
+      "loss": 0.3814,
+      "step": 23713
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.396771613566329e-05,
+      "loss": 0.4465,
+      "step": 23714
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3962108638835435e-05,
+      "loss": 0.3647,
+      "step": 23715
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.395650218335285e-05,
+      "loss": 0.3332,
+      "step": 23716
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3950896769283439e-05,
+      "loss": 0.4283,
+      "step": 23717
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.394529239669501e-05,
+      "loss": 0.454,
+      "step": 23718
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3939689065655404e-05,
+      "loss": 0.4119,
+      "step": 23719
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3934086776232447e-05,
+      "loss": 0.428,
+      "step": 23720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3928485528493906e-05,
+      "loss": 0.461,
+      "step": 23721
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3922885322507628e-05,
+      "loss": 0.4391,
+      "step": 23722
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.391728615834137e-05,
+      "loss": 0.4078,
+      "step": 23723
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3911688036062897e-05,
+      "loss": 0.4672,
+      "step": 23724
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3906090955739982e-05,
+      "loss": 0.372,
+      "step": 23725
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3900494917440332e-05,
+      "loss": 0.377,
+      "step": 23726
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.389489992123172e-05,
+      "loss": 0.3612,
+      "step": 23727
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3889305967181843e-05,
+      "loss": 0.3722,
+      "step": 23728
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3883713055358415e-05,
+      "loss": 0.4113,
+      "step": 23729
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3878121185829118e-05,
+      "loss": 0.3293,
+      "step": 23730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3872530358661617e-05,
+      "loss": 0.4524,
+      "step": 23731
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3866940573923627e-05,
+      "loss": 0.3967,
+      "step": 23732
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3861351831682767e-05,
+      "loss": 0.408,
+      "step": 23733
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3855764132006699e-05,
+      "loss": 0.4065,
+      "step": 23734
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3850177474963033e-05,
+      "loss": 0.4158,
+      "step": 23735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3844591860619383e-05,
+      "loss": 0.4111,
+      "step": 23736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3839007289043381e-05,
+      "loss": 0.4321,
+      "step": 23737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3833423760302611e-05,
+      "loss": 0.3987,
+      "step": 23738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3827841274464637e-05,
+      "loss": 0.3739,
+      "step": 23739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.382225983159704e-05,
+      "loss": 0.3972,
+      "step": 23740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3816679431767354e-05,
+      "loss": 0.5106,
+      "step": 23741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3811100075043149e-05,
+      "loss": 0.4089,
+      "step": 23742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3805521761491935e-05,
+      "loss": 0.4466,
+      "step": 23743
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3799944491181238e-05,
+      "loss": 0.423,
+      "step": 23744
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3794368264178548e-05,
+      "loss": 0.5015,
+      "step": 23745
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3788793080551376e-05,
+      "loss": 0.3858,
+      "step": 23746
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.378321894036716e-05,
+      "loss": 0.4583,
+      "step": 23747
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.377764584369341e-05,
+      "loss": 0.3799,
+      "step": 23748
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3772073790597572e-05,
+      "loss": 0.3896,
+      "step": 23749
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3766502781147084e-05,
+      "loss": 0.3935,
+      "step": 23750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3760932815409355e-05,
+      "loss": 0.408,
+      "step": 23751
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3755363893451789e-05,
+      "loss": 0.3972,
+      "step": 23752
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3749796015341843e-05,
+      "loss": 0.3529,
+      "step": 23753
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3744229181146873e-05,
+      "loss": 0.3458,
+      "step": 23754
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3738663390934259e-05,
+      "loss": 0.4063,
+      "step": 23755
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.373309864477137e-05,
+      "loss": 0.4751,
+      "step": 23756
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3727534942725529e-05,
+      "loss": 0.3879,
+      "step": 23757
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3721972284864116e-05,
+      "loss": 0.4252,
+      "step": 23758
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3716410671254443e-05,
+      "loss": 0.3455,
+      "step": 23759
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3710850101963835e-05,
+      "loss": 0.378,
+      "step": 23760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3705290577059571e-05,
+      "loss": 0.4591,
+      "step": 23761
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.369973209660893e-05,
+      "loss": 0.4405,
+      "step": 23762
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3694174660679226e-05,
+      "loss": 0.4589,
+      "step": 23763
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3688618269337706e-05,
+      "loss": 0.3567,
+      "step": 23764
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3683062922651624e-05,
+      "loss": 0.4123,
+      "step": 23765
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3677508620688218e-05,
+      "loss": 0.4297,
+      "step": 23766
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3671955363514676e-05,
+      "loss": 0.4536,
+      "step": 23767
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3666403151198271e-05,
+      "loss": 0.4854,
+      "step": 23768
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3660851983806177e-05,
+      "loss": 0.4512,
+      "step": 23769
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.365530186140559e-05,
+      "loss": 0.4988,
+      "step": 23770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3649752784063662e-05,
+      "loss": 0.392,
+      "step": 23771
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3644204751847555e-05,
+      "loss": 0.413,
+      "step": 23772
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3638657764824459e-05,
+      "loss": 0.4144,
+      "step": 23773
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3633111823061473e-05,
+      "loss": 0.416,
+      "step": 23774
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3627566926625745e-05,
+      "loss": 0.4288,
+      "step": 23775
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3622023075584367e-05,
+      "loss": 0.444,
+      "step": 23776
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.361648027000445e-05,
+      "loss": 0.3348,
+      "step": 23777
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3610938509953065e-05,
+      "loss": 0.3984,
+      "step": 23778
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.360539779549731e-05,
+      "loss": 0.4681,
+      "step": 23779
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3599858126704223e-05,
+      "loss": 0.4327,
+      "step": 23780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3594319503640884e-05,
+      "loss": 0.5296,
+      "step": 23781
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3588781926374294e-05,
+      "loss": 0.4235,
+      "step": 23782
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3583245394971467e-05,
+      "loss": 0.4309,
+      "step": 23783
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3577709909499469e-05,
+      "loss": 0.4447,
+      "step": 23784
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.357217547002526e-05,
+      "loss": 0.3914,
+      "step": 23785
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3566642076615832e-05,
+      "loss": 0.5382,
+      "step": 23786
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.356110972933815e-05,
+      "loss": 0.4847,
+      "step": 23787
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3555578428259163e-05,
+      "loss": 0.3661,
+      "step": 23788
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3550048173445861e-05,
+      "loss": 0.5457,
+      "step": 23789
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3544518964965148e-05,
+      "loss": 0.4141,
+      "step": 23790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3538990802883955e-05,
+      "loss": 0.3834,
+      "step": 23791
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3533463687269176e-05,
+      "loss": 0.3859,
+      "step": 23792
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3527937618187714e-05,
+      "loss": 0.4291,
+      "step": 23793
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3522412595706468e-05,
+      "loss": 0.3736,
+      "step": 23794
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3516888619892321e-05,
+      "loss": 0.3563,
+      "step": 23795
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3511365690812084e-05,
+      "loss": 0.4679,
+      "step": 23796
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3505843808532604e-05,
+      "loss": 0.4526,
+      "step": 23797
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3500322973120771e-05,
+      "loss": 0.3916,
+      "step": 23798
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3494803184643357e-05,
+      "loss": 0.4457,
+      "step": 23799
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3489284443167193e-05,
+      "loss": 0.4021,
+      "step": 23800
+    },
+    {
+      "epoch": 0.83,
+      "eval_loss": 0.4073745310306549,
+      "eval_runtime": 855.4661,
+      "eval_samples_per_second": 28.125,
+      "eval_steps_per_second": 14.063,
+      "step": 23800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3483766748759075e-05,
+      "loss": 0.4471,
+      "step": 23801
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3478250101485746e-05,
+      "loss": 0.4388,
+      "step": 23802
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.347273450141403e-05,
+      "loss": 0.3849,
+      "step": 23803
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3467219948610654e-05,
+      "loss": 0.4806,
+      "step": 23804
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3461706443142386e-05,
+      "loss": 0.3887,
+      "step": 23805
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3456193985075927e-05,
+      "loss": 0.4331,
+      "step": 23806
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3450682574477991e-05,
+      "loss": 0.3766,
+      "step": 23807
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3445172211415324e-05,
+      "loss": 0.4088,
+      "step": 23808
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3439662895954608e-05,
+      "loss": 0.4469,
+      "step": 23809
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.343415462816251e-05,
+      "loss": 0.448,
+      "step": 23810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.34286474081057e-05,
+      "loss": 0.4422,
+      "step": 23811
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3423141235850845e-05,
+      "loss": 0.3428,
+      "step": 23812
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3417636111464549e-05,
+      "loss": 0.4365,
+      "step": 23813
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3412132035013502e-05,
+      "loss": 0.4878,
+      "step": 23814
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3406629006564298e-05,
+      "loss": 0.4558,
+      "step": 23815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3401127026183535e-05,
+      "loss": 0.4318,
+      "step": 23816
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3395626093937808e-05,
+      "loss": 0.4063,
+      "step": 23817
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3390126209893672e-05,
+      "loss": 0.3717,
+      "step": 23818
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3384627374117742e-05,
+      "loss": 0.418,
+      "step": 23819
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3379129586676542e-05,
+      "loss": 0.3755,
+      "step": 23820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.337363284763663e-05,
+      "loss": 0.3478,
+      "step": 23821
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.336813715706452e-05,
+      "loss": 0.4568,
+      "step": 23822
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3362642515026713e-05,
+      "loss": 0.4683,
+      "step": 23823
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3357148921589757e-05,
+      "loss": 0.4673,
+      "step": 23824
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.335165637682012e-05,
+      "loss": 0.3687,
+      "step": 23825
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3346164880784274e-05,
+      "loss": 0.3836,
+      "step": 23826
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3340674433548694e-05,
+      "loss": 0.4399,
+      "step": 23827
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3335185035179799e-05,
+      "loss": 0.3821,
+      "step": 23828
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3329696685744086e-05,
+      "loss": 0.431,
+      "step": 23829
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.332420938530795e-05,
+      "loss": 0.3936,
+      "step": 23830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3318723133937816e-05,
+      "loss": 0.4171,
+      "step": 23831
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3313237931700073e-05,
+      "loss": 0.4135,
+      "step": 23832
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3307753778661091e-05,
+      "loss": 0.5208,
+      "step": 23833
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3302270674887307e-05,
+      "loss": 0.3871,
+      "step": 23834
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3296788620445044e-05,
+      "loss": 0.3887,
+      "step": 23835
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3291307615400661e-05,
+      "loss": 0.3773,
+      "step": 23836
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3285827659820493e-05,
+      "loss": 0.4299,
+      "step": 23837
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3280348753770844e-05,
+      "loss": 0.3601,
+      "step": 23838
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.327487089731807e-05,
+      "loss": 0.398,
+      "step": 23839
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3269394090528465e-05,
+      "loss": 0.4525,
+      "step": 23840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3263918333468295e-05,
+      "loss": 0.4425,
+      "step": 23841
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3258443626203842e-05,
+      "loss": 0.4675,
+      "step": 23842
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3252969968801354e-05,
+      "loss": 0.459,
+      "step": 23843
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.324749736132711e-05,
+      "loss": 0.397,
+      "step": 23844
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3242025803847336e-05,
+      "loss": 0.4148,
+      "step": 23845
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3236555296428255e-05,
+      "loss": 0.4261,
+      "step": 23846
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3231085839136081e-05,
+      "loss": 0.4191,
+      "step": 23847
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3225617432036996e-05,
+      "loss": 0.3866,
+      "step": 23848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.322015007519719e-05,
+      "loss": 0.392,
+      "step": 23849
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3214683768682856e-05,
+      "loss": 0.3929,
+      "step": 23850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3209218512560151e-05,
+      "loss": 0.3482,
+      "step": 23851
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.320375430689521e-05,
+      "loss": 0.3728,
+      "step": 23852
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3198291151754171e-05,
+      "loss": 0.4491,
+      "step": 23853
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3192829047203136e-05,
+      "loss": 0.3986,
+      "step": 23854
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3187367993308264e-05,
+      "loss": 0.4056,
+      "step": 23855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3181907990135622e-05,
+      "loss": 0.4451,
+      "step": 23856
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3176449037751293e-05,
+      "loss": 0.477,
+      "step": 23857
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3170991136221355e-05,
+      "loss": 0.5128,
+      "step": 23858
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3165534285611847e-05,
+      "loss": 0.4593,
+      "step": 23859
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3160078485988858e-05,
+      "loss": 0.3717,
+      "step": 23860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3154623737418382e-05,
+      "loss": 0.4012,
+      "step": 23861
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3149170039966463e-05,
+      "loss": 0.4571,
+      "step": 23862
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3143717393699095e-05,
+      "loss": 0.4587,
+      "step": 23863
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3138265798682259e-05,
+      "loss": 0.4296,
+      "step": 23864
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3132815254981978e-05,
+      "loss": 0.4341,
+      "step": 23865
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3127365762664202e-05,
+      "loss": 0.4624,
+      "step": 23866
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3121917321794885e-05,
+      "loss": 0.4641,
+      "step": 23867
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3116469932439968e-05,
+      "loss": 0.4279,
+      "step": 23868
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.311102359466536e-05,
+      "loss": 0.3481,
+      "step": 23869
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3105578308537037e-05,
+      "loss": 0.495,
+      "step": 23870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3100134074120873e-05,
+      "loss": 0.3899,
+      "step": 23871
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3094690891482764e-05,
+      "loss": 0.4294,
+      "step": 23872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3089248760688588e-05,
+      "loss": 0.3891,
+      "step": 23873
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3083807681804194e-05,
+      "loss": 0.45,
+      "step": 23874
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3078367654895485e-05,
+      "loss": 0.4593,
+      "step": 23875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3072928680028262e-05,
+      "loss": 0.4926,
+      "step": 23876
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3067490757268386e-05,
+      "loss": 0.4901,
+      "step": 23877
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3062053886681646e-05,
+      "loss": 0.5398,
+      "step": 23878
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3056618068333858e-05,
+      "loss": 0.4994,
+      "step": 23879
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3051183302290792e-05,
+      "loss": 0.4736,
+      "step": 23880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3045749588618272e-05,
+      "loss": 0.3613,
+      "step": 23881
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3040316927382035e-05,
+      "loss": 0.5006,
+      "step": 23882
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3034885318647839e-05,
+      "loss": 0.4302,
+      "step": 23883
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.302945476248143e-05,
+      "loss": 0.4179,
+      "step": 23884
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3024025258948502e-05,
+      "loss": 0.4328,
+      "step": 23885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3018596808114824e-05,
+      "loss": 0.3969,
+      "step": 23886
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3013169410046078e-05,
+      "loss": 0.5142,
+      "step": 23887
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3007743064807943e-05,
+      "loss": 0.389,
+      "step": 23888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.30023177724661e-05,
+      "loss": 0.4535,
+      "step": 23889
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2996893533086196e-05,
+      "loss": 0.5456,
+      "step": 23890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2991470346733913e-05,
+      "loss": 0.4584,
+      "step": 23891
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2986048213474888e-05,
+      "loss": 0.4009,
+      "step": 23892
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2980627133374734e-05,
+      "loss": 0.4103,
+      "step": 23893
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2975207106499065e-05,
+      "loss": 0.4093,
+      "step": 23894
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2969788132913464e-05,
+      "loss": 0.4529,
+      "step": 23895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2964370212683551e-05,
+      "loss": 0.5802,
+      "step": 23896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2958953345874902e-05,
+      "loss": 0.4124,
+      "step": 23897
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.295353753255305e-05,
+      "loss": 0.3993,
+      "step": 23898
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2948122772783566e-05,
+      "loss": 0.4147,
+      "step": 23899
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2942709066631952e-05,
+      "loss": 0.456,
+      "step": 23900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2937296414163791e-05,
+      "loss": 0.3118,
+      "step": 23901
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2931884815444551e-05,
+      "loss": 0.3774,
+      "step": 23902
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2926474270539756e-05,
+      "loss": 0.4248,
+      "step": 23903
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2921064779514869e-05,
+      "loss": 0.4651,
+      "step": 23904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2915656342435345e-05,
+      "loss": 0.4789,
+      "step": 23905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2910248959366699e-05,
+      "loss": 0.3881,
+      "step": 23906
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2904842630374347e-05,
+      "loss": 0.3637,
+      "step": 23907
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2899437355523724e-05,
+      "loss": 0.4329,
+      "step": 23908
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2894033134880269e-05,
+      "loss": 0.4712,
+      "step": 23909
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2888629968509335e-05,
+      "loss": 0.3863,
+      "step": 23910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2883227856476388e-05,
+      "loss": 0.3622,
+      "step": 23911
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2877826798846793e-05,
+      "loss": 0.4287,
+      "step": 23912
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.28724267956859e-05,
+      "loss": 0.4104,
+      "step": 23913
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2867027847059088e-05,
+      "loss": 0.4907,
+      "step": 23914
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2861629953031684e-05,
+      "loss": 0.4499,
+      "step": 23915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2856233113669004e-05,
+      "loss": 0.4433,
+      "step": 23916
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2850837329036425e-05,
+      "loss": 0.43,
+      "step": 23917
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.284544259919922e-05,
+      "loss": 0.4058,
+      "step": 23918
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.284004892422268e-05,
+      "loss": 0.3421,
+      "step": 23919
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2834656304172088e-05,
+      "loss": 0.4068,
+      "step": 23920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.28292647391127e-05,
+      "loss": 0.4865,
+      "step": 23921
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2823874229109812e-05,
+      "loss": 0.3608,
+      "step": 23922
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2818484774228635e-05,
+      "loss": 0.4121,
+      "step": 23923
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2813096374534417e-05,
+      "loss": 0.3924,
+      "step": 23924
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2807709030092362e-05,
+      "loss": 0.3524,
+      "step": 23925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2802322740967654e-05,
+      "loss": 0.3676,
+      "step": 23926
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2796937507225537e-05,
+      "loss": 0.3883,
+      "step": 23927
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.279155332893116e-05,
+      "loss": 0.3536,
+      "step": 23928
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2786170206149718e-05,
+      "loss": 0.4502,
+      "step": 23929
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2780788138946287e-05,
+      "loss": 0.4402,
+      "step": 23930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2775407127386085e-05,
+      "loss": 0.4153,
+      "step": 23931
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2770027171534226e-05,
+      "loss": 0.4788,
+      "step": 23932
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2764648271455804e-05,
+      "loss": 0.3229,
+      "step": 23933
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2759270427215942e-05,
+      "loss": 0.4349,
+      "step": 23934
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2753893638879689e-05,
+      "loss": 0.4281,
+      "step": 23935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.274851790651218e-05,
+      "loss": 0.4191,
+      "step": 23936
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2743143230178457e-05,
+      "loss": 0.3298,
+      "step": 23937
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2737769609943562e-05,
+      "loss": 0.4262,
+      "step": 23938
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2732397045872535e-05,
+      "loss": 0.5367,
+      "step": 23939
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.272702553803039e-05,
+      "loss": 0.4034,
+      "step": 23940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2721655086482176e-05,
+      "loss": 0.453,
+      "step": 23941
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2716285691292873e-05,
+      "loss": 0.4072,
+      "step": 23942
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2710917352527463e-05,
+      "loss": 0.372,
+      "step": 23943
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2705550070250938e-05,
+      "loss": 0.4337,
+      "step": 23944
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2700183844528224e-05,
+      "loss": 0.3543,
+      "step": 23945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2694818675424314e-05,
+      "loss": 0.5176,
+      "step": 23946
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2689454563004133e-05,
+      "loss": 0.4657,
+      "step": 23947
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2684091507332586e-05,
+      "loss": 0.3478,
+      "step": 23948
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2678729508474595e-05,
+      "loss": 0.3995,
+      "step": 23949
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.267336856649507e-05,
+      "loss": 0.4833,
+      "step": 23950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2668008681458853e-05,
+      "loss": 0.3815,
+      "step": 23951
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2662649853430874e-05,
+      "loss": 0.4407,
+      "step": 23952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2657292082475959e-05,
+      "loss": 0.4232,
+      "step": 23953
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2651935368658973e-05,
+      "loss": 0.474,
+      "step": 23954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2646579712044736e-05,
+      "loss": 0.3486,
+      "step": 23955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2641225112698052e-05,
+      "loss": 0.4163,
+      "step": 23956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2635871570683765e-05,
+      "loss": 0.4427,
+      "step": 23957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2630519086066661e-05,
+      "loss": 0.4009,
+      "step": 23958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.262516765891152e-05,
+      "loss": 0.425,
+      "step": 23959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2619817289283097e-05,
+      "loss": 0.3765,
+      "step": 23960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2614467977246159e-05,
+      "loss": 0.4224,
+      "step": 23961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2609119722865459e-05,
+      "loss": 0.4695,
+      "step": 23962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2603772526205737e-05,
+      "loss": 0.4413,
+      "step": 23963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2598426387331696e-05,
+      "loss": 0.4161,
+      "step": 23964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2593081306308042e-05,
+      "loss": 0.3983,
+      "step": 23965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2587737283199441e-05,
+      "loss": 0.4206,
+      "step": 23966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2582394318070633e-05,
+      "loss": 0.398,
+      "step": 23967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2577052410986257e-05,
+      "loss": 0.4461,
+      "step": 23968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2571711562010969e-05,
+      "loss": 0.5116,
+      "step": 23969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2566371771209407e-05,
+      "loss": 0.4416,
+      "step": 23970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2561033038646175e-05,
+      "loss": 0.4134,
+      "step": 23971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2555695364385944e-05,
+      "loss": 0.3498,
+      "step": 23972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2550358748493286e-05,
+      "loss": 0.3529,
+      "step": 23973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2545023191032801e-05,
+      "loss": 0.4881,
+      "step": 23974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2539688692069062e-05,
+      "loss": 0.3828,
+      "step": 23975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2534355251666608e-05,
+      "loss": 0.4626,
+      "step": 23976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.252902286989004e-05,
+      "loss": 0.4919,
+      "step": 23977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2523691546803873e-05,
+      "loss": 0.4019,
+      "step": 23978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2518361282472645e-05,
+      "loss": 0.4429,
+      "step": 23979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.251303207696085e-05,
+      "loss": 0.5075,
+      "step": 23980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2507703930333014e-05,
+      "loss": 0.5402,
+      "step": 23981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2502376842653585e-05,
+      "loss": 0.4993,
+      "step": 23982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2497050813987078e-05,
+      "loss": 0.3317,
+      "step": 23983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.249172584439795e-05,
+      "loss": 0.4935,
+      "step": 23984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2486401933950643e-05,
+      "loss": 0.4307,
+      "step": 23985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2481079082709602e-05,
+      "loss": 0.4181,
+      "step": 23986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.247575729073921e-05,
+      "loss": 0.4496,
+      "step": 23987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2470436558103938e-05,
+      "loss": 0.3565,
+      "step": 23988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2465116884868167e-05,
+      "loss": 0.4529,
+      "step": 23989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2459798271096268e-05,
+      "loss": 0.4197,
+      "step": 23990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2454480716852624e-05,
+      "loss": 0.4999,
+      "step": 23991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.244916422220156e-05,
+      "loss": 0.3789,
+      "step": 23992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.244384878720749e-05,
+      "loss": 0.3733,
+      "step": 23993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.243853441193471e-05,
+      "loss": 0.4502,
+      "step": 23994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2433221096447534e-05,
+      "loss": 0.4274,
+      "step": 23995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2427908840810298e-05,
+      "loss": 0.4087,
+      "step": 23996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2422597645087252e-05,
+      "loss": 0.388,
+      "step": 23997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2417287509342734e-05,
+      "loss": 0.3807,
+      "step": 23998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2411978433640991e-05,
+      "loss": 0.4827,
+      "step": 23999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2406670418046273e-05,
+      "loss": 0.376,
+      "step": 24000
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 0.4063355028629303,
+      "eval_runtime": 853.5803,
+      "eval_samples_per_second": 28.187,
+      "eval_steps_per_second": 14.094,
+      "step": 24000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.240136346262284e-05,
+      "loss": 0.3569,
+      "step": 24001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2396057567434882e-05,
+      "loss": 0.453,
+      "step": 24002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2390752732546684e-05,
+      "loss": 0.4724,
+      "step": 24003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2385448958022417e-05,
+      "loss": 0.3738,
+      "step": 24004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2380146243926272e-05,
+      "loss": 0.4059,
+      "step": 24005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2374844590322432e-05,
+      "loss": 0.3786,
+      "step": 24006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2369543997275045e-05,
+      "loss": 0.4133,
+      "step": 24007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2364244464848318e-05,
+      "loss": 0.4292,
+      "step": 24008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2358945993106352e-05,
+      "loss": 0.4167,
+      "step": 24009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2353648582113286e-05,
+      "loss": 0.4662,
+      "step": 24010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2348352231933246e-05,
+      "loss": 0.4155,
+      "step": 24011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2343056942630304e-05,
+      "loss": 0.4337,
+      "step": 24012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2337762714268586e-05,
+      "loss": 0.4426,
+      "step": 24013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2332469546912162e-05,
+      "loss": 0.4091,
+      "step": 24014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2327177440625092e-05,
+      "loss": 0.3436,
+      "step": 24015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2321886395471428e-05,
+      "loss": 0.4246,
+      "step": 24016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2316596411515213e-05,
+      "loss": 0.4038,
+      "step": 24017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2311307488820445e-05,
+      "loss": 0.5013,
+      "step": 24018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2306019627451181e-05,
+      "loss": 0.4176,
+      "step": 24019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2300732827471406e-05,
+      "loss": 0.345,
+      "step": 24020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.22954470889451e-05,
+      "loss": 0.416,
+      "step": 24021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2290162411936246e-05,
+      "loss": 0.4238,
+      "step": 24022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.228487879650878e-05,
+      "loss": 0.3528,
+      "step": 24023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2279596242726699e-05,
+      "loss": 0.3497,
+      "step": 24024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2274314750653915e-05,
+      "loss": 0.341,
+      "step": 24025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2269034320354356e-05,
+      "loss": 0.5035,
+      "step": 24026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2263754951891927e-05,
+      "loss": 0.4279,
+      "step": 24027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2258476645330496e-05,
+      "loss": 0.4096,
+      "step": 24028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2253199400734005e-05,
+      "loss": 0.4333,
+      "step": 24029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2247923218166302e-05,
+      "loss": 0.4423,
+      "step": 24030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2242648097691256e-05,
+      "loss": 0.4187,
+      "step": 24031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2237374039372685e-05,
+      "loss": 0.4506,
+      "step": 24032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2232101043274436e-05,
+      "loss": 0.4336,
+      "step": 24033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2226829109460335e-05,
+      "loss": 0.499,
+      "step": 24034
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2221558237994213e-05,
+      "loss": 0.4002,
+      "step": 24035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2216288428939826e-05,
+      "loss": 0.4078,
+      "step": 24036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.221101968236097e-05,
+      "loss": 0.3566,
+      "step": 24037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2205751998321402e-05,
+      "loss": 0.3986,
+      "step": 24038
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2200485376884918e-05,
+      "loss": 0.4974,
+      "step": 24039
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2195219818115222e-05,
+      "loss": 0.4023,
+      "step": 24040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2189955322076074e-05,
+      "loss": 0.5045,
+      "step": 24041
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2184691888831168e-05,
+      "loss": 0.3935,
+      "step": 24042
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2179429518444196e-05,
+      "loss": 0.4633,
+      "step": 24043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2174168210978897e-05,
+      "loss": 0.3812,
+      "step": 24044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2168907966498922e-05,
+      "loss": 0.4205,
+      "step": 24045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.216364878506795e-05,
+      "loss": 0.3933,
+      "step": 24046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.215839066674962e-05,
+      "loss": 0.3683,
+      "step": 24047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.215313361160758e-05,
+      "loss": 0.3965,
+      "step": 24048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2147877619705428e-05,
+      "loss": 0.3813,
+      "step": 24049
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2142622691106841e-05,
+      "loss": 0.4455,
+      "step": 24050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2137368825875373e-05,
+      "loss": 0.3819,
+      "step": 24051
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2132116024074635e-05,
+      "loss": 0.3664,
+      "step": 24052
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2126864285768202e-05,
+      "loss": 0.3857,
+      "step": 24053
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2121613611019612e-05,
+      "loss": 0.462,
+      "step": 24054
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2116363999892443e-05,
+      "loss": 0.4291,
+      "step": 24055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.211111545245024e-05,
+      "loss": 0.4127,
+      "step": 24056
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2105867968756501e-05,
+      "loss": 0.3435,
+      "step": 24057
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2100621548874758e-05,
+      "loss": 0.3759,
+      "step": 24058
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.209537619286848e-05,
+      "loss": 0.4493,
+      "step": 24059
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2090131900801194e-05,
+      "loss": 0.53,
+      "step": 24060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2084888672736349e-05,
+      "loss": 0.3676,
+      "step": 24061
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2079646508737418e-05,
+      "loss": 0.3664,
+      "step": 24062
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.207440540886785e-05,
+      "loss": 0.458,
+      "step": 24063
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.206916537319106e-05,
+      "loss": 0.402,
+      "step": 24064
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2063926401770498e-05,
+      "loss": 0.3619,
+      "step": 24065
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2058688494669546e-05,
+      "loss": 0.4209,
+      "step": 24066
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2053451651951608e-05,
+      "loss": 0.422,
+      "step": 24067
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2048215873680057e-05,
+      "loss": 0.4107,
+      "step": 24068
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2042981159918287e-05,
+      "loss": 0.4611,
+      "step": 24069
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2037747510729658e-05,
+      "loss": 0.4944,
+      "step": 24070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2032514926177495e-05,
+      "loss": 0.3791,
+      "step": 24071
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.202728340632514e-05,
+      "loss": 0.4391,
+      "step": 24072
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2022052951235884e-05,
+      "loss": 0.5107,
+      "step": 24073
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2016823560973078e-05,
+      "loss": 0.4345,
+      "step": 24074
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2011595235599993e-05,
+      "loss": 0.3752,
+      "step": 24075
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2006367975179922e-05,
+      "loss": 0.3789,
+      "step": 24076
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2001141779776103e-05,
+      "loss": 0.3708,
+      "step": 24077
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.19959166494518e-05,
+      "loss": 0.4146,
+      "step": 24078
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.199069258427028e-05,
+      "loss": 0.3755,
+      "step": 24079
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1985469584294761e-05,
+      "loss": 0.447,
+      "step": 24080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1980247649588438e-05,
+      "loss": 0.5009,
+      "step": 24081
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1975026780214537e-05,
+      "loss": 0.4304,
+      "step": 24082
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1969806976236243e-05,
+      "loss": 0.4057,
+      "step": 24083
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1964588237716701e-05,
+      "loss": 0.3693,
+      "step": 24084
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.195937056471913e-05,
+      "loss": 0.5407,
+      "step": 24085
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1954153957306647e-05,
+      "loss": 0.431,
+      "step": 24086
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.19489384155424e-05,
+      "loss": 0.4076,
+      "step": 24087
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1943723939489516e-05,
+      "loss": 0.4699,
+      "step": 24088
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1938510529211078e-05,
+      "loss": 0.429,
+      "step": 24089
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1933298184770225e-05,
+      "loss": 0.4078,
+      "step": 24090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.192808690623004e-05,
+      "loss": 0.3783,
+      "step": 24091
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1922876693653585e-05,
+      "loss": 0.4758,
+      "step": 24092
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1917667547103928e-05,
+      "loss": 0.4865,
+      "step": 24093
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1912459466644077e-05,
+      "loss": 0.4705,
+      "step": 24094
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1907252452337136e-05,
+      "loss": 0.4295,
+      "step": 24095
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.190204650424609e-05,
+      "loss": 0.3859,
+      "step": 24096
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1896841622433952e-05,
+      "loss": 0.4385,
+      "step": 24097
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1891637806963707e-05,
+      "loss": 0.436,
+      "step": 24098
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1886435057898337e-05,
+      "loss": 0.3626,
+      "step": 24099
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1881233375300849e-05,
+      "loss": 0.3982,
+      "step": 24100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1876032759234168e-05,
+      "loss": 0.4588,
+      "step": 24101
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1870833209761257e-05,
+      "loss": 0.4363,
+      "step": 24102
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1865634726945029e-05,
+      "loss": 0.4438,
+      "step": 24103
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1860437310848405e-05,
+      "loss": 0.4369,
+      "step": 24104
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1855240961534308e-05,
+      "loss": 0.4144,
+      "step": 24105
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1850045679065624e-05,
+      "loss": 0.364,
+      "step": 24106
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1844851463505236e-05,
+      "loss": 0.4201,
+      "step": 24107
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.183965831491599e-05,
+      "loss": 0.4343,
+      "step": 24108
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1834466233360752e-05,
+      "loss": 0.4714,
+      "step": 24109
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.182927521890238e-05,
+      "loss": 0.3399,
+      "step": 24110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1824085271603703e-05,
+      "loss": 0.3922,
+      "step": 24111
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1818896391527512e-05,
+      "loss": 0.4071,
+      "step": 24112
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1813708578736627e-05,
+      "loss": 0.3725,
+      "step": 24113
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1808521833293807e-05,
+      "loss": 0.3807,
+      "step": 24114
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1803336155261879e-05,
+      "loss": 0.4531,
+      "step": 24115
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1798151544703584e-05,
+      "loss": 0.4492,
+      "step": 24116
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.179296800168167e-05,
+      "loss": 0.3993,
+      "step": 24117
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1787785526258876e-05,
+      "loss": 0.3479,
+      "step": 24118
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1782604118497931e-05,
+      "loss": 0.4267,
+      "step": 24119
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1777423778461517e-05,
+      "loss": 0.4801,
+      "step": 24120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1772244506212383e-05,
+      "loss": 0.3824,
+      "step": 24121
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.17670663018132e-05,
+      "loss": 0.3519,
+      "step": 24122
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1761889165326633e-05,
+      "loss": 0.4317,
+      "step": 24123
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1756713096815342e-05,
+      "loss": 0.3844,
+      "step": 24124
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1751538096341952e-05,
+      "loss": 0.496,
+      "step": 24125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1746364163969148e-05,
+      "loss": 0.4061,
+      "step": 24126
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1741191299759535e-05,
+      "loss": 0.434,
+      "step": 24127
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1736019503775698e-05,
+      "loss": 0.4337,
+      "step": 24128
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1730848776080261e-05,
+      "loss": 0.3947,
+      "step": 24129
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1725679116735766e-05,
+      "loss": 0.3922,
+      "step": 24130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.172051052580484e-05,
+      "loss": 0.4111,
+      "step": 24131
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.171534300335001e-05,
+      "loss": 0.3648,
+      "step": 24132
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1710176549433816e-05,
+      "loss": 0.403,
+      "step": 24133
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1705011164118796e-05,
+      "loss": 0.4269,
+      "step": 24134
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1699846847467455e-05,
+      "loss": 0.3814,
+      "step": 24135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1694683599542333e-05,
+      "loss": 0.3686,
+      "step": 24136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.168952142040589e-05,
+      "loss": 0.3688,
+      "step": 24137
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1684360310120624e-05,
+      "loss": 0.4994,
+      "step": 24138
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.167920026874899e-05,
+      "loss": 0.3828,
+      "step": 24139
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1674041296353422e-05,
+      "loss": 0.5079,
+      "step": 24140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.166888339299641e-05,
+      "loss": 0.4148,
+      "step": 24141
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1663726558740362e-05,
+      "loss": 0.4405,
+      "step": 24142
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1658570793647683e-05,
+      "loss": 0.3534,
+      "step": 24143
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1653416097780778e-05,
+      "loss": 0.3656,
+      "step": 24144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1648262471202009e-05,
+      "loss": 0.3975,
+      "step": 24145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1643109913973815e-05,
+      "loss": 0.3794,
+      "step": 24146
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1637958426158513e-05,
+      "loss": 0.3959,
+      "step": 24147
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1632808007818475e-05,
+      "loss": 0.4773,
+      "step": 24148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1627658659016016e-05,
+      "loss": 0.4361,
+      "step": 24149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1622510379813478e-05,
+      "loss": 0.4053,
+      "step": 24150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1617363170273144e-05,
+      "loss": 0.3691,
+      "step": 24151
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.161221703045735e-05,
+      "loss": 0.4083,
+      "step": 24152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1607071960428384e-05,
+      "loss": 0.4162,
+      "step": 24153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1601927960248483e-05,
+      "loss": 0.3971,
+      "step": 24154
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.159678502997993e-05,
+      "loss": 0.401,
+      "step": 24155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1591643169684952e-05,
+      "loss": 0.4687,
+      "step": 24156
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.15865023794258e-05,
+      "loss": 0.4083,
+      "step": 24157
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1581362659264705e-05,
+      "loss": 0.5051,
+      "step": 24158
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1576224009263859e-05,
+      "loss": 0.4869,
+      "step": 24159
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1571086429485456e-05,
+      "loss": 0.4582,
+      "step": 24160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.156594991999167e-05,
+      "loss": 0.4202,
+      "step": 24161
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1560814480844684e-05,
+      "loss": 0.3881,
+      "step": 24162
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.155568011210666e-05,
+      "loss": 0.4708,
+      "step": 24163
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1550546813839735e-05,
+      "loss": 0.4843,
+      "step": 24164
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.154541458610603e-05,
+      "loss": 0.4113,
+      "step": 24165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1540283428967646e-05,
+      "loss": 0.4632,
+      "step": 24166
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1535153342486726e-05,
+      "loss": 0.5063,
+      "step": 24167
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1530024326725353e-05,
+      "loss": 0.4441,
+      "step": 24168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1524896381745587e-05,
+      "loss": 0.359,
+      "step": 24169
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1519769507609513e-05,
+      "loss": 0.4508,
+      "step": 24170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1514643704379147e-05,
+      "loss": 0.3721,
+      "step": 24171
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1509518972116572e-05,
+      "loss": 0.4499,
+      "step": 24172
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1504395310883798e-05,
+      "loss": 0.4266,
+      "step": 24173
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1499272720742827e-05,
+      "loss": 0.3531,
+      "step": 24174
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1494151201755676e-05,
+      "loss": 0.4513,
+      "step": 24175
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.148903075398431e-05,
+      "loss": 0.3448,
+      "step": 24176
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.148391137749073e-05,
+      "loss": 0.444,
+      "step": 24177
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1478793072336892e-05,
+      "loss": 0.3792,
+      "step": 24178
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.147367583858474e-05,
+      "loss": 0.4098,
+      "step": 24179
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.146855967629621e-05,
+      "loss": 0.3509,
+      "step": 24180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1463444585533212e-05,
+      "loss": 0.45,
+      "step": 24181
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1458330566357656e-05,
+      "loss": 0.3631,
+      "step": 24182
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1453217618831469e-05,
+      "loss": 0.412,
+      "step": 24183
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1448105743016514e-05,
+      "loss": 0.4107,
+      "step": 24184
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1442994938974661e-05,
+      "loss": 0.434,
+      "step": 24185
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1437885206767773e-05,
+      "loss": 0.447,
+      "step": 24186
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1432776546457669e-05,
+      "loss": 0.4886,
+      "step": 24187
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1427668958106219e-05,
+      "loss": 0.3653,
+      "step": 24188
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.142256244177523e-05,
+      "loss": 0.3857,
+      "step": 24189
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.141745699752651e-05,
+      "loss": 0.3499,
+      "step": 24190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.141235262542184e-05,
+      "loss": 0.4349,
+      "step": 24191
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1407249325522983e-05,
+      "loss": 0.3095,
+      "step": 24192
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1402147097891758e-05,
+      "loss": 0.4011,
+      "step": 24193
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1397045942589879e-05,
+      "loss": 0.4921,
+      "step": 24194
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1391945859679098e-05,
+      "loss": 0.4366,
+      "step": 24195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1386846849221145e-05,
+      "loss": 0.4027,
+      "step": 24196
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1381748911277712e-05,
+      "loss": 0.4498,
+      "step": 24197
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1376652045910552e-05,
+      "loss": 0.3971,
+      "step": 24198
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1371556253181337e-05,
+      "loss": 0.419,
+      "step": 24199
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1366461533151706e-05,
+      "loss": 0.5067,
+      "step": 24200
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.4052059054374695,
+      "eval_runtime": 844.0209,
+      "eval_samples_per_second": 28.506,
+      "eval_steps_per_second": 14.253,
+      "step": 24200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1361367885883322e-05,
+      "loss": 0.3842,
+      "step": 24201
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1356275311437892e-05,
+      "loss": 0.4163,
+      "step": 24202
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.135118380987701e-05,
+      "loss": 0.3728,
+      "step": 24203
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1346093381262314e-05,
+      "loss": 0.4277,
+      "step": 24204
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1341004025655411e-05,
+      "loss": 0.4486,
+      "step": 24205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1335915743117875e-05,
+      "loss": 0.4039,
+      "step": 24206
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1330828533711347e-05,
+      "loss": 0.4059,
+      "step": 24207
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1325742397497364e-05,
+      "loss": 0.4502,
+      "step": 24208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1320657334537488e-05,
+      "loss": 0.4093,
+      "step": 24209
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.131557334489326e-05,
+      "loss": 0.3741,
+      "step": 24210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1310490428626209e-05,
+      "loss": 0.3989,
+      "step": 24211
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1305408585797871e-05,
+      "loss": 0.4225,
+      "step": 24212
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1300327816469769e-05,
+      "loss": 0.4321,
+      "step": 24213
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1295248120703362e-05,
+      "loss": 0.4506,
+      "step": 24214
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1290169498560143e-05,
+      "loss": 0.3811,
+      "step": 24215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1285091950101578e-05,
+      "loss": 0.4312,
+      "step": 24216
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1280015475389139e-05,
+      "loss": 0.3981,
+      "step": 24217
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1274940074484253e-05,
+      "loss": 0.3919,
+      "step": 24218
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1269865747448349e-05,
+      "loss": 0.4286,
+      "step": 24219
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1264792494342857e-05,
+      "loss": 0.4326,
+      "step": 24220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1259720315229172e-05,
+      "loss": 0.4259,
+      "step": 24221
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1254649210168655e-05,
+      "loss": 0.4646,
+      "step": 24222
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1249579179222736e-05,
+      "loss": 0.3506,
+      "step": 24223
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1244510222452742e-05,
+      "loss": 0.4182,
+      "step": 24224
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1239442339920047e-05,
+      "loss": 0.444,
+      "step": 24225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1234375531685981e-05,
+      "loss": 0.4399,
+      "step": 24226
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1229309797811838e-05,
+      "loss": 0.3993,
+      "step": 24227
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.122424513835898e-05,
+      "loss": 0.3556,
+      "step": 24228
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1219181553388692e-05,
+      "loss": 0.4147,
+      "step": 24229
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1214119042962257e-05,
+      "loss": 0.4517,
+      "step": 24230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1209057607140938e-05,
+      "loss": 0.5012,
+      "step": 24231
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1203997245985976e-05,
+      "loss": 0.4132,
+      "step": 24232
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1198937959558676e-05,
+      "loss": 0.4576,
+      "step": 24233
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1193879747920234e-05,
+      "loss": 0.4804,
+      "step": 24234
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1188822611131877e-05,
+      "loss": 0.4187,
+      "step": 24235
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1183766549254815e-05,
+      "loss": 0.4994,
+      "step": 24236
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.117871156235023e-05,
+      "loss": 0.4241,
+      "step": 24237
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.117365765047933e-05,
+      "loss": 0.3669,
+      "step": 24238
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1168604813703266e-05,
+      "loss": 0.4151,
+      "step": 24239
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1163553052083208e-05,
+      "loss": 0.4133,
+      "step": 24240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1158502365680278e-05,
+      "loss": 0.4319,
+      "step": 24241
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1153452754555604e-05,
+      "loss": 0.4213,
+      "step": 24242
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1148404218770347e-05,
+      "loss": 0.4732,
+      "step": 24243
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1143356758385571e-05,
+      "loss": 0.4189,
+      "step": 24244
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.113831037346238e-05,
+      "loss": 0.4411,
+      "step": 24245
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1133265064061849e-05,
+      "loss": 0.4035,
+      "step": 24246
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1128220830245018e-05,
+      "loss": 0.4076,
+      "step": 24247
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1123177672072992e-05,
+      "loss": 0.4314,
+      "step": 24248
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1118135589606782e-05,
+      "loss": 0.4865,
+      "step": 24249
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1113094582907412e-05,
+      "loss": 0.4581,
+      "step": 24250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1108054652035894e-05,
+      "loss": 0.3896,
+      "step": 24251
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.110301579705324e-05,
+      "loss": 0.4558,
+      "step": 24252
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1097978018020405e-05,
+      "loss": 0.5011,
+      "step": 24253
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1092941314998406e-05,
+      "loss": 0.3993,
+      "step": 24254
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1087905688048183e-05,
+      "loss": 0.4647,
+      "step": 24255
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1082871137230689e-05,
+      "loss": 0.4493,
+      "step": 24256
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.107783766260685e-05,
+      "loss": 0.4216,
+      "step": 24257
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1072805264237573e-05,
+      "loss": 0.3977,
+      "step": 24258
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1067773942183813e-05,
+      "loss": 0.4114,
+      "step": 24259
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1062743696506428e-05,
+      "loss": 0.4163,
+      "step": 24260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1057714527266328e-05,
+      "loss": 0.4347,
+      "step": 24261
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1052686434524351e-05,
+      "loss": 0.3719,
+      "step": 24262
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.104765941834136e-05,
+      "loss": 0.4174,
+      "step": 24263
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.104263347877823e-05,
+      "loss": 0.4778,
+      "step": 24264
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1037608615895767e-05,
+      "loss": 0.4352,
+      "step": 24265
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1032584829754788e-05,
+      "loss": 0.375,
+      "step": 24266
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1027562120416101e-05,
+      "loss": 0.4626,
+      "step": 24267
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1022540487940481e-05,
+      "loss": 0.3706,
+      "step": 24268
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1017519932388743e-05,
+      "loss": 0.3377,
+      "step": 24269
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1012500453821629e-05,
+      "loss": 0.4329,
+      "step": 24270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1007482052299888e-05,
+      "loss": 0.4365,
+      "step": 24271
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1002464727884277e-05,
+      "loss": 0.4521,
+      "step": 24272
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0997448480635485e-05,
+      "loss": 0.437,
+      "step": 24273
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.099243331061428e-05,
+      "loss": 0.4481,
+      "step": 24274
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0987419217881334e-05,
+      "loss": 0.4128,
+      "step": 24275
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.098240620249733e-05,
+      "loss": 0.4197,
+      "step": 24276
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0977394264522944e-05,
+      "loss": 0.384,
+      "step": 24277
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0972383404018826e-05,
+      "loss": 0.3826,
+      "step": 24278
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0967373621045652e-05,
+      "loss": 0.4005,
+      "step": 24279
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0962364915664048e-05,
+      "loss": 0.38,
+      "step": 24280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0957357287934632e-05,
+      "loss": 0.3727,
+      "step": 24281
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0952350737918004e-05,
+      "loss": 0.4195,
+      "step": 24282
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.094734526567478e-05,
+      "loss": 0.4275,
+      "step": 24283
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0942340871265499e-05,
+      "loss": 0.4635,
+      "step": 24284
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0937337554750792e-05,
+      "loss": 0.3758,
+      "step": 24285
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0932335316191178e-05,
+      "loss": 0.4946,
+      "step": 24286
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0927334155647218e-05,
+      "loss": 0.3827,
+      "step": 24287
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.092233407317943e-05,
+      "loss": 0.4886,
+      "step": 24288
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.091733506884831e-05,
+      "loss": 0.4905,
+      "step": 24289
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0912337142714413e-05,
+      "loss": 0.4185,
+      "step": 24290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0907340294838209e-05,
+      "loss": 0.3796,
+      "step": 24291
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0902344525280173e-05,
+      "loss": 0.4645,
+      "step": 24292
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0897349834100767e-05,
+      "loss": 0.4182,
+      "step": 24293
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0892356221360433e-05,
+      "loss": 0.3728,
+      "step": 24294
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0887363687119634e-05,
+      "loss": 0.3879,
+      "step": 24295
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0882372231438798e-05,
+      "loss": 0.4115,
+      "step": 24296
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0877381854378322e-05,
+      "loss": 0.3771,
+      "step": 24297
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0872392555998613e-05,
+      "loss": 0.4435,
+      "step": 24298
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0867404336360032e-05,
+      "loss": 0.4639,
+      "step": 24299
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0862417195523e-05,
+      "loss": 0.5001,
+      "step": 24300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0857431133547858e-05,
+      "loss": 0.4128,
+      "step": 24301
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0852446150494943e-05,
+      "loss": 0.4016,
+      "step": 24302
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.084746224642461e-05,
+      "loss": 0.4217,
+      "step": 24303
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0842479421397146e-05,
+      "loss": 0.4224,
+      "step": 24304
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0837497675472896e-05,
+      "loss": 0.393,
+      "step": 24305
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.083251700871215e-05,
+      "loss": 0.381,
+      "step": 24306
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0827537421175182e-05,
+      "loss": 0.3559,
+      "step": 24307
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0822558912922265e-05,
+      "loss": 0.3986,
+      "step": 24308
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.081758148401364e-05,
+      "loss": 0.4458,
+      "step": 24309
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0812605134509568e-05,
+      "loss": 0.4964,
+      "step": 24310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0807629864470282e-05,
+      "loss": 0.4689,
+      "step": 24311
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0802655673955997e-05,
+      "loss": 0.4208,
+      "step": 24312
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0797682563026912e-05,
+      "loss": 0.5098,
+      "step": 24313
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.07927105317432e-05,
+      "loss": 0.4594,
+      "step": 24314
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.078773958016508e-05,
+      "loss": 0.3939,
+      "step": 24315
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0782769708352691e-05,
+      "loss": 0.4111,
+      "step": 24316
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0777800916366187e-05,
+      "loss": 0.4212,
+      "step": 24317
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.077283320426572e-05,
+      "loss": 0.4341,
+      "step": 24318
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0767866572111395e-05,
+      "loss": 0.5138,
+      "step": 24319
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0762901019963322e-05,
+      "loss": 0.4601,
+      "step": 24320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0757936547881631e-05,
+      "loss": 0.4831,
+      "step": 24321
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0752973155926394e-05,
+      "loss": 0.4502,
+      "step": 24322
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0748010844157686e-05,
+      "loss": 0.4418,
+      "step": 24323
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0743049612635559e-05,
+      "loss": 0.4091,
+      "step": 24324
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0738089461420042e-05,
+      "loss": 0.4715,
+      "step": 24325
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0733130390571211e-05,
+      "loss": 0.4128,
+      "step": 24326
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0728172400149073e-05,
+      "loss": 0.4259,
+      "step": 24327
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0723215490213634e-05,
+      "loss": 0.4436,
+      "step": 24328
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0718259660824892e-05,
+      "loss": 0.4448,
+      "step": 24329
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.07133049120428e-05,
+      "loss": 0.4095,
+      "step": 24330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0708351243927362e-05,
+      "loss": 0.389,
+      "step": 24331
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0703398656538532e-05,
+      "loss": 0.3915,
+      "step": 24332
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0698447149936242e-05,
+      "loss": 0.4412,
+      "step": 24333
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.069349672418043e-05,
+      "loss": 0.3688,
+      "step": 24334
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0688547379330993e-05,
+      "loss": 0.3431,
+      "step": 24335
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0683599115447863e-05,
+      "loss": 0.4175,
+      "step": 24336
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0678651932590911e-05,
+      "loss": 0.4378,
+      "step": 24337
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0673705830820013e-05,
+      "loss": 0.4784,
+      "step": 24338
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0668760810195023e-05,
+      "loss": 0.4364,
+      "step": 24339
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0663816870775823e-05,
+      "loss": 0.4162,
+      "step": 24340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0658874012622244e-05,
+      "loss": 0.4158,
+      "step": 24341
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0653932235794107e-05,
+      "loss": 0.4315,
+      "step": 24342
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0648991540351206e-05,
+      "loss": 0.3897,
+      "step": 24343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0644051926353349e-05,
+      "loss": 0.3734,
+      "step": 24344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0639113393860334e-05,
+      "loss": 0.4401,
+      "step": 24345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0634175942931934e-05,
+      "loss": 0.4375,
+      "step": 24346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0629239573627903e-05,
+      "loss": 0.4018,
+      "step": 24347
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0624304286007991e-05,
+      "loss": 0.3159,
+      "step": 24348
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0619370080131897e-05,
+      "loss": 0.3522,
+      "step": 24349
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0614436956059392e-05,
+      "loss": 0.4826,
+      "step": 24350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0609504913850165e-05,
+      "loss": 0.3753,
+      "step": 24351
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0604573953563913e-05,
+      "loss": 0.3774,
+      "step": 24352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0599644075260307e-05,
+      "loss": 0.4522,
+      "step": 24353
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0594715278999012e-05,
+      "loss": 0.3701,
+      "step": 24354
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0589787564839682e-05,
+      "loss": 0.5032,
+      "step": 24355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0584860932841977e-05,
+      "loss": 0.4407,
+      "step": 24356
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.057993538306552e-05,
+      "loss": 0.3739,
+      "step": 24357
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0575010915569915e-05,
+      "loss": 0.407,
+      "step": 24358
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0570087530414785e-05,
+      "loss": 0.4326,
+      "step": 24359
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0565165227659679e-05,
+      "loss": 0.4037,
+      "step": 24360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0560244007364218e-05,
+      "loss": 0.3941,
+      "step": 24361
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0555323869587952e-05,
+      "loss": 0.4527,
+      "step": 24362
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0550404814390424e-05,
+      "loss": 0.4101,
+      "step": 24363
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0545486841831177e-05,
+      "loss": 0.534,
+      "step": 24364
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0540569951969703e-05,
+      "loss": 0.4144,
+      "step": 24365
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.053565414486557e-05,
+      "loss": 0.4364,
+      "step": 24366
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0530739420578239e-05,
+      "loss": 0.3849,
+      "step": 24367
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0525825779167209e-05,
+      "loss": 0.5377,
+      "step": 24368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.052091322069194e-05,
+      "loss": 0.3803,
+      "step": 24369
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0516001745211879e-05,
+      "loss": 0.3801,
+      "step": 24370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0511091352786506e-05,
+      "loss": 0.4431,
+      "step": 24371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0506182043475232e-05,
+      "loss": 0.378,
+      "step": 24372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0501273817337476e-05,
+      "loss": 0.4051,
+      "step": 24373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0496366674432656e-05,
+      "loss": 0.4365,
+      "step": 24374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0491460614820125e-05,
+      "loss": 0.4654,
+      "step": 24375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0486555638559325e-05,
+      "loss": 0.4031,
+      "step": 24376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0481651745709576e-05,
+      "loss": 0.4562,
+      "step": 24377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0476748936330261e-05,
+      "loss": 0.5177,
+      "step": 24378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.047184721048069e-05,
+      "loss": 0.4066,
+      "step": 24379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0466946568220193e-05,
+      "loss": 0.4487,
+      "step": 24380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0462047009608111e-05,
+      "loss": 0.4086,
+      "step": 24381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0457148534703742e-05,
+      "loss": 0.3467,
+      "step": 24382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.045225114356635e-05,
+      "loss": 0.4056,
+      "step": 24383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0447354836255229e-05,
+      "loss": 0.3904,
+      "step": 24384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0442459612829637e-05,
+      "loss": 0.4436,
+      "step": 24385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0437565473348798e-05,
+      "loss": 0.3658,
+      "step": 24386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0432672417871981e-05,
+      "loss": 0.4428,
+      "step": 24387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0427780446458402e-05,
+      "loss": 0.3801,
+      "step": 24388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.042288955916726e-05,
+      "loss": 0.512,
+      "step": 24389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0417999756057761e-05,
+      "loss": 0.5091,
+      "step": 24390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.041311103718906e-05,
+      "loss": 0.4391,
+      "step": 24391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0408223402620365e-05,
+      "loss": 0.515,
+      "step": 24392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0403336852410816e-05,
+      "loss": 0.3916,
+      "step": 24393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0398451386619557e-05,
+      "loss": 0.4496,
+      "step": 24394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0393567005305726e-05,
+      "loss": 0.4195,
+      "step": 24395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0388683708528413e-05,
+      "loss": 0.3839,
+      "step": 24396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0383801496346757e-05,
+      "loss": 0.3532,
+      "step": 24397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0378920368819834e-05,
+      "loss": 0.4303,
+      "step": 24398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0374040326006717e-05,
+      "loss": 0.4278,
+      "step": 24399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0369161367966485e-05,
+      "loss": 0.4721,
+      "step": 24400
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.4041050672531128,
+      "eval_runtime": 845.7214,
+      "eval_samples_per_second": 28.449,
+      "eval_steps_per_second": 14.225,
+      "step": 24400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0364283494758153e-05,
+      "loss": 0.4544,
+      "step": 24401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.035940670644081e-05,
+      "loss": 0.4083,
+      "step": 24402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0354531003073464e-05,
+      "loss": 0.3558,
+      "step": 24403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0349656384715112e-05,
+      "loss": 0.3974,
+      "step": 24404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0344782851424761e-05,
+      "loss": 0.4026,
+      "step": 24405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0339910403261388e-05,
+      "loss": 0.3921,
+      "step": 24406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0335039040284e-05,
+      "loss": 0.3831,
+      "step": 24407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0330168762551518e-05,
+      "loss": 0.4527,
+      "step": 24408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0325299570122915e-05,
+      "loss": 0.4679,
+      "step": 24409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0320431463057112e-05,
+      "loss": 0.5135,
+      "step": 24410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0315564441413006e-05,
+      "loss": 0.4633,
+      "step": 24411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.031069850524955e-05,
+      "loss": 0.4309,
+      "step": 24412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0305833654625618e-05,
+      "loss": 0.4755,
+      "step": 24413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0300969889600099e-05,
+      "loss": 0.4628,
+      "step": 24414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0296107210231843e-05,
+      "loss": 0.388,
+      "step": 24415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0291245616579693e-05,
+      "loss": 0.4533,
+      "step": 24416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0286385108702522e-05,
+      "loss": 0.3296,
+      "step": 24417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0281525686659166e-05,
+      "loss": 0.3645,
+      "step": 24418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0276667350508406e-05,
+      "loss": 0.3353,
+      "step": 24419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0271810100309065e-05,
+      "loss": 0.3564,
+      "step": 24420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0266953936119927e-05,
+      "loss": 0.3829,
+      "step": 24421
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0262098857999746e-05,
+      "loss": 0.4364,
+      "step": 24422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.025724486600732e-05,
+      "loss": 0.43,
+      "step": 24423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0252391960201391e-05,
+      "loss": 0.4753,
+      "step": 24424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0247540140640687e-05,
+      "loss": 0.3512,
+      "step": 24425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0242689407383933e-05,
+      "loss": 0.4026,
+      "step": 24426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.023783976048982e-05,
+      "loss": 0.4265,
+      "step": 24427
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0232991200017083e-05,
+      "loss": 0.4095,
+      "step": 24428
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0228143726024386e-05,
+      "loss": 0.3576,
+      "step": 24429
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0223297338570391e-05,
+      "loss": 0.4147,
+      "step": 24430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0218452037713777e-05,
+      "loss": 0.4413,
+      "step": 24431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0213607823513149e-05,
+      "loss": 0.4808,
+      "step": 24432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0208764696027196e-05,
+      "loss": 0.4785,
+      "step": 24433
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0203922655314491e-05,
+      "loss": 0.4806,
+      "step": 24434
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0199081701433655e-05,
+      "loss": 0.4567,
+      "step": 24435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0194241834443285e-05,
+      "loss": 0.4367,
+      "step": 24436
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0189403054401925e-05,
+      "loss": 0.5509,
+      "step": 24437
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.018456536136818e-05,
+      "loss": 0.4188,
+      "step": 24438
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0179728755400608e-05,
+      "loss": 0.5136,
+      "step": 24439
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0174893236557714e-05,
+      "loss": 0.3932,
+      "step": 24440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0170058804898042e-05,
+      "loss": 0.3688,
+      "step": 24441
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0165225460480087e-05,
+      "loss": 0.3686,
+      "step": 24442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0160393203362385e-05,
+      "loss": 0.4608,
+      "step": 24443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0155562033603394e-05,
+      "loss": 0.4407,
+      "step": 24444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0150731951261594e-05,
+      "loss": 0.4056,
+      "step": 24445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0145902956395447e-05,
+      "loss": 0.4431,
+      "step": 24446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0141075049063386e-05,
+      "loss": 0.3852,
+      "step": 24447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0136248229323875e-05,
+      "loss": 0.3726,
+      "step": 24448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.013142249723531e-05,
+      "loss": 0.3762,
+      "step": 24449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0126597852856111e-05,
+      "loss": 0.3613,
+      "step": 24450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0121774296244656e-05,
+      "loss": 0.3837,
+      "step": 24451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0116951827459353e-05,
+      "loss": 0.3947,
+      "step": 24452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.011213044655852e-05,
+      "loss": 0.3971,
+      "step": 24453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0107310153600568e-05,
+      "loss": 0.3922,
+      "step": 24454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0102490948643827e-05,
+      "loss": 0.3643,
+      "step": 24455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0097672831746608e-05,
+      "loss": 0.4363,
+      "step": 24456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0092855802967227e-05,
+      "loss": 0.3926,
+      "step": 24457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0088039862363973e-05,
+      "loss": 0.4135,
+      "step": 24458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0083225009995179e-05,
+      "loss": 0.4864,
+      "step": 24459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0078411245919084e-05,
+      "loss": 0.3609,
+      "step": 24460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0073598570193977e-05,
+      "loss": 0.5854,
+      "step": 24461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0068786982878087e-05,
+      "loss": 0.5016,
+      "step": 24462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0063976484029636e-05,
+      "loss": 0.3986,
+      "step": 24463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0059167073706877e-05,
+      "loss": 0.3861,
+      "step": 24464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0054358751968019e-05,
+      "loss": 0.3858,
+      "step": 24465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0049551518871247e-05,
+      "loss": 0.4212,
+      "step": 24466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.004474537447475e-05,
+      "loss": 0.4287,
+      "step": 24467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0039940318836694e-05,
+      "loss": 0.3739,
+      "step": 24468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.003513635201524e-05,
+      "loss": 0.4937,
+      "step": 24469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0030333474068521e-05,
+      "loss": 0.4504,
+      "step": 24470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.002553168505469e-05,
+      "loss": 0.5122,
+      "step": 24471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0020730985031824e-05,
+      "loss": 0.4261,
+      "step": 24472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0015931374058062e-05,
+      "loss": 0.4801,
+      "step": 24473
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0011132852191507e-05,
+      "loss": 0.5024,
+      "step": 24474
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.000633541949021e-05,
+      "loss": 0.5576,
+      "step": 24475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0001539076012244e-05,
+      "loss": 0.4282,
+      "step": 24476
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.996743821815634e-06,
+      "loss": 0.5116,
+      "step": 24477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.991949656958477e-06,
+      "loss": 0.5082,
+      "step": 24478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.987156581498758e-06,
+      "loss": 0.4136,
+      "step": 24479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.982364595494499e-06,
+      "loss": 0.3535,
+      "step": 24480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.977573699003706e-06,
+      "loss": 0.3564,
+      "step": 24481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.972783892084325e-06,
+      "loss": 0.3765,
+      "step": 24482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.967995174794386e-06,
+      "loss": 0.4378,
+      "step": 24483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.963207547191823e-06,
+      "loss": 0.3214,
+      "step": 24484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.958421009334573e-06,
+      "loss": 0.455,
+      "step": 24485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.953635561280594e-06,
+      "loss": 0.4307,
+      "step": 24486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.948851203087772e-06,
+      "loss": 0.3862,
+      "step": 24487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.944067934814028e-06,
+      "loss": 0.3983,
+      "step": 24488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.939285756517269e-06,
+      "loss": 0.4354,
+      "step": 24489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.934504668255362e-06,
+      "loss": 0.4007,
+      "step": 24490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.929724670086182e-06,
+      "loss": 0.4413,
+      "step": 24491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.924945762067584e-06,
+      "loss": 0.421,
+      "step": 24492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.920167944257374e-06,
+      "loss": 0.4202,
+      "step": 24493
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.915391216713432e-06,
+      "loss": 0.3971,
+      "step": 24494
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.910615579493555e-06,
+      "loss": 0.3559,
+      "step": 24495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.905841032655527e-06,
+      "loss": 0.469,
+      "step": 24496
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.901067576257162e-06,
+      "loss": 0.4116,
+      "step": 24497
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.896295210356187e-06,
+      "loss": 0.3701,
+      "step": 24498
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.891523935010415e-06,
+      "loss": 0.3597,
+      "step": 24499
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.886753750277578e-06,
+      "loss": 0.3864,
+      "step": 24500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.881984656215416e-06,
+      "loss": 0.4747,
+      "step": 24501
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.877216652881638e-06,
+      "loss": 0.3779,
+      "step": 24502
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.872449740333933e-06,
+      "loss": 0.4394,
+      "step": 24503
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.867683918630044e-06,
+      "loss": 0.5133,
+      "step": 24504
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.862919187827635e-06,
+      "loss": 0.4074,
+      "step": 24505
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.85815554798437e-06,
+      "loss": 0.4591,
+      "step": 24506
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.853392999157907e-06,
+      "loss": 0.5345,
+      "step": 24507
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.848631541405873e-06,
+      "loss": 0.5652,
+      "step": 24508
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.843871174785924e-06,
+      "loss": 0.3629,
+      "step": 24509
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.839111899355669e-06,
+      "loss": 0.3572,
+      "step": 24510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.834353715172706e-06,
+      "loss": 0.512,
+      "step": 24511
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.829596622294624e-06,
+      "loss": 0.4444,
+      "step": 24512
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.824840620778974e-06,
+      "loss": 0.4199,
+      "step": 24513
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.82008571068338e-06,
+      "loss": 0.441,
+      "step": 24514
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.81533189206535e-06,
+      "loss": 0.3716,
+      "step": 24515
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.810579164982436e-06,
+      "loss": 0.4687,
+      "step": 24516
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.805827529492161e-06,
+      "loss": 0.4222,
+      "step": 24517
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.801076985652003e-06,
+      "loss": 0.4268,
+      "step": 24518
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.796327533519501e-06,
+      "loss": 0.5625,
+      "step": 24519
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.791579173152132e-06,
+      "loss": 0.4793,
+      "step": 24520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.786831904607363e-06,
+      "loss": 0.4742,
+      "step": 24521
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.782085727942647e-06,
+      "loss": 0.423,
+      "step": 24522
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.777340643215427e-06,
+      "loss": 0.4033,
+      "step": 24523
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.772596650483124e-06,
+      "loss": 0.3444,
+      "step": 24524
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.76785374980319e-06,
+      "loss": 0.445,
+      "step": 24525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.763111941233006e-06,
+      "loss": 0.4136,
+      "step": 24526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.758371224829965e-06,
+      "loss": 0.4591,
+      "step": 24527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.753631600651458e-06,
+      "loss": 0.3883,
+      "step": 24528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.748893068754816e-06,
+      "loss": 0.5092,
+      "step": 24529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.744155629197426e-06,
+      "loss": 0.3909,
+      "step": 24530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.739419282036632e-06,
+      "loss": 0.4447,
+      "step": 24531
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.734684027329733e-06,
+      "loss": 0.4326,
+      "step": 24532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.729949865134059e-06,
+      "loss": 0.4138,
+      "step": 24533
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.725216795506874e-06,
+      "loss": 0.4643,
+      "step": 24534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.720484818505515e-06,
+      "loss": 0.3765,
+      "step": 24535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.715753934187233e-06,
+      "loss": 0.386,
+      "step": 24536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.711024142609292e-06,
+      "loss": 0.455,
+      "step": 24537
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.706295443828927e-06,
+      "loss": 0.4495,
+      "step": 24538
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.701567837903359e-06,
+      "loss": 0.4158,
+      "step": 24539
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.696841324889838e-06,
+      "loss": 0.3909,
+      "step": 24540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.692115904845568e-06,
+      "loss": 0.3746,
+      "step": 24541
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.687391577827732e-06,
+      "loss": 0.4649,
+      "step": 24542
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.682668343893497e-06,
+      "loss": 0.3716,
+      "step": 24543
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.67794620310004e-06,
+      "loss": 0.4393,
+      "step": 24544
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.673225155504528e-06,
+      "loss": 0.4516,
+      "step": 24545
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.66850520116409e-06,
+      "loss": 0.4288,
+      "step": 24546
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.66378634013586e-06,
+      "loss": 0.3669,
+      "step": 24547
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.659068572476948e-06,
+      "loss": 0.3466,
+      "step": 24548
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.65435189824443e-06,
+      "loss": 0.3913,
+      "step": 24549
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.649636317495437e-06,
+      "loss": 0.3603,
+      "step": 24550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.644921830287023e-06,
+      "loss": 0.4115,
+      "step": 24551
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.640208436676257e-06,
+      "loss": 0.4558,
+      "step": 24552
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.635496136720168e-06,
+      "loss": 0.4949,
+      "step": 24553
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.6307849304758e-06,
+      "loss": 0.3497,
+      "step": 24554
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.626074818000164e-06,
+      "loss": 0.4622,
+      "step": 24555
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.62136579935029e-06,
+      "loss": 0.4295,
+      "step": 24556
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.616657874583168e-06,
+      "loss": 0.489,
+      "step": 24557
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.611951043755785e-06,
+      "loss": 0.4114,
+      "step": 24558
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.607245306925084e-06,
+      "loss": 0.4838,
+      "step": 24559
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.602540664148018e-06,
+      "loss": 0.4398,
+      "step": 24560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.597837115481568e-06,
+      "loss": 0.3499,
+      "step": 24561
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.59313466098264e-06,
+      "loss": 0.4105,
+      "step": 24562
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.588433300708145e-06,
+      "loss": 0.446,
+      "step": 24563
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.583733034714981e-06,
+      "loss": 0.3893,
+      "step": 24564
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.579033863060038e-06,
+      "loss": 0.4724,
+      "step": 24565
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.574335785800226e-06,
+      "loss": 0.4323,
+      "step": 24566
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.569638802992365e-06,
+      "loss": 0.4003,
+      "step": 24567
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.564942914693331e-06,
+      "loss": 0.4503,
+      "step": 24568
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.560248120959935e-06,
+      "loss": 0.4016,
+      "step": 24569
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.555554421849011e-06,
+      "loss": 0.4065,
+      "step": 24570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.550861817417377e-06,
+      "loss": 0.3697,
+      "step": 24571
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.546170307721825e-06,
+      "loss": 0.4605,
+      "step": 24572
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.541479892819138e-06,
+      "loss": 0.3669,
+      "step": 24573
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.536790572766085e-06,
+      "loss": 0.3495,
+      "step": 24574
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.532102347619398e-06,
+      "loss": 0.393,
+      "step": 24575
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.527415217435864e-06,
+      "loss": 0.3623,
+      "step": 24576
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.522729182272184e-06,
+      "loss": 0.417,
+      "step": 24577
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.518044242185086e-06,
+      "loss": 0.4778,
+      "step": 24578
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.513360397231274e-06,
+      "loss": 0.4222,
+      "step": 24579
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.508677647467412e-06,
+      "loss": 0.4922,
+      "step": 24580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.503995992950221e-06,
+      "loss": 0.4592,
+      "step": 24581
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.499315433736345e-06,
+      "loss": 0.4136,
+      "step": 24582
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.494635969882426e-06,
+      "loss": 0.4337,
+      "step": 24583
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.48995760144511e-06,
+      "loss": 0.4512,
+      "step": 24584
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.485280328481016e-06,
+      "loss": 0.4744,
+      "step": 24585
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.480604151046735e-06,
+      "loss": 0.4536,
+      "step": 24586
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.475929069198908e-06,
+      "loss": 0.4494,
+      "step": 24587
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.471255082994091e-06,
+      "loss": 0.2934,
+      "step": 24588
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.466582192488872e-06,
+      "loss": 0.3601,
+      "step": 24589
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.461910397739781e-06,
+      "loss": 0.4161,
+      "step": 24590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.457239698803377e-06,
+      "loss": 0.4153,
+      "step": 24591
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.452570095736202e-06,
+      "loss": 0.3908,
+      "step": 24592
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.447901588594765e-06,
+      "loss": 0.3817,
+      "step": 24593
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.443234177435567e-06,
+      "loss": 0.4261,
+      "step": 24594
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.438567862315107e-06,
+      "loss": 0.3551,
+      "step": 24595
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.433902643289838e-06,
+      "loss": 0.4182,
+      "step": 24596
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.429238520416262e-06,
+      "loss": 0.4471,
+      "step": 24597
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.424575493750809e-06,
+      "loss": 0.4546,
+      "step": 24598
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.419913563349924e-06,
+      "loss": 0.4501,
+      "step": 24599
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.41525272927003e-06,
+      "loss": 0.4015,
+      "step": 24600
+    },
+    {
+      "epoch": 0.86,
+      "eval_loss": 0.40309014916419983,
+      "eval_runtime": 847.4282,
+      "eval_samples_per_second": 28.392,
+      "eval_steps_per_second": 14.196,
+      "step": 24600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.410592991567525e-06,
+      "loss": 0.4846,
+      "step": 24601
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.405934350298828e-06,
+      "loss": 0.3699,
+      "step": 24602
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.401276805520342e-06,
+      "loss": 0.3733,
+      "step": 24603
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.396620357288388e-06,
+      "loss": 0.391,
+      "step": 24604
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.391965005659342e-06,
+      "loss": 0.4502,
+      "step": 24605
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.38731075068956e-06,
+      "loss": 0.4119,
+      "step": 24606
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.382657592435384e-06,
+      "loss": 0.4034,
+      "step": 24607
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.378005530953116e-06,
+      "loss": 0.506,
+      "step": 24608
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.373354566299064e-06,
+      "loss": 0.3382,
+      "step": 24609
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.368704698529496e-06,
+      "loss": 0.5032,
+      "step": 24610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.364055927700744e-06,
+      "loss": 0.4292,
+      "step": 24611
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.35940825386904e-06,
+      "loss": 0.3717,
+      "step": 24612
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.35476167709065e-06,
+      "loss": 0.3905,
+      "step": 24613
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.350116197421799e-06,
+      "loss": 0.44,
+      "step": 24614
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.345471814918694e-06,
+      "loss": 0.4823,
+      "step": 24615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.340828529637602e-06,
+      "loss": 0.4048,
+      "step": 24616
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.33618634163469e-06,
+      "loss": 0.3235,
+      "step": 24617
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.331545250966145e-06,
+      "loss": 0.4432,
+      "step": 24618
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.326905257688135e-06,
+      "loss": 0.4264,
+      "step": 24619
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.322266361856813e-06,
+      "loss": 0.4004,
+      "step": 24620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.317628563528357e-06,
+      "loss": 0.367,
+      "step": 24621
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.31299186275888e-06,
+      "loss": 0.3443,
+      "step": 24622
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.3083562596045e-06,
+      "loss": 0.4242,
+      "step": 24623
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.30372175412133e-06,
+      "loss": 0.4167,
+      "step": 24624
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.29908834636546e-06,
+      "loss": 0.4238,
+      "step": 24625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.29445603639294e-06,
+      "loss": 0.4108,
+      "step": 24626
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.289824824259897e-06,
+      "loss": 0.3837,
+      "step": 24627
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.285194710022338e-06,
+      "loss": 0.3778,
+      "step": 24628
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.280565693736321e-06,
+      "loss": 0.4735,
+      "step": 24629
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.275937775457865e-06,
+      "loss": 0.3973,
+      "step": 24630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.271310955242973e-06,
+      "loss": 0.3743,
+      "step": 24631
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.266685233147665e-06,
+      "loss": 0.3807,
+      "step": 24632
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.26206060922793e-06,
+      "loss": 0.3328,
+      "step": 24633
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.257437083539732e-06,
+      "loss": 0.4418,
+      "step": 24634
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.25281465613903e-06,
+      "loss": 0.4934,
+      "step": 24635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.248193327081755e-06,
+      "loss": 0.4293,
+      "step": 24636
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.243573096423874e-06,
+      "loss": 0.3047,
+      "step": 24637
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.238953964221298e-06,
+      "loss": 0.424,
+      "step": 24638
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.234335930529914e-06,
+      "loss": 0.3659,
+      "step": 24639
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.229718995405633e-06,
+      "loss": 0.3859,
+      "step": 24640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.225103158904313e-06,
+      "loss": 0.3801,
+      "step": 24641
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.220488421081852e-06,
+      "loss": 0.4538,
+      "step": 24642
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.215874781994093e-06,
+      "loss": 0.4015,
+      "step": 24643
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.211262241696883e-06,
+      "loss": 0.3799,
+      "step": 24644
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.20665080024603e-06,
+      "loss": 0.349,
+      "step": 24645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.202040457697337e-06,
+      "loss": 0.3943,
+      "step": 24646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.197431214106645e-06,
+      "loss": 0.4557,
+      "step": 24647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.192823069529722e-06,
+      "loss": 0.3506,
+      "step": 24648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.188216024022344e-06,
+      "loss": 0.427,
+      "step": 24649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.18361007764027e-06,
+      "loss": 0.4623,
+      "step": 24650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.17900523043922e-06,
+      "loss": 0.3891,
+      "step": 24651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.174401482474982e-06,
+      "loss": 0.3529,
+      "step": 24652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.169798833803234e-06,
+      "loss": 0.5083,
+      "step": 24653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.16519728447971e-06,
+      "loss": 0.4677,
+      "step": 24654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.160596834560086e-06,
+      "loss": 0.4533,
+      "step": 24655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.155997484100055e-06,
+      "loss": 0.4905,
+      "step": 24656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.151399233155255e-06,
+      "loss": 0.3545,
+      "step": 24657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.146802081781391e-06,
+      "loss": 0.3698,
+      "step": 24658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.142206030034062e-06,
+      "loss": 0.4107,
+      "step": 24659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.137611077968923e-06,
+      "loss": 0.4209,
+      "step": 24660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.133017225641571e-06,
+      "loss": 0.4534,
+      "step": 24661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.128424473107589e-06,
+      "loss": 0.4258,
+      "step": 24662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.123832820422606e-06,
+      "loss": 0.3876,
+      "step": 24663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.119242267642181e-06,
+      "loss": 0.3412,
+      "step": 24664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.114652814821866e-06,
+      "loss": 0.4481,
+      "step": 24665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.110064462017209e-06,
+      "loss": 0.486,
+      "step": 24666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.105477209283742e-06,
+      "loss": 0.4047,
+      "step": 24667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.100891056677007e-06,
+      "loss": 0.4358,
+      "step": 24668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.096306004252498e-06,
+      "loss": 0.5151,
+      "step": 24669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.091722052065709e-06,
+      "loss": 0.3966,
+      "step": 24670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.087139200172124e-06,
+      "loss": 0.5623,
+      "step": 24671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.082557448627183e-06,
+      "loss": 0.4025,
+      "step": 24672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.077976797486386e-06,
+      "loss": 0.4195,
+      "step": 24673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.07339724680516e-06,
+      "loss": 0.4436,
+      "step": 24674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.068818796638923e-06,
+      "loss": 0.407,
+      "step": 24675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.064241447043098e-06,
+      "loss": 0.3757,
+      "step": 24676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.059665198073052e-06,
+      "loss": 0.3707,
+      "step": 24677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.055090049784231e-06,
+      "loss": 0.4172,
+      "step": 24678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.050516002231969e-06,
+      "loss": 0.4046,
+      "step": 24679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.045943055471651e-06,
+      "loss": 0.489,
+      "step": 24680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.041371209558603e-06,
+      "loss": 0.3859,
+      "step": 24681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.036800464548157e-06,
+      "loss": 0.3619,
+      "step": 24682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.032230820495668e-06,
+      "loss": 0.385,
+      "step": 24683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.027662277456417e-06,
+      "loss": 0.46,
+      "step": 24684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.023094835485701e-06,
+      "loss": 0.399,
+      "step": 24685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.018528494638811e-06,
+      "loss": 0.4152,
+      "step": 24686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.013963254971002e-06,
+      "loss": 0.3221,
+      "step": 24687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.00939911653752e-06,
+      "loss": 0.3674,
+      "step": 24688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.004836079393642e-06,
+      "loss": 0.3507,
+      "step": 24689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.000274143594578e-06,
+      "loss": 0.4306,
+      "step": 24690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.99571330919553e-06,
+      "loss": 0.3599,
+      "step": 24691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.991153576251721e-06,
+      "loss": 0.4177,
+      "step": 24692
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.986594944818305e-06,
+      "loss": 0.3715,
+      "step": 24693
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.982037414950495e-06,
+      "loss": 0.3584,
+      "step": 24694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.977480986703434e-06,
+      "loss": 0.4145,
+      "step": 24695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.97292566013227e-06,
+      "loss": 0.3718,
+      "step": 24696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.968371435292144e-06,
+      "loss": 0.4178,
+      "step": 24697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.963818312238159e-06,
+      "loss": 0.5303,
+      "step": 24698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.959266291025448e-06,
+      "loss": 0.4405,
+      "step": 24699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.9547153717091e-06,
+      "loss": 0.3891,
+      "step": 24700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.950165554344192e-06,
+      "loss": 0.4673,
+      "step": 24701
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.945616838985782e-06,
+      "loss": 0.4501,
+      "step": 24702
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.941069225688925e-06,
+      "loss": 0.4295,
+      "step": 24703
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.936522714508678e-06,
+      "loss": 0.4495,
+      "step": 24704
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.931977305500072e-06,
+      "loss": 0.4353,
+      "step": 24705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.9274329987181e-06,
+      "loss": 0.3787,
+      "step": 24706
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.922889794217782e-06,
+      "loss": 0.4078,
+      "step": 24707
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.918347692054086e-06,
+      "loss": 0.3667,
+      "step": 24708
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.913806692282012e-06,
+      "loss": 0.4868,
+      "step": 24709
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.909266794956506e-06,
+      "loss": 0.4605,
+      "step": 24710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.904728000132523e-06,
+      "loss": 0.368,
+      "step": 24711
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.900190307864986e-06,
+      "loss": 0.4746,
+      "step": 24712
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.895653718208807e-06,
+      "loss": 0.5005,
+      "step": 24713
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.891118231218932e-06,
+      "loss": 0.4664,
+      "step": 24714
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.886583846950225e-06,
+      "loss": 0.4212,
+      "step": 24715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.882050565457589e-06,
+      "loss": 0.4656,
+      "step": 24716
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.87751838679587e-06,
+      "loss": 0.3811,
+      "step": 24717
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.87298731101991e-06,
+      "loss": 0.4499,
+      "step": 24718
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.868457338184588e-06,
+      "loss": 0.4249,
+      "step": 24719
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.863928468344718e-06,
+      "loss": 0.3731,
+      "step": 24720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.85940070155512e-06,
+      "loss": 0.4071,
+      "step": 24721
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.854874037870575e-06,
+      "loss": 0.4286,
+      "step": 24722
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.850348477345882e-06,
+      "loss": 0.3393,
+      "step": 24723
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.845824020035799e-06,
+      "loss": 0.381,
+      "step": 24724
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.841300665995123e-06,
+      "loss": 0.455,
+      "step": 24725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.83677841527858e-06,
+      "loss": 0.409,
+      "step": 24726
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.832257267940913e-06,
+      "loss": 0.4148,
+      "step": 24727
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.827737224036824e-06,
+      "loss": 0.3575,
+      "step": 24728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.823218283621027e-06,
+      "loss": 0.3674,
+      "step": 24729
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.81870044674823e-06,
+      "loss": 0.3868,
+      "step": 24730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.814183713473112e-06,
+      "loss": 0.4544,
+      "step": 24731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.809668083850331e-06,
+      "loss": 0.3762,
+      "step": 24732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.805153557934542e-06,
+      "loss": 0.4556,
+      "step": 24733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.80064013578038e-06,
+      "loss": 0.4663,
+      "step": 24734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.796127817442489e-06,
+      "loss": 0.3902,
+      "step": 24735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.791616602975483e-06,
+      "loss": 0.352,
+      "step": 24736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.78710649243395e-06,
+      "loss": 0.3764,
+      "step": 24737
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.78259748587248e-06,
+      "loss": 0.4397,
+      "step": 24738
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.778089583345661e-06,
+      "loss": 0.4279,
+      "step": 24739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.77358278490803e-06,
+      "loss": 0.3886,
+      "step": 24740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.769077090614142e-06,
+      "loss": 0.4674,
+      "step": 24741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.764572500518542e-06,
+      "loss": 0.4398,
+      "step": 24742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.760069014675731e-06,
+      "loss": 0.4321,
+      "step": 24743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.755566633140245e-06,
+      "loss": 0.3818,
+      "step": 24744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.751065355966547e-06,
+      "loss": 0.4037,
+      "step": 24745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.746565183209143e-06,
+      "loss": 0.4192,
+      "step": 24746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.742066114922499e-06,
+      "loss": 0.3196,
+      "step": 24747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.737568151161024e-06,
+      "loss": 0.3965,
+      "step": 24748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.733071291979223e-06,
+      "loss": 0.4338,
+      "step": 24749
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.728575537431483e-06,
+      "loss": 0.4965,
+      "step": 24750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.724080887572227e-06,
+      "loss": 0.496,
+      "step": 24751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.719587342455859e-06,
+      "loss": 0.3685,
+      "step": 24752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.715094902136756e-06,
+      "loss": 0.3825,
+      "step": 24753
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.710603566669306e-06,
+      "loss": 0.4424,
+      "step": 24754
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.706113336107857e-06,
+      "loss": 0.4639,
+      "step": 24755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.701624210506765e-06,
+      "loss": 0.3906,
+      "step": 24756
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.697136189920362e-06,
+      "loss": 0.3697,
+      "step": 24757
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.692649274402963e-06,
+      "loss": 0.4353,
+      "step": 24758
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.688163464008858e-06,
+      "loss": 0.444,
+      "step": 24759
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.68367875879238e-06,
+      "loss": 0.4142,
+      "step": 24760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.679195158807784e-06,
+      "loss": 0.3453,
+      "step": 24761
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.67471266410933e-06,
+      "loss": 0.4022,
+      "step": 24762
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.670231274751295e-06,
+      "loss": 0.4456,
+      "step": 24763
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.66575099078788e-06,
+      "loss": 0.5193,
+      "step": 24764
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.661271812273353e-06,
+      "loss": 0.3841,
+      "step": 24765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.656793739261915e-06,
+      "loss": 0.4831,
+      "step": 24766
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.652316771807767e-06,
+      "loss": 0.3795,
+      "step": 24767
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.647840909965078e-06,
+      "loss": 0.4104,
+      "step": 24768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.643366153788013e-06,
+      "loss": 0.405,
+      "step": 24769
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.638892503330775e-06,
+      "loss": 0.3532,
+      "step": 24770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.634419958647477e-06,
+      "loss": 0.4041,
+      "step": 24771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.629948519792274e-06,
+      "loss": 0.3523,
+      "step": 24772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.625478186819258e-06,
+      "loss": 0.5137,
+      "step": 24773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.621008959782539e-06,
+      "loss": 0.4044,
+      "step": 24774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.616540838736231e-06,
+      "loss": 0.4095,
+      "step": 24775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.612073823734401e-06,
+      "loss": 0.4754,
+      "step": 24776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.607607914831117e-06,
+      "loss": 0.4077,
+      "step": 24777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.603143112080436e-06,
+      "loss": 0.417,
+      "step": 24778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.59867941553636e-06,
+      "loss": 0.4193,
+      "step": 24779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.594216825252966e-06,
+      "loss": 0.4628,
+      "step": 24780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.589755341284256e-06,
+      "loss": 0.5183,
+      "step": 24781
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.58529496368421e-06,
+      "loss": 0.4239,
+      "step": 24782
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.580835692506817e-06,
+      "loss": 0.4306,
+      "step": 24783
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.576377527806035e-06,
+      "loss": 0.3973,
+      "step": 24784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.571920469635864e-06,
+      "loss": 0.4082,
+      "step": 24785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.567464518050228e-06,
+      "loss": 0.3449,
+      "step": 24786
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.56300967310305e-06,
+      "loss": 0.3419,
+      "step": 24787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.558555934848266e-06,
+      "loss": 0.3266,
+      "step": 24788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.554103303339756e-06,
+      "loss": 0.4454,
+      "step": 24789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.549651778631418e-06,
+      "loss": 0.3553,
+      "step": 24790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.545201360777155e-06,
+      "loss": 0.4225,
+      "step": 24791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.540752049830825e-06,
+      "loss": 0.3721,
+      "step": 24792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.53630384584626e-06,
+      "loss": 0.426,
+      "step": 24793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.531856748877309e-06,
+      "loss": 0.3487,
+      "step": 24794
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.527410758977772e-06,
+      "loss": 0.3639,
+      "step": 24795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.522965876201505e-06,
+      "loss": 0.454,
+      "step": 24796
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.518522100602288e-06,
+      "loss": 0.3657,
+      "step": 24797
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.514079432233913e-06,
+      "loss": 0.3438,
+      "step": 24798
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.509637871150122e-06,
+      "loss": 0.3852,
+      "step": 24799
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.505197417404687e-06,
+      "loss": 0.3576,
+      "step": 24800
+    },
+    {
+      "epoch": 0.87,
+      "eval_loss": 0.40213853120803833,
+      "eval_runtime": 846.1675,
+      "eval_samples_per_second": 28.434,
+      "eval_steps_per_second": 14.217,
+      "step": 24800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.500758071051362e-06,
+      "loss": 0.4403,
+      "step": 24801
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.496319832143884e-06,
+      "loss": 0.3533,
+      "step": 24802
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.491882700735954e-06,
+      "loss": 0.4735,
+      "step": 24803
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.487446676881282e-06,
+      "loss": 0.4552,
+      "step": 24804
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.483011760633542e-06,
+      "loss": 0.4295,
+      "step": 24805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.47857795204644e-06,
+      "loss": 0.4298,
+      "step": 24806
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.474145251173626e-06,
+      "loss": 0.3833,
+      "step": 24807
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.469713658068757e-06,
+      "loss": 0.4714,
+      "step": 24808
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.465283172785465e-06,
+      "loss": 0.408,
+      "step": 24809
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.460853795377343e-06,
+      "loss": 0.4131,
+      "step": 24810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.45642552589806e-06,
+      "loss": 0.4861,
+      "step": 24811
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.451998364401181e-06,
+      "loss": 0.4073,
+      "step": 24812
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.4475723109403e-06,
+      "loss": 0.3789,
+      "step": 24813
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.44314736556897e-06,
+      "loss": 0.5924,
+      "step": 24814
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.438723528340742e-06,
+      "loss": 0.4621,
+      "step": 24815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.434300799309191e-06,
+      "loss": 0.3977,
+      "step": 24816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.429879178527833e-06,
+      "loss": 0.4799,
+      "step": 24817
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.425458666050178e-06,
+      "loss": 0.3617,
+      "step": 24818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.42103926192973e-06,
+      "loss": 0.4612,
+      "step": 24819
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.416620966219968e-06,
+      "loss": 0.3884,
+      "step": 24820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.412203778974404e-06,
+      "loss": 0.4048,
+      "step": 24821
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.407787700246471e-06,
+      "loss": 0.394,
+      "step": 24822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.40337273008962e-06,
+      "loss": 0.5365,
+      "step": 24823
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.398958868557305e-06,
+      "loss": 0.422,
+      "step": 24824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.394546115702928e-06,
+      "loss": 0.3953,
+      "step": 24825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.39013447157988e-06,
+      "loss": 0.5007,
+      "step": 24826
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.385723936241608e-06,
+      "loss": 0.3752,
+      "step": 24827
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.381314509741467e-06,
+      "loss": 0.423,
+      "step": 24828
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.376906192132828e-06,
+      "loss": 0.3731,
+      "step": 24829
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.372498983469034e-06,
+      "loss": 0.4828,
+      "step": 24830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.368092883803435e-06,
+      "loss": 0.4026,
+      "step": 24831
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.363687893189375e-06,
+      "loss": 0.4481,
+      "step": 24832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.359284011680157e-06,
+      "loss": 0.324,
+      "step": 24833
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.354881239329081e-06,
+      "loss": 0.4608,
+      "step": 24834
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.350479576189429e-06,
+      "loss": 0.3834,
+      "step": 24835
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.346079022314479e-06,
+      "loss": 0.4821,
+      "step": 24836
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.34167957775751e-06,
+      "loss": 0.431,
+      "step": 24837
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.337281242571749e-06,
+      "loss": 0.43,
+      "step": 24838
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.332884016810449e-06,
+      "loss": 0.4086,
+      "step": 24839
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.328487900526816e-06,
+      "loss": 0.4168,
+      "step": 24840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.324092893774038e-06,
+      "loss": 0.4245,
+      "step": 24841
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.319698996605363e-06,
+      "loss": 0.396,
+      "step": 24842
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.315306209073925e-06,
+      "loss": 0.3993,
+      "step": 24843
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.310914531232927e-06,
+      "loss": 0.4989,
+      "step": 24844
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.306523963135493e-06,
+      "loss": 0.564,
+      "step": 24845
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.302134504834758e-06,
+      "loss": 0.3714,
+      "step": 24846
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.297746156383879e-06,
+      "loss": 0.3847,
+      "step": 24847
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.29335891783597e-06,
+      "loss": 0.4633,
+      "step": 24848
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.288972789244109e-06,
+      "loss": 0.4544,
+      "step": 24849
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.28458777066139e-06,
+      "loss": 0.3709,
+      "step": 24850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.280203862140868e-06,
+      "loss": 0.5008,
+      "step": 24851
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.275821063735645e-06,
+      "loss": 0.408,
+      "step": 24852
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.271439375498747e-06,
+      "loss": 0.492,
+      "step": 24853
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.267058797483196e-06,
+      "loss": 0.3947,
+      "step": 24854
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.26267932974203e-06,
+      "loss": 0.4362,
+      "step": 24855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.258300972328248e-06,
+      "loss": 0.5343,
+      "step": 24856
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.25392372529481e-06,
+      "loss": 0.4296,
+      "step": 24857
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.24954758869475e-06,
+      "loss": 0.3454,
+      "step": 24858
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.245172562581017e-06,
+      "loss": 0.3967,
+      "step": 24859
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.240798647006554e-06,
+      "loss": 0.3822,
+      "step": 24860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.236425842024287e-06,
+      "loss": 0.4781,
+      "step": 24861
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.232054147687151e-06,
+      "loss": 0.4367,
+      "step": 24862
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.227683564048084e-06,
+      "loss": 0.3903,
+      "step": 24863
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.223314091159962e-06,
+      "loss": 0.4963,
+      "step": 24864
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.218945729075666e-06,
+      "loss": 0.4267,
+      "step": 24865
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.21457847784809e-06,
+      "loss": 0.3713,
+      "step": 24866
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.210212337530044e-06,
+      "loss": 0.4645,
+      "step": 24867
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.20584730817443e-06,
+      "loss": 0.3448,
+      "step": 24868
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.201483389834052e-06,
+      "loss": 0.4019,
+      "step": 24869
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.197120582561724e-06,
+      "loss": 0.3977,
+      "step": 24870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.192758886410268e-06,
+      "loss": 0.464,
+      "step": 24871
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.188398301432465e-06,
+      "loss": 0.4396,
+      "step": 24872
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.184038827681084e-06,
+      "loss": 0.397,
+      "step": 24873
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.179680465208895e-06,
+      "loss": 0.3909,
+      "step": 24874
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.175323214068664e-06,
+      "loss": 0.3994,
+      "step": 24875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.170967074313085e-06,
+      "loss": 0.4108,
+      "step": 24876
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.166612045994936e-06,
+      "loss": 0.4618,
+      "step": 24877
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.1622581291669e-06,
+      "loss": 0.4758,
+      "step": 24878
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.157905323881676e-06,
+      "loss": 0.364,
+      "step": 24879
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.153553630191956e-06,
+      "loss": 0.4098,
+      "step": 24880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.149203048150377e-06,
+      "loss": 0.482,
+      "step": 24881
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.144853577809641e-06,
+      "loss": 0.3952,
+      "step": 24882
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.140505219222372e-06,
+      "loss": 0.3721,
+      "step": 24883
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.136157972441206e-06,
+      "loss": 0.374,
+      "step": 24884
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.131811837518744e-06,
+      "loss": 0.4845,
+      "step": 24885
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.127466814507578e-06,
+      "loss": 0.3843,
+      "step": 24886
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.123122903460334e-06,
+      "loss": 0.3811,
+      "step": 24887
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.118780104429568e-06,
+      "loss": 0.4296,
+      "step": 24888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.11443841746785e-06,
+      "loss": 0.4867,
+      "step": 24889
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.110097842627728e-06,
+      "loss": 0.3952,
+      "step": 24890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.105758379961726e-06,
+      "loss": 0.4322,
+      "step": 24891
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.101420029522356e-06,
+      "loss": 0.4706,
+      "step": 24892
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.097082791362154e-06,
+      "loss": 0.421,
+      "step": 24893
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.092746665533613e-06,
+      "loss": 0.3535,
+      "step": 24894
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.088411652089201e-06,
+      "loss": 0.3462,
+      "step": 24895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.084077751081387e-06,
+      "loss": 0.5381,
+      "step": 24896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.079744962562619e-06,
+      "loss": 0.3924,
+      "step": 24897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.075413286585364e-06,
+      "loss": 0.4638,
+      "step": 24898
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.071082723202028e-06,
+      "loss": 0.3692,
+      "step": 24899
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.066753272465032e-06,
+      "loss": 0.4905,
+      "step": 24900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.062424934426783e-06,
+      "loss": 0.439,
+      "step": 24901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.058097709139644e-06,
+      "loss": 0.4189,
+      "step": 24902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.053771596656023e-06,
+      "loss": 0.3947,
+      "step": 24903
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.049446597028265e-06,
+      "loss": 0.4025,
+      "step": 24904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.045122710308716e-06,
+      "loss": 0.4097,
+      "step": 24905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.040799936549703e-06,
+      "loss": 0.4338,
+      "step": 24906
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.036478275803538e-06,
+      "loss": 0.4389,
+      "step": 24907
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.032157728122559e-06,
+      "loss": 0.4379,
+      "step": 24908
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.027838293559043e-06,
+      "loss": 0.3945,
+      "step": 24909
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.023519972165273e-06,
+      "loss": 0.4482,
+      "step": 24910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.019202763993505e-06,
+      "loss": 0.3587,
+      "step": 24911
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.014886669095977e-06,
+      "loss": 0.4502,
+      "step": 24912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.010571687524971e-06,
+      "loss": 0.5079,
+      "step": 24913
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.006257819332685e-06,
+      "loss": 0.3224,
+      "step": 24914
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.001945064571336e-06,
+      "loss": 0.4664,
+      "step": 24915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.997633423293128e-06,
+      "loss": 0.398,
+      "step": 24916
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.993322895550216e-06,
+      "loss": 0.4017,
+      "step": 24917
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.989013481394814e-06,
+      "loss": 0.4005,
+      "step": 24918
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.98470518087907e-06,
+      "loss": 0.3621,
+      "step": 24919
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.980397994055122e-06,
+      "loss": 0.3424,
+      "step": 24920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.976091920975093e-06,
+      "loss": 0.4114,
+      "step": 24921
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.971786961691097e-06,
+      "loss": 0.3939,
+      "step": 24922
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.96748311625527e-06,
+      "loss": 0.5262,
+      "step": 24923
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.96318038471967e-06,
+      "loss": 0.5309,
+      "step": 24924
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.958878767136401e-06,
+      "loss": 0.3934,
+      "step": 24925
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.95457826355751e-06,
+      "loss": 0.3434,
+      "step": 24926
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.950278874035055e-06,
+      "loss": 0.3393,
+      "step": 24927
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.945980598621039e-06,
+      "loss": 0.4199,
+      "step": 24928
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.941683437367542e-06,
+      "loss": 0.3986,
+      "step": 24929
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.937387390326545e-06,
+      "loss": 0.4507,
+      "step": 24930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.933092457550051e-06,
+      "loss": 0.4282,
+      "step": 24931
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.928798639090029e-06,
+      "loss": 0.4288,
+      "step": 24932
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.92450593499845e-06,
+      "loss": 0.484,
+      "step": 24933
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.920214345327293e-06,
+      "loss": 0.4308,
+      "step": 24934
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.915923870128482e-06,
+      "loss": 0.5114,
+      "step": 24935
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.911634509453958e-06,
+      "loss": 0.4591,
+      "step": 24936
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.907346263355619e-06,
+      "loss": 0.4546,
+      "step": 24937
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.90305913188536e-06,
+      "loss": 0.4499,
+      "step": 24938
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.898773115095116e-06,
+      "loss": 0.4123,
+      "step": 24939
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.894488213036722e-06,
+      "loss": 0.428,
+      "step": 24940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.890204425762048e-06,
+      "loss": 0.4271,
+      "step": 24941
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.885921753322944e-06,
+      "loss": 0.4356,
+      "step": 24942
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.881640195771234e-06,
+      "loss": 0.3914,
+      "step": 24943
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.877359753158754e-06,
+      "loss": 0.4637,
+      "step": 24944
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.873080425537316e-06,
+      "loss": 0.3884,
+      "step": 24945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.868802212958703e-06,
+      "loss": 0.4116,
+      "step": 24946
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.864525115474708e-06,
+      "loss": 0.3759,
+      "step": 24947
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.860249133137054e-06,
+      "loss": 0.4354,
+      "step": 24948
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.855974265997556e-06,
+      "loss": 0.4042,
+      "step": 24949
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.85170051410793e-06,
+      "loss": 0.4398,
+      "step": 24950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.847427877519897e-06,
+      "loss": 0.3743,
+      "step": 24951
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.843156356285164e-06,
+      "loss": 0.356,
+      "step": 24952
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.838885950455432e-06,
+      "loss": 0.3649,
+      "step": 24953
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.834616660082406e-06,
+      "loss": 0.4364,
+      "step": 24954
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.830348485217753e-06,
+      "loss": 0.4622,
+      "step": 24955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.826081425913124e-06,
+      "loss": 0.4788,
+      "step": 24956
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.821815482220163e-06,
+      "loss": 0.3816,
+      "step": 24957
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.817550654190509e-06,
+      "loss": 0.4148,
+      "step": 24958
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.813286941875764e-06,
+      "loss": 0.4437,
+      "step": 24959
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.809024345327553e-06,
+      "loss": 0.4999,
+      "step": 24960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.804762864597471e-06,
+      "loss": 0.4729,
+      "step": 24961
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.800502499737084e-06,
+      "loss": 0.3668,
+      "step": 24962
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.796243250797963e-06,
+      "loss": 0.449,
+      "step": 24963
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.791985117831636e-06,
+      "loss": 0.364,
+      "step": 24964
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.787728100889669e-06,
+      "loss": 0.3575,
+      "step": 24965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.783472200023589e-06,
+      "loss": 0.4373,
+      "step": 24966
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.77921741528489e-06,
+      "loss": 0.4174,
+      "step": 24967
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.774963746725073e-06,
+      "loss": 0.4415,
+      "step": 24968
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.770711194395596e-06,
+      "loss": 0.3747,
+      "step": 24969
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.766459758347988e-06,
+      "loss": 0.4322,
+      "step": 24970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.76220943863366e-06,
+      "loss": 0.4493,
+      "step": 24971
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.757960235304074e-06,
+      "loss": 0.3522,
+      "step": 24972
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.753712148410653e-06,
+      "loss": 0.3798,
+      "step": 24973
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.7494651780048e-06,
+      "loss": 0.4504,
+      "step": 24974
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.745219324137942e-06,
+      "loss": 0.4118,
+      "step": 24975
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.74097458686146e-06,
+      "loss": 0.4621,
+      "step": 24976
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.736730966226724e-06,
+      "loss": 0.4286,
+      "step": 24977
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.732488462285104e-06,
+      "loss": 0.4209,
+      "step": 24978
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.728247075087924e-06,
+      "loss": 0.4907,
+      "step": 24979
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.724006804686545e-06,
+      "loss": 0.4055,
+      "step": 24980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.719767651132292e-06,
+      "loss": 0.397,
+      "step": 24981
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.715529614476457e-06,
+      "loss": 0.4058,
+      "step": 24982
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.711292694770345e-06,
+      "loss": 0.3672,
+      "step": 24983
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.707056892065212e-06,
+      "loss": 0.5295,
+      "step": 24984
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.702822206412353e-06,
+      "loss": 0.3888,
+      "step": 24985
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.698588637863025e-06,
+      "loss": 0.4224,
+      "step": 24986
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.694356186468444e-06,
+      "loss": 0.4745,
+      "step": 24987
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.690124852279856e-06,
+      "loss": 0.5304,
+      "step": 24988
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.68589463534848e-06,
+      "loss": 0.4404,
+      "step": 24989
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.681665535725468e-06,
+      "loss": 0.455,
+      "step": 24990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.677437553462063e-06,
+      "loss": 0.3541,
+      "step": 24991
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.673210688609423e-06,
+      "loss": 0.4578,
+      "step": 24992
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.668984941218693e-06,
+      "loss": 0.4492,
+      "step": 24993
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.664760311341024e-06,
+      "loss": 0.3606,
+      "step": 24994
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.66053679902754e-06,
+      "loss": 0.418,
+      "step": 24995
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.65631440432939e-06,
+      "loss": 0.4807,
+      "step": 24996
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.652093127297654e-06,
+      "loss": 0.379,
+      "step": 24997
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.647872967983426e-06,
+      "loss": 0.4522,
+      "step": 24998
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.643653926437787e-06,
+      "loss": 0.3343,
+      "step": 24999
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.639436002711786e-06,
+      "loss": 0.3975,
+      "step": 25000
+    },
+    {
+      "epoch": 0.87,
+      "eval_loss": 0.4014919400215149,
+      "eval_runtime": 845.6441,
+      "eval_samples_per_second": 28.452,
+      "eval_steps_per_second": 14.226,
+      "step": 25000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.635219196856503e-06,
+      "loss": 0.415,
+      "step": 25001
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.631003508922963e-06,
+      "loss": 0.4527,
+      "step": 25002
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.626788938962193e-06,
+      "loss": 0.4387,
+      "step": 25003
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.622575487025185e-06,
+      "loss": 0.4192,
+      "step": 25004
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.618363153162944e-06,
+      "loss": 0.4292,
+      "step": 25005
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.614151937426472e-06,
+      "loss": 0.4012,
+      "step": 25006
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.609941839866741e-06,
+      "loss": 0.4288,
+      "step": 25007
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.605732860534675e-06,
+      "loss": 0.3819,
+      "step": 25008
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.6015249994812135e-06,
+      "loss": 0.4055,
+      "step": 25009
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.597318256757324e-06,
+      "loss": 0.3677,
+      "step": 25010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.593112632413901e-06,
+      "loss": 0.4701,
+      "step": 25011
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.588908126501859e-06,
+      "loss": 0.3751,
+      "step": 25012
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.584704739072068e-06,
+      "loss": 0.3982,
+      "step": 25013
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.580502470175399e-06,
+      "loss": 0.4151,
+      "step": 25014
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.576301319862733e-06,
+      "loss": 0.4467,
+      "step": 25015
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.572101288184919e-06,
+      "loss": 0.3669,
+      "step": 25016
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.567902375192781e-06,
+      "loss": 0.4269,
+      "step": 25017
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.563704580937147e-06,
+      "loss": 0.3891,
+      "step": 25018
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.559507905468799e-06,
+      "loss": 0.428,
+      "step": 25019
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5553123488385616e-06,
+      "loss": 0.3806,
+      "step": 25020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.551117911097205e-06,
+      "loss": 0.4179,
+      "step": 25021
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.546924592295501e-06,
+      "loss": 0.415,
+      "step": 25022
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5427323924841975e-06,
+      "loss": 0.4005,
+      "step": 25023
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.53854131171402e-06,
+      "loss": 0.4078,
+      "step": 25024
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.534351350035718e-06,
+      "loss": 0.4171,
+      "step": 25025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.530162507500005e-06,
+      "loss": 0.4626,
+      "step": 25026
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.525974784157552e-06,
+      "loss": 0.3457,
+      "step": 25027
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.521788180059075e-06,
+      "loss": 0.4053,
+      "step": 25028
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.517602695255232e-06,
+      "loss": 0.4084,
+      "step": 25029
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.513418329796662e-06,
+      "loss": 0.4874,
+      "step": 25030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.509235083734045e-06,
+      "loss": 0.4616,
+      "step": 25031
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5050529571179975e-06,
+      "loss": 0.3894,
+      "step": 25032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.500871949999133e-06,
+      "loss": 0.3708,
+      "step": 25033
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4966920624280675e-06,
+      "loss": 0.4244,
+      "step": 25034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4925132944553615e-06,
+      "loss": 0.4382,
+      "step": 25035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.488335646131628e-06,
+      "loss": 0.3596,
+      "step": 25036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.484159117507417e-06,
+      "loss": 0.5097,
+      "step": 25037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.479983708633287e-06,
+      "loss": 0.4488,
+      "step": 25038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.475809419559754e-06,
+      "loss": 0.4564,
+      "step": 25039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.471636250337332e-06,
+      "loss": 0.3757,
+      "step": 25040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.467464201016583e-06,
+      "loss": 0.3953,
+      "step": 25041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.463293271647953e-06,
+      "loss": 0.4352,
+      "step": 25042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.459123462281947e-06,
+      "loss": 0.3957,
+      "step": 25043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.454954772969036e-06,
+      "loss": 0.4635,
+      "step": 25044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4507872037596345e-06,
+      "loss": 0.3901,
+      "step": 25045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4466207547042474e-06,
+      "loss": 0.3379,
+      "step": 25046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.442455425853256e-06,
+      "loss": 0.3921,
+      "step": 25047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4382912172570984e-06,
+      "loss": 0.3189,
+      "step": 25048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4341281289661665e-06,
+      "loss": 0.4601,
+      "step": 25049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.429966161030832e-06,
+      "loss": 0.3728,
+      "step": 25050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.425805313501511e-06,
+      "loss": 0.4605,
+      "step": 25051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.421645586428516e-06,
+      "loss": 0.4599,
+      "step": 25052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.417486979862231e-06,
+      "loss": 0.381,
+      "step": 25053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4133294938529715e-06,
+      "loss": 0.4824,
+      "step": 25054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.40917312845103e-06,
+      "loss": 0.457,
+      "step": 25055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.405017883706766e-06,
+      "loss": 0.4721,
+      "step": 25056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.400863759670451e-06,
+      "loss": 0.3605,
+      "step": 25057
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.396710756392344e-06,
+      "loss": 0.3617,
+      "step": 25058
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3925588739227394e-06,
+      "loss": 0.4763,
+      "step": 25059
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.388408112311873e-06,
+      "loss": 0.5159,
+      "step": 25060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3842584716099615e-06,
+      "loss": 0.3759,
+      "step": 25061
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.380109951867264e-06,
+      "loss": 0.3546,
+      "step": 25062
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.375962553133997e-06,
+      "loss": 0.4772,
+      "step": 25063
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.37181627546033e-06,
+      "loss": 0.4066,
+      "step": 25064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3676711188964576e-06,
+      "loss": 0.3541,
+      "step": 25065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.363527083492539e-06,
+      "loss": 0.3675,
+      "step": 25066
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.359384169298744e-06,
+      "loss": 0.355,
+      "step": 25067
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3552423763652235e-06,
+      "loss": 0.433,
+      "step": 25068
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3511017047421026e-06,
+      "loss": 0.43,
+      "step": 25069
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3469621544794866e-06,
+      "loss": 0.3544,
+      "step": 25070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.342823725627468e-06,
+      "loss": 0.386,
+      "step": 25071
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.338686418236162e-06,
+      "loss": 0.433,
+      "step": 25072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.33455023235563e-06,
+      "loss": 0.4602,
+      "step": 25073
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.330415168035942e-06,
+      "loss": 0.4624,
+      "step": 25074
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3262812253271364e-06,
+      "loss": 0.4563,
+      "step": 25075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.322148404279238e-06,
+      "loss": 0.368,
+      "step": 25076
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.318016704942298e-06,
+      "loss": 0.4483,
+      "step": 25077
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.313886127366298e-06,
+      "loss": 0.4314,
+      "step": 25078
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.3097566716012536e-06,
+      "loss": 0.4887,
+      "step": 25079
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.305628337697123e-06,
+      "loss": 0.3947,
+      "step": 25080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.301501125703869e-06,
+      "loss": 0.3709,
+      "step": 25081
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.297375035671472e-06,
+      "loss": 0.4127,
+      "step": 25082
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.293250067649859e-06,
+      "loss": 0.4721,
+      "step": 25083
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2891262216889465e-06,
+      "loss": 0.4409,
+      "step": 25084
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.285003497838661e-06,
+      "loss": 0.3837,
+      "step": 25085
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.280881896148872e-06,
+      "loss": 0.4283,
+      "step": 25086
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2767614166695086e-06,
+      "loss": 0.4381,
+      "step": 25087
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.272642059450419e-06,
+      "loss": 0.4074,
+      "step": 25088
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.268523824541451e-06,
+      "loss": 0.4521,
+      "step": 25089
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2644067119924776e-06,
+      "loss": 0.3529,
+      "step": 25090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.260290721853302e-06,
+      "loss": 0.4431,
+      "step": 25091
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.256175854173742e-06,
+      "loss": 0.3985,
+      "step": 25092
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.252062109003621e-06,
+      "loss": 0.3778,
+      "step": 25093
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.247949486392724e-06,
+      "loss": 0.4749,
+      "step": 25094
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.243837986390822e-06,
+      "loss": 0.3701,
+      "step": 25095
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.239727609047675e-06,
+      "loss": 0.4015,
+      "step": 25096
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.235618354413021e-06,
+      "loss": 0.4395,
+      "step": 25097
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.23151022253663e-06,
+      "loss": 0.4144,
+      "step": 25098
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.227403213468198e-06,
+      "loss": 0.459,
+      "step": 25099
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.22329732725745e-06,
+      "loss": 0.3879,
+      "step": 25100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.219192563954058e-06,
+      "loss": 0.376,
+      "step": 25101
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.215088923607705e-06,
+      "loss": 0.4129,
+      "step": 25102
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2109864062680945e-06,
+      "loss": 0.3859,
+      "step": 25103
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.206885011984843e-06,
+      "loss": 0.4144,
+      "step": 25104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.202784740807611e-06,
+      "loss": 0.4228,
+      "step": 25105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.198685592786014e-06,
+      "loss": 0.4161,
+      "step": 25106
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.194587567969657e-06,
+      "loss": 0.3936,
+      "step": 25107
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.190490666408167e-06,
+      "loss": 0.3716,
+      "step": 25108
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.186394888151126e-06,
+      "loss": 0.5258,
+      "step": 25109
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.182300233248085e-06,
+      "loss": 0.4374,
+      "step": 25110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.1782067017486134e-06,
+      "loss": 0.4026,
+      "step": 25111
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.17411429370225e-06,
+      "loss": 0.3755,
+      "step": 25112
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.170023009158544e-06,
+      "loss": 0.42,
+      "step": 25113
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.165932848167001e-06,
+      "loss": 0.4222,
+      "step": 25114
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.161843810777136e-06,
+      "loss": 0.5252,
+      "step": 25115
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.1577558970384315e-06,
+      "loss": 0.3437,
+      "step": 25116
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.1536691070003495e-06,
+      "loss": 0.3856,
+      "step": 25117
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.149583440712382e-06,
+      "loss": 0.4934,
+      "step": 25118
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.145498898223968e-06,
+      "loss": 0.4332,
+      "step": 25119
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.141415479584535e-06,
+      "loss": 0.4083,
+      "step": 25120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.137333184843531e-06,
+      "loss": 0.5106,
+      "step": 25121
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.133252014050318e-06,
+      "loss": 0.4699,
+      "step": 25122
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.1291719672543444e-06,
+      "loss": 0.3884,
+      "step": 25123
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.125093044504971e-06,
+      "loss": 0.3901,
+      "step": 25124
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.12101524585157e-06,
+      "loss": 0.4287,
+      "step": 25125
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.1169385713434785e-06,
+      "loss": 0.3912,
+      "step": 25126
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.112863021030069e-06,
+      "loss": 0.4359,
+      "step": 25127
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.108788594960625e-06,
+      "loss": 0.3117,
+      "step": 25128
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.104715293184505e-06,
+      "loss": 0.358,
+      "step": 25129
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.100643115750982e-06,
+      "loss": 0.3656,
+      "step": 25130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.096572062709373e-06,
+      "loss": 0.4076,
+      "step": 25131
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.092502134108914e-06,
+      "loss": 0.4654,
+      "step": 25132
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.088433329998867e-06,
+      "loss": 0.3952,
+      "step": 25133
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.084365650428515e-06,
+      "loss": 0.3774,
+      "step": 25134
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.080299095447074e-06,
+      "loss": 0.431,
+      "step": 25135
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.0762336651037486e-06,
+      "loss": 0.4051,
+      "step": 25136
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.072169359447744e-06,
+      "loss": 0.3872,
+      "step": 25137
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.068106178528256e-06,
+      "loss": 0.421,
+      "step": 25138
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.064044122394486e-06,
+      "loss": 0.4687,
+      "step": 25139
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.059983191095576e-06,
+      "loss": 0.4571,
+      "step": 25140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.055923384680674e-06,
+      "loss": 0.3682,
+      "step": 25141
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.05186470319893e-06,
+      "loss": 0.3986,
+      "step": 25142
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.047807146699459e-06,
+      "loss": 0.4397,
+      "step": 25143
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.043750715231367e-06,
+      "loss": 0.3646,
+      "step": 25144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.039695408843772e-06,
+      "loss": 0.5304,
+      "step": 25145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.035641227585732e-06,
+      "loss": 0.392,
+      "step": 25146
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.031588171506309e-06,
+      "loss": 0.3758,
+      "step": 25147
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.0275362406545864e-06,
+      "loss": 0.3705,
+      "step": 25148
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.023485435079602e-06,
+      "loss": 0.3768,
+      "step": 25149
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.019435754830373e-06,
+      "loss": 0.445,
+      "step": 25150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.0153871999559136e-06,
+      "loss": 0.3442,
+      "step": 25151
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.011339770505221e-06,
+      "loss": 0.377,
+      "step": 25152
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.007293466527298e-06,
+      "loss": 0.4491,
+      "step": 25153
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.003248288071118e-06,
+      "loss": 0.3857,
+      "step": 25154
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.99920423518563e-06,
+      "loss": 0.4978,
+      "step": 25155
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.9951613079197845e-06,
+      "loss": 0.3901,
+      "step": 25156
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.991119506322497e-06,
+      "loss": 0.3619,
+      "step": 25157
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.987078830442717e-06,
+      "loss": 0.419,
+      "step": 25158
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.98303928032934e-06,
+      "loss": 0.3676,
+      "step": 25159
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.979000856031259e-06,
+      "loss": 0.4611,
+      "step": 25160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.974963557597347e-06,
+      "loss": 0.397,
+      "step": 25161
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.970927385076464e-06,
+      "loss": 0.3576,
+      "step": 25162
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.96689233851745e-06,
+      "loss": 0.3844,
+      "step": 25163
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.962858417969187e-06,
+      "loss": 0.48,
+      "step": 25164
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.95882562348047e-06,
+      "loss": 0.4288,
+      "step": 25165
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.954793955100103e-06,
+      "loss": 0.4946,
+      "step": 25166
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.950763412876893e-06,
+      "loss": 0.348,
+      "step": 25167
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.946733996859611e-06,
+      "loss": 0.3984,
+      "step": 25168
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.942705707097052e-06,
+      "loss": 0.399,
+      "step": 25169
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.9386785436379445e-06,
+      "loss": 0.3968,
+      "step": 25170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.9346525065310585e-06,
+      "loss": 0.4188,
+      "step": 25171
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.93062759582509e-06,
+      "loss": 0.4432,
+      "step": 25172
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.926603811568766e-06,
+      "loss": 0.4333,
+      "step": 25173
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.922581153810792e-06,
+      "loss": 0.4541,
+      "step": 25174
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.918559622599874e-06,
+      "loss": 0.3144,
+      "step": 25175
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.91453921798465e-06,
+      "loss": 0.4188,
+      "step": 25176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.910519940013804e-06,
+      "loss": 0.4527,
+      "step": 25177
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.906501788735964e-06,
+      "loss": 0.3992,
+      "step": 25178
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.90248476419979e-06,
+      "loss": 0.4677,
+      "step": 25179
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.898468866453878e-06,
+      "loss": 0.374,
+      "step": 25180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.894454095546854e-06,
+      "loss": 0.3444,
+      "step": 25181
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.890440451527291e-06,
+      "loss": 0.4036,
+      "step": 25182
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.886427934443762e-06,
+      "loss": 0.3996,
+      "step": 25183
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.882416544344849e-06,
+      "loss": 0.4041,
+      "step": 25184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.878406281279115e-06,
+      "loss": 0.4331,
+      "step": 25185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.874397145295075e-06,
+      "loss": 0.3769,
+      "step": 25186
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.8703891364412575e-06,
+      "loss": 0.3398,
+      "step": 25187
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.866382254766157e-06,
+      "loss": 0.3995,
+      "step": 25188
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.862376500318302e-06,
+      "loss": 0.4205,
+      "step": 25189
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.858371873146163e-06,
+      "loss": 0.4238,
+      "step": 25190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.854368373298214e-06,
+      "loss": 0.4582,
+      "step": 25191
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.850366000822894e-06,
+      "loss": 0.435,
+      "step": 25192
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.846364755768664e-06,
+      "loss": 0.4333,
+      "step": 25193
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.842364638183918e-06,
+      "loss": 0.4648,
+      "step": 25194
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.838365648117118e-06,
+      "loss": 0.4442,
+      "step": 25195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.834367785616646e-06,
+      "loss": 0.3347,
+      "step": 25196
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.830371050730877e-06,
+      "loss": 0.4296,
+      "step": 25197
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.826375443508204e-06,
+      "loss": 0.4663,
+      "step": 25198
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.822380963996955e-06,
+      "loss": 0.374,
+      "step": 25199
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.8183876122455245e-06,
+      "loss": 0.3898,
+      "step": 25200
+    },
+    {
+      "epoch": 0.88,
+      "eval_loss": 0.40058866143226624,
+      "eval_runtime": 845.477,
+      "eval_samples_per_second": 28.457,
+      "eval_steps_per_second": 14.229,
+      "step": 25200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.814395388302208e-06,
+      "loss": 0.4653,
+      "step": 25201
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.8104042922153445e-06,
+      "loss": 0.3972,
+      "step": 25202
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.806414324033228e-06,
+      "loss": 0.4011,
+      "step": 25203
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.802425483804142e-06,
+      "loss": 0.4361,
+      "step": 25204
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.798437771576394e-06,
+      "loss": 0.3901,
+      "step": 25205
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.794451187398232e-06,
+      "loss": 0.5281,
+      "step": 25206
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.7904657313178965e-06,
+      "loss": 0.3537,
+      "step": 25207
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.786481403383649e-06,
+      "loss": 0.385,
+      "step": 25208
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.782498203643672e-06,
+      "loss": 0.4244,
+      "step": 25209
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.7785161321462284e-06,
+      "loss": 0.4022,
+      "step": 25210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.774535188939479e-06,
+      "loss": 0.3818,
+      "step": 25211
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.7705553740716185e-06,
+      "loss": 0.4967,
+      "step": 25212
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.76657668759082e-06,
+      "loss": 0.3576,
+      "step": 25213
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.7625991295452105e-06,
+      "loss": 0.3856,
+      "step": 25214
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.758622699982975e-06,
+      "loss": 0.3915,
+      "step": 25215
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.754647398952218e-06,
+      "loss": 0.415,
+      "step": 25216
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.750673226501059e-06,
+      "loss": 0.4128,
+      "step": 25217
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.74670018267759e-06,
+      "loss": 0.4689,
+      "step": 25218
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.742728267529885e-06,
+      "loss": 0.3626,
+      "step": 25219
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.738757481106061e-06,
+      "loss": 0.4386,
+      "step": 25220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.734787823454148e-06,
+      "loss": 0.3863,
+      "step": 25221
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.730819294622204e-06,
+      "loss": 0.4098,
+      "step": 25222
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.726851894658248e-06,
+      "loss": 0.4369,
+      "step": 25223
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.722885623610287e-06,
+      "loss": 0.3665,
+      "step": 25224
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.718920481526369e-06,
+      "loss": 0.4559,
+      "step": 25225
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.714956468454459e-06,
+      "loss": 0.4169,
+      "step": 25226
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.710993584442538e-06,
+      "loss": 0.4845,
+      "step": 25227
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.707031829538557e-06,
+      "loss": 0.4123,
+      "step": 25228
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.703071203790489e-06,
+      "loss": 0.4051,
+      "step": 25229
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.699111707246242e-06,
+      "loss": 0.4398,
+      "step": 25230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.6951533399537745e-06,
+      "loss": 0.436,
+      "step": 25231
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.691196101960973e-06,
+      "loss": 0.4068,
+      "step": 25232
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.6872399933157415e-06,
+      "loss": 0.4378,
+      "step": 25233
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.683285014065965e-06,
+      "loss": 0.4548,
+      "step": 25234
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.679331164259483e-06,
+      "loss": 0.4847,
+      "step": 25235
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.675378443944202e-06,
+      "loss": 0.378,
+      "step": 25236
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.671426853167917e-06,
+      "loss": 0.3669,
+      "step": 25237
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.667476391978489e-06,
+      "loss": 0.3747,
+      "step": 25238
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.663527060423702e-06,
+      "loss": 0.3854,
+      "step": 25239
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.659578858551363e-06,
+      "loss": 0.4273,
+      "step": 25240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.655631786409278e-06,
+      "loss": 0.4034,
+      "step": 25241
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.651685844045208e-06,
+      "loss": 0.3641,
+      "step": 25242
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.6477410315069155e-06,
+      "loss": 0.4322,
+      "step": 25243
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.64379734884214e-06,
+      "loss": 0.4452,
+      "step": 25244
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.639854796098599e-06,
+      "loss": 0.4427,
+      "step": 25245
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.635913373324032e-06,
+      "loss": 0.3467,
+      "step": 25246
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.631973080566145e-06,
+      "loss": 0.5035,
+      "step": 25247
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.628033917872623e-06,
+      "loss": 0.3406,
+      "step": 25248
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.624095885291148e-06,
+      "loss": 0.406,
+      "step": 25249
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.620158982869351e-06,
+      "loss": 0.4472,
+      "step": 25250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.616223210654926e-06,
+      "loss": 0.3568,
+      "step": 25251
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.6122885686954905e-06,
+      "loss": 0.4213,
+      "step": 25252
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.608355057038662e-06,
+      "loss": 0.3943,
+      "step": 25253
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.604422675732058e-06,
+      "loss": 0.4637,
+      "step": 25254
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.600491424823252e-06,
+      "loss": 0.3836,
+      "step": 25255
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.596561304359861e-06,
+      "loss": 0.4302,
+      "step": 25256
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.592632314389436e-06,
+      "loss": 0.3627,
+      "step": 25257
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.588704454959527e-06,
+      "loss": 0.4246,
+      "step": 25258
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.584777726117664e-06,
+      "loss": 0.4038,
+      "step": 25259
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.580852127911397e-06,
+      "loss": 0.3723,
+      "step": 25260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.57692766038821e-06,
+      "loss": 0.5283,
+      "step": 25261
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.573004323595633e-06,
+      "loss": 0.3975,
+      "step": 25262
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.569082117581149e-06,
+      "loss": 0.4871,
+      "step": 25263
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.5651610423922095e-06,
+      "loss": 0.4526,
+      "step": 25264
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.561241098076276e-06,
+      "loss": 0.4154,
+      "step": 25265
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.557322284680789e-06,
+      "loss": 0.4495,
+      "step": 25266
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.5534046022532105e-06,
+      "loss": 0.411,
+      "step": 25267
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.549488050840924e-06,
+      "loss": 0.3449,
+      "step": 25268
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.545572630491348e-06,
+      "loss": 0.4649,
+      "step": 25269
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.541658341251877e-06,
+      "loss": 0.3874,
+      "step": 25270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.537745183169852e-06,
+      "loss": 0.3774,
+      "step": 25271
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.533833156292679e-06,
+      "loss": 0.4119,
+      "step": 25272
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.529922260667698e-06,
+      "loss": 0.553,
+      "step": 25273
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.526012496342226e-06,
+      "loss": 0.4553,
+      "step": 25274
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.522103863363604e-06,
+      "loss": 0.4546,
+      "step": 25275
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.5181963617791275e-06,
+      "loss": 0.5348,
+      "step": 25276
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.51428999163608e-06,
+      "loss": 0.4192,
+      "step": 25277
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.510384752981769e-06,
+      "loss": 0.4311,
+      "step": 25278
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.5064806458634335e-06,
+      "loss": 0.3849,
+      "step": 25279
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.502577670328325e-06,
+      "loss": 0.3262,
+      "step": 25280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.498675826423717e-06,
+      "loss": 0.3927,
+      "step": 25281
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.494775114196816e-06,
+      "loss": 0.4938,
+      "step": 25282
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.490875533694818e-06,
+      "loss": 0.3779,
+      "step": 25283
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.48697708496494e-06,
+      "loss": 0.4275,
+      "step": 25284
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.483079768054346e-06,
+      "loss": 0.3958,
+      "step": 25285
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.47918358301024e-06,
+      "loss": 0.5076,
+      "step": 25286
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.475288529879753e-06,
+      "loss": 0.3777,
+      "step": 25287
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.471394608710035e-06,
+      "loss": 0.3847,
+      "step": 25288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.467501819548216e-06,
+      "loss": 0.5072,
+      "step": 25289
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.463610162441403e-06,
+      "loss": 0.3728,
+      "step": 25290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.459719637436712e-06,
+      "loss": 0.4455,
+      "step": 25291
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.455830244581229e-06,
+      "loss": 0.3998,
+      "step": 25292
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.451941983922027e-06,
+      "loss": 0.4661,
+      "step": 25293
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4480548555061585e-06,
+      "loss": 0.3789,
+      "step": 25294
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4441688593806836e-06,
+      "loss": 0.3942,
+      "step": 25295
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.440283995592611e-06,
+      "loss": 0.3682,
+      "step": 25296
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.436400264188991e-06,
+      "loss": 0.3429,
+      "step": 25297
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.43251766521682e-06,
+      "loss": 0.4358,
+      "step": 25298
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4286361987230925e-06,
+      "loss": 0.3639,
+      "step": 25299
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.424755864754772e-06,
+      "loss": 0.3787,
+      "step": 25300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.420876663358832e-06,
+      "loss": 0.3763,
+      "step": 25301
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4169985945822355e-06,
+      "loss": 0.3207,
+      "step": 25302
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.413121658471899e-06,
+      "loss": 0.4506,
+      "step": 25303
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.409245855074775e-06,
+      "loss": 0.35,
+      "step": 25304
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.405371184437748e-06,
+      "loss": 0.3977,
+      "step": 25305
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.401497646607702e-06,
+      "loss": 0.4463,
+      "step": 25306
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.397625241631566e-06,
+      "loss": 0.4282,
+      "step": 25307
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.393753969556182e-06,
+      "loss": 0.4052,
+      "step": 25308
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3898838304284e-06,
+      "loss": 0.3742,
+      "step": 25309
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.386014824295072e-06,
+      "loss": 0.4325,
+      "step": 25310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.382146951203005e-06,
+      "loss": 0.4666,
+      "step": 25311
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.37828021119905e-06,
+      "loss": 0.4574,
+      "step": 25312
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3744146043299815e-06,
+      "loss": 0.4633,
+      "step": 25313
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.370550130642594e-06,
+      "loss": 0.354,
+      "step": 25314
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.366686790183662e-06,
+      "loss": 0.3831,
+      "step": 25315
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3628245829999265e-06,
+      "loss": 0.4092,
+      "step": 25316
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3589635091381605e-06,
+      "loss": 0.4352,
+      "step": 25317
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.355103568645093e-06,
+      "loss": 0.4218,
+      "step": 25318
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.351244761567421e-06,
+      "loss": 0.3864,
+      "step": 25319
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.347387087951862e-06,
+      "loss": 0.4055,
+      "step": 25320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3435305478451e-06,
+      "loss": 0.4143,
+      "step": 25321
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.339675141293821e-06,
+      "loss": 0.4447,
+      "step": 25322
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3358208683446885e-06,
+      "loss": 0.4245,
+      "step": 25323
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.331967729044352e-06,
+      "loss": 0.3586,
+      "step": 25324
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.328115723439443e-06,
+      "loss": 0.4244,
+      "step": 25325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.324264851576567e-06,
+      "loss": 0.4124,
+      "step": 25326
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.320415113502365e-06,
+      "loss": 0.3897,
+      "step": 25327
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3165665092634215e-06,
+      "loss": 0.3164,
+      "step": 25328
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.312719038906312e-06,
+      "loss": 0.4178,
+      "step": 25329
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.308872702477597e-06,
+      "loss": 0.3982,
+      "step": 25330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3050275000238414e-06,
+      "loss": 0.3321,
+      "step": 25331
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3011834315915615e-06,
+      "loss": 0.4369,
+      "step": 25332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.297340497227322e-06,
+      "loss": 0.4361,
+      "step": 25333
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.293498696977607e-06,
+      "loss": 0.3891,
+      "step": 25334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.2896580308889345e-06,
+      "loss": 0.3741,
+      "step": 25335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.285818499007779e-06,
+      "loss": 0.3696,
+      "step": 25336
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.281980101380591e-06,
+      "loss": 0.438,
+      "step": 25337
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.2781428380538576e-06,
+      "loss": 0.3403,
+      "step": 25338
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.274306709074018e-06,
+      "loss": 0.4804,
+      "step": 25339
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.270471714487502e-06,
+      "loss": 0.4571,
+      "step": 25340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.266637854340718e-06,
+      "loss": 0.4498,
+      "step": 25341
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.262805128680049e-06,
+      "loss": 0.4175,
+      "step": 25342
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.258973537551937e-06,
+      "loss": 0.3645,
+      "step": 25343
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.2551430810027126e-06,
+      "loss": 0.4437,
+      "step": 25344
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.251313759078758e-06,
+      "loss": 0.4682,
+      "step": 25345
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.247485571826406e-06,
+      "loss": 0.4646,
+      "step": 25346
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.243658519291984e-06,
+      "loss": 0.4347,
+      "step": 25347
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.239832601521844e-06,
+      "loss": 0.4177,
+      "step": 25348
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.236007818562284e-06,
+      "loss": 0.3507,
+      "step": 25349
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.2321841704595766e-06,
+      "loss": 0.3961,
+      "step": 25350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.228361657260007e-06,
+      "loss": 0.3614,
+      "step": 25351
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.224540279009839e-06,
+      "loss": 0.4514,
+      "step": 25352
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.220720035755334e-06,
+      "loss": 0.3164,
+      "step": 25353
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.216900927542735e-06,
+      "loss": 0.4225,
+      "step": 25354
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.2130829544182476e-06,
+      "loss": 0.3767,
+      "step": 25355
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.209266116428103e-06,
+      "loss": 0.4042,
+      "step": 25356
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.205450413618463e-06,
+      "loss": 0.4243,
+      "step": 25357
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.201635846035536e-06,
+      "loss": 0.4594,
+      "step": 25358
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.197822413725507e-06,
+      "loss": 0.3765,
+      "step": 25359
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.194010116734495e-06,
+      "loss": 0.4064,
+      "step": 25360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.190198955108673e-06,
+      "loss": 0.3569,
+      "step": 25361
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.186388928894149e-06,
+      "loss": 0.3852,
+      "step": 25362
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.18258003813702e-06,
+      "loss": 0.4409,
+      "step": 25363
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.1787722828834275e-06,
+      "loss": 0.4282,
+      "step": 25364
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.174965663179433e-06,
+      "loss": 0.4221,
+      "step": 25365
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.171160179071123e-06,
+      "loss": 0.419,
+      "step": 25366
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.167355830604549e-06,
+      "loss": 0.4584,
+      "step": 25367
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.163552617825741e-06,
+      "loss": 0.372,
+      "step": 25368
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.159750540780751e-06,
+      "loss": 0.3807,
+      "step": 25369
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.155949599515598e-06,
+      "loss": 0.3763,
+      "step": 25370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.15214979407629e-06,
+      "loss": 0.3903,
+      "step": 25371
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.1483511245087886e-06,
+      "loss": 0.426,
+      "step": 25372
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.144553590859081e-06,
+      "loss": 0.4305,
+      "step": 25373
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.140757193173152e-06,
+      "loss": 0.4412,
+      "step": 25374
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.136961931496943e-06,
+      "loss": 0.4263,
+      "step": 25375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.133167805876372e-06,
+      "loss": 0.5039,
+      "step": 25376
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.129374816357369e-06,
+      "loss": 0.3571,
+      "step": 25377
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.125582962985843e-06,
+      "loss": 0.4595,
+      "step": 25378
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.121792245807689e-06,
+      "loss": 0.4004,
+      "step": 25379
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.118002664868794e-06,
+      "loss": 0.4231,
+      "step": 25380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.1142142202150195e-06,
+      "loss": 0.3818,
+      "step": 25381
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.1104269118922085e-06,
+      "loss": 0.3887,
+      "step": 25382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.106640739946201e-06,
+      "loss": 0.3998,
+      "step": 25383
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.102855704422839e-06,
+      "loss": 0.3654,
+      "step": 25384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.099071805367929e-06,
+      "loss": 0.4491,
+      "step": 25385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.095289042827268e-06,
+      "loss": 0.376,
+      "step": 25386
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.091507416846631e-06,
+      "loss": 0.4245,
+      "step": 25387
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.08772692747177e-06,
+      "loss": 0.4646,
+      "step": 25388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0839475747484945e-06,
+      "loss": 0.3901,
+      "step": 25389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.08016935872251e-06,
+      "loss": 0.3843,
+      "step": 25390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.076392279439558e-06,
+      "loss": 0.4417,
+      "step": 25391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.072616336945347e-06,
+      "loss": 0.4979,
+      "step": 25392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.068841531285574e-06,
+      "loss": 0.3417,
+      "step": 25393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.065067862505925e-06,
+      "loss": 0.4594,
+      "step": 25394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.061295330652095e-06,
+      "loss": 0.4393,
+      "step": 25395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.057523935769738e-06,
+      "loss": 0.4895,
+      "step": 25396
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.053753677904484e-06,
+      "loss": 0.4081,
+      "step": 25397
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.049984557101984e-06,
+      "loss": 0.5156,
+      "step": 25398
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.046216573407826e-06,
+      "loss": 0.4317,
+      "step": 25399
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.042449726867661e-06,
+      "loss": 0.4235,
+      "step": 25400
+    },
+    {
+      "epoch": 0.89,
+      "eval_loss": 0.3999830186367035,
+      "eval_runtime": 854.0011,
+      "eval_samples_per_second": 28.173,
+      "eval_steps_per_second": 14.087,
+      "step": 25400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.038684017527052e-06,
+      "loss": 0.3954,
+      "step": 25401
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.034919445431575e-06,
+      "loss": 0.3378,
+      "step": 25402
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.031156010626815e-06,
+      "loss": 0.4704,
+      "step": 25403
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0273937131582804e-06,
+      "loss": 0.3616,
+      "step": 25404
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0236325530715565e-06,
+      "loss": 0.4178,
+      "step": 25405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.019872530412141e-06,
+      "loss": 0.3775,
+      "step": 25406
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0161136452255515e-06,
+      "loss": 0.3443,
+      "step": 25407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0123558975572645e-06,
+      "loss": 0.3732,
+      "step": 25408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.008599287452776e-06,
+      "loss": 0.4008,
+      "step": 25409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.004843814957562e-06,
+      "loss": 0.4525,
+      "step": 25410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.001089480117073e-06,
+      "loss": 0.3706,
+      "step": 25411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.997336282976728e-06,
+      "loss": 0.4359,
+      "step": 25412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9935842235819605e-06,
+      "loss": 0.4753,
+      "step": 25413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.989833301978198e-06,
+      "loss": 0.4392,
+      "step": 25414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.986083518210839e-06,
+      "loss": 0.409,
+      "step": 25415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.982334872325246e-06,
+      "loss": 0.5522,
+      "step": 25416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.978587364366817e-06,
+      "loss": 0.3451,
+      "step": 25417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.97484099438087e-06,
+      "loss": 0.3717,
+      "step": 25418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.971095762412804e-06,
+      "loss": 0.3677,
+      "step": 25419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.967351668507914e-06,
+      "loss": 0.4029,
+      "step": 25420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9636087127115214e-06,
+      "loss": 0.3539,
+      "step": 25421
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9598668950689216e-06,
+      "loss": 0.4256,
+      "step": 25422
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.956126215625401e-06,
+      "loss": 0.3796,
+      "step": 25423
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.952386674426269e-06,
+      "loss": 0.416,
+      "step": 25424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.948648271516744e-06,
+      "loss": 0.4346,
+      "step": 25425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9449110069421e-06,
+      "loss": 0.429,
+      "step": 25426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.941174880747558e-06,
+      "loss": 0.3841,
+      "step": 25427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.937439892978325e-06,
+      "loss": 0.3639,
+      "step": 25428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.933706043679644e-06,
+      "loss": 0.4132,
+      "step": 25429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.929973332896677e-06,
+      "loss": 0.4297,
+      "step": 25430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.926241760674611e-06,
+      "loss": 0.4111,
+      "step": 25431
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.922511327058611e-06,
+      "loss": 0.4333,
+      "step": 25432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.918782032093828e-06,
+      "loss": 0.4438,
+      "step": 25433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.915053875825383e-06,
+      "loss": 0.4413,
+      "step": 25434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.911326858298427e-06,
+      "loss": 0.4057,
+      "step": 25435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.907600979558059e-06,
+      "loss": 0.459,
+      "step": 25436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9038762396493755e-06,
+      "loss": 0.4213,
+      "step": 25437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.90015263861744e-06,
+      "loss": 0.4477,
+      "step": 25438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.896430176507328e-06,
+      "loss": 0.3936,
+      "step": 25439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.892708853364115e-06,
+      "loss": 0.3861,
+      "step": 25440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.888988669232831e-06,
+      "loss": 0.4102,
+      "step": 25441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.885269624158496e-06,
+      "loss": 0.3772,
+      "step": 25442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.881551718186118e-06,
+      "loss": 0.3862,
+      "step": 25443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.877834951360695e-06,
+      "loss": 0.3834,
+      "step": 25444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.874119323727234e-06,
+      "loss": 0.4828,
+      "step": 25445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.87040483533069e-06,
+      "loss": 0.3932,
+      "step": 25446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.866691486216036e-06,
+      "loss": 0.4933,
+      "step": 25447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.862979276428193e-06,
+      "loss": 0.4191,
+      "step": 25448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8592682060120914e-06,
+      "loss": 0.3919,
+      "step": 25449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.855558275012662e-06,
+      "loss": 0.4166,
+      "step": 25450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.851849483474814e-06,
+      "loss": 0.4198,
+      "step": 25451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8481418314434215e-06,
+      "loss": 0.4802,
+      "step": 25452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.84443531896337e-06,
+      "loss": 0.4076,
+      "step": 25453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.840729946079493e-06,
+      "loss": 0.345,
+      "step": 25454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.837025712836675e-06,
+      "loss": 0.431,
+      "step": 25455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.833322619279735e-06,
+      "loss": 0.4098,
+      "step": 25456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.829620665453484e-06,
+      "loss": 0.3614,
+      "step": 25457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.825919851402739e-06,
+      "loss": 0.3892,
+      "step": 25458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.822220177172277e-06,
+      "loss": 0.3663,
+      "step": 25459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.818521642806895e-06,
+      "loss": 0.4322,
+      "step": 25460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.814824248351358e-06,
+      "loss": 0.3769,
+      "step": 25461
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.811127993850407e-06,
+      "loss": 0.4417,
+      "step": 25462
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8074328793487845e-06,
+      "loss": 0.3651,
+      "step": 25463
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.80373890489121e-06,
+      "loss": 0.3873,
+      "step": 25464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8000460705223694e-06,
+      "loss": 0.4112,
+      "step": 25465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.796354376287005e-06,
+      "loss": 0.4615,
+      "step": 25466
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.792663822229771e-06,
+      "loss": 0.4131,
+      "step": 25467
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.788974408395354e-06,
+      "loss": 0.4216,
+      "step": 25468
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.785286134828382e-06,
+      "loss": 0.4323,
+      "step": 25469
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.78159900157349e-06,
+      "loss": 0.4291,
+      "step": 25470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.777913008675339e-06,
+      "loss": 0.4745,
+      "step": 25471
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.77422815617853e-06,
+      "loss": 0.4611,
+      "step": 25472
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.770544444127657e-06,
+      "loss": 0.4304,
+      "step": 25473
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.766861872567297e-06,
+      "loss": 0.4539,
+      "step": 25474
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.763180441542026e-06,
+      "loss": 0.3359,
+      "step": 25475
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.759500151096419e-06,
+      "loss": 0.4312,
+      "step": 25476
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.755821001274997e-06,
+      "loss": 0.425,
+      "step": 25477
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.752142992122311e-06,
+      "loss": 0.3673,
+      "step": 25478
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.748466123682861e-06,
+      "loss": 0.4746,
+      "step": 25479
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7447903960011316e-06,
+      "loss": 0.3561,
+      "step": 25480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.741115809121655e-06,
+      "loss": 0.3975,
+      "step": 25481
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.737442363088885e-06,
+      "loss": 0.4429,
+      "step": 25482
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.733770057947274e-06,
+      "loss": 0.4167,
+      "step": 25483
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.730098893741287e-06,
+      "loss": 0.3545,
+      "step": 25484
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.726428870515332e-06,
+      "loss": 0.4494,
+      "step": 25485
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.722759988313853e-06,
+      "loss": 0.5121,
+      "step": 25486
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.719092247181246e-06,
+      "loss": 0.3954,
+      "step": 25487
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.715425647161909e-06,
+      "loss": 0.428,
+      "step": 25488
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.711760188300208e-06,
+      "loss": 0.4206,
+      "step": 25489
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.708095870640506e-06,
+      "loss": 0.3527,
+      "step": 25490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7044326942271795e-06,
+      "loss": 0.3622,
+      "step": 25491
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7007706591045375e-06,
+      "loss": 0.3732,
+      "step": 25492
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.69710976531691e-06,
+      "loss": 0.4074,
+      "step": 25493
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.693450012908619e-06,
+      "loss": 0.3806,
+      "step": 25494
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.689791401923938e-06,
+      "loss": 0.3978,
+      "step": 25495
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.686133932407156e-06,
+      "loss": 0.4398,
+      "step": 25496
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.682477604402547e-06,
+      "loss": 0.4733,
+      "step": 25497
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.678822417954377e-06,
+      "loss": 0.3635,
+      "step": 25498
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.675168373106854e-06,
+      "loss": 0.4194,
+      "step": 25499
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.671515469904231e-06,
+      "loss": 0.3764,
+      "step": 25500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.667863708390697e-06,
+      "loss": 0.4597,
+      "step": 25501
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.66421308861047e-06,
+      "loss": 0.4011,
+      "step": 25502
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.660563610607739e-06,
+      "loss": 0.3871,
+      "step": 25503
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.656915274426655e-06,
+      "loss": 0.4225,
+      "step": 25504
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.653268080111385e-06,
+      "loss": 0.351,
+      "step": 25505
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.649622027706058e-06,
+      "loss": 0.4866,
+      "step": 25506
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.64597711725483e-06,
+      "loss": 0.4817,
+      "step": 25507
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.6423333488018095e-06,
+      "loss": 0.5346,
+      "step": 25508
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.638690722391083e-06,
+      "loss": 0.3815,
+      "step": 25509
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.635049238066747e-06,
+      "loss": 0.364,
+      "step": 25510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.63140889587287e-06,
+      "loss": 0.4391,
+      "step": 25511
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.627769695853535e-06,
+      "loss": 0.3792,
+      "step": 25512
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.624131638052765e-06,
+      "loss": 0.431,
+      "step": 25513
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.620494722514602e-06,
+      "loss": 0.464,
+      "step": 25514
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.616858949283055e-06,
+      "loss": 0.3876,
+      "step": 25515
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.6132243184021325e-06,
+      "loss": 0.3753,
+      "step": 25516
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.609590829915834e-06,
+      "loss": 0.3656,
+      "step": 25517
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.605958483868135e-06,
+      "loss": 0.4263,
+      "step": 25518
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.602327280302999e-06,
+      "loss": 0.4431,
+      "step": 25519
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5986972192643705e-06,
+      "loss": 0.5063,
+      "step": 25520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.59506830079618e-06,
+      "loss": 0.3685,
+      "step": 25521
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.591440524942359e-06,
+      "loss": 0.4089,
+      "step": 25522
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5878138917468296e-06,
+      "loss": 0.4331,
+      "step": 25523
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.584188401253454e-06,
+      "loss": 0.4188,
+      "step": 25524
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.580564053506143e-06,
+      "loss": 0.4052,
+      "step": 25525
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.576940848548728e-06,
+      "loss": 0.4069,
+      "step": 25526
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.573318786425097e-06,
+      "loss": 0.3851,
+      "step": 25527
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.56969786717908e-06,
+      "loss": 0.39,
+      "step": 25528
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5660780908544876e-06,
+      "loss": 0.3434,
+      "step": 25529
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.56245945749515e-06,
+      "loss": 0.455,
+      "step": 25530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5588419671448565e-06,
+      "loss": 0.3893,
+      "step": 25531
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5552256198473705e-06,
+      "loss": 0.4331,
+      "step": 25532
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.551610415646502e-06,
+      "loss": 0.3979,
+      "step": 25533
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.547996354585983e-06,
+      "loss": 0.4142,
+      "step": 25534
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.544383436709555e-06,
+      "loss": 0.4611,
+      "step": 25535
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.540771662060951e-06,
+      "loss": 0.4317,
+      "step": 25536
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5371610306838796e-06,
+      "loss": 0.3986,
+      "step": 25537
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5335515426220504e-06,
+      "loss": 0.3615,
+      "step": 25538
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.529943197919152e-06,
+      "loss": 0.4058,
+      "step": 25539
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.526335996618859e-06,
+      "loss": 0.4473,
+      "step": 25540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5227299387648145e-06,
+      "loss": 0.3631,
+      "step": 25541
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.519125024400662e-06,
+      "loss": 0.4461,
+      "step": 25542
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.515521253570055e-06,
+      "loss": 0.4183,
+      "step": 25543
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.511918626316603e-06,
+      "loss": 0.376,
+      "step": 25544
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.508317142683905e-06,
+      "loss": 0.3354,
+      "step": 25545
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.504716802715549e-06,
+      "loss": 0.3884,
+      "step": 25546
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5011176064551196e-06,
+      "loss": 0.3813,
+      "step": 25547
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.4975195539461735e-06,
+      "loss": 0.3716,
+      "step": 25548
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.493922645232263e-06,
+      "loss": 0.4536,
+      "step": 25549
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.490326880356922e-06,
+      "loss": 0.4105,
+      "step": 25550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.486732259363647e-06,
+      "loss": 0.407,
+      "step": 25551
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.4831387822959825e-06,
+      "loss": 0.3959,
+      "step": 25552
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.479546449197415e-06,
+      "loss": 0.4355,
+      "step": 25553
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.475955260111409e-06,
+      "loss": 0.4731,
+      "step": 25554
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.472365215081432e-06,
+      "loss": 0.4005,
+      "step": 25555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.4687763141509255e-06,
+      "loss": 0.4251,
+      "step": 25556
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.465188557363354e-06,
+      "loss": 0.3574,
+      "step": 25557
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.4616019447621406e-06,
+      "loss": 0.4505,
+      "step": 25558
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.45801647639067e-06,
+      "loss": 0.4452,
+      "step": 25559
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.454432152292355e-06,
+      "loss": 0.4431,
+      "step": 25560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.450848972510558e-06,
+      "loss": 0.4075,
+      "step": 25561
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.4472669370886794e-06,
+      "loss": 0.4418,
+      "step": 25562
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.443686046070051e-06,
+      "loss": 0.3694,
+      "step": 25563
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.4401062994980265e-06,
+      "loss": 0.4003,
+      "step": 25564
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.4365276974159274e-06,
+      "loss": 0.3477,
+      "step": 25565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.432950239867063e-06,
+      "loss": 0.4021,
+      "step": 25566
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.429373926894721e-06,
+      "loss": 0.3999,
+      "step": 25567
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.425798758542222e-06,
+      "loss": 0.4348,
+      "step": 25568
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.42222473485281e-06,
+      "loss": 0.4157,
+      "step": 25569
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.418651855869749e-06,
+      "loss": 0.4263,
+      "step": 25570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.415080121636274e-06,
+      "loss": 0.4268,
+      "step": 25571
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.411509532195624e-06,
+      "loss": 0.3712,
+      "step": 25572
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.407940087591024e-06,
+      "loss": 0.4321,
+      "step": 25573
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.40437178786567e-06,
+      "loss": 0.4341,
+      "step": 25574
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.400804633062739e-06,
+      "loss": 0.4097,
+      "step": 25575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.39723862322542e-06,
+      "loss": 0.3939,
+      "step": 25576
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.393673758396845e-06,
+      "loss": 0.4804,
+      "step": 25577
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.390110038620211e-06,
+      "loss": 0.3497,
+      "step": 25578
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.386547463938607e-06,
+      "loss": 0.3565,
+      "step": 25579
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.382986034395176e-06,
+      "loss": 0.3833,
+      "step": 25580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.379425750033018e-06,
+      "loss": 0.3659,
+      "step": 25581
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.375866610895208e-06,
+      "loss": 0.4691,
+      "step": 25582
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.372308617024846e-06,
+      "loss": 0.3476,
+      "step": 25583
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.368751768464986e-06,
+      "loss": 0.4484,
+      "step": 25584
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.365196065258682e-06,
+      "loss": 0.4593,
+      "step": 25585
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.361641507448967e-06,
+      "loss": 0.2884,
+      "step": 25586
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.358088095078839e-06,
+      "loss": 0.4039,
+      "step": 25587
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.354535828191354e-06,
+      "loss": 0.392,
+      "step": 25588
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.350984706829476e-06,
+      "loss": 0.434,
+      "step": 25589
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.347434731036182e-06,
+      "loss": 0.3884,
+      "step": 25590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.343885900854462e-06,
+      "loss": 0.4409,
+      "step": 25591
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.340338216327234e-06,
+      "loss": 0.3928,
+      "step": 25592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3367916774974655e-06,
+      "loss": 0.3854,
+      "step": 25593
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.333246284408088e-06,
+      "loss": 0.3688,
+      "step": 25594
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.32970203710198e-06,
+      "loss": 0.442,
+      "step": 25595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.326158935622072e-06,
+      "loss": 0.4536,
+      "step": 25596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.322616980011219e-06,
+      "loss": 0.3977,
+      "step": 25597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.319076170312287e-06,
+      "loss": 0.404,
+      "step": 25598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.315536506568175e-06,
+      "loss": 0.3778,
+      "step": 25599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.311997988821693e-06,
+      "loss": 0.3808,
+      "step": 25600
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 0.39919450879096985,
+      "eval_runtime": 854.1993,
+      "eval_samples_per_second": 28.167,
+      "eval_steps_per_second": 14.083,
+      "step": 25600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.308460617115662e-06,
+      "loss": 0.5306,
+      "step": 25601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3049243914929155e-06,
+      "loss": 0.3223,
+      "step": 25602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.30138931199623e-06,
+      "loss": 0.2896,
+      "step": 25603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.297855378668426e-06,
+      "loss": 0.4289,
+      "step": 25604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2943225915522474e-06,
+      "loss": 0.4389,
+      "step": 25605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.290790950690471e-06,
+      "loss": 0.4765,
+      "step": 25606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.287260456125831e-06,
+      "loss": 0.3954,
+      "step": 25607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.283731107901036e-06,
+      "loss": 0.4747,
+      "step": 25608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.280202906058851e-06,
+      "loss": 0.3987,
+      "step": 25609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2766758506419545e-06,
+      "loss": 0.3867,
+      "step": 25610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.273149941693034e-06,
+      "loss": 0.3949,
+      "step": 25611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.269625179254756e-06,
+      "loss": 0.479,
+      "step": 25612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.266101563369785e-06,
+      "loss": 0.3741,
+      "step": 25613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.262579094080799e-06,
+      "loss": 0.3416,
+      "step": 25614
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.259057771430398e-06,
+      "loss": 0.4174,
+      "step": 25615
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.255537595461213e-06,
+      "loss": 0.3925,
+      "step": 25616
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.252018566215855e-06,
+      "loss": 0.3878,
+      "step": 25617
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.24850068373689e-06,
+      "loss": 0.3559,
+      "step": 25618
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.24498394806694e-06,
+      "loss": 0.4005,
+      "step": 25619
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.241468359248536e-06,
+      "loss": 0.3716,
+      "step": 25620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.237953917324234e-06,
+      "loss": 0.3823,
+      "step": 25621
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.234440622336578e-06,
+      "loss": 0.3105,
+      "step": 25622
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.230928474328079e-06,
+      "loss": 0.377,
+      "step": 25623
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.227417473341267e-06,
+      "loss": 0.3957,
+      "step": 25624
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.223907619418611e-06,
+      "loss": 0.4039,
+      "step": 25625
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.22039891260262e-06,
+      "loss": 0.3708,
+      "step": 25626
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2168913529357376e-06,
+      "loss": 0.3832,
+      "step": 25627
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.213384940460408e-06,
+      "loss": 0.3332,
+      "step": 25628
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2098796752190986e-06,
+      "loss": 0.4463,
+      "step": 25629
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2063755572542285e-06,
+      "loss": 0.3749,
+      "step": 25630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2028725866082095e-06,
+      "loss": 0.4679,
+      "step": 25631
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.199370763323419e-06,
+      "loss": 0.4201,
+      "step": 25632
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.195870087442267e-06,
+      "loss": 0.3999,
+      "step": 25633
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.192370559007098e-06,
+      "loss": 0.3785,
+      "step": 25634
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1888721780602885e-06,
+      "loss": 0.4065,
+      "step": 25635
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.185374944644184e-06,
+      "loss": 0.4015,
+      "step": 25636
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.181878858801092e-06,
+      "loss": 0.3855,
+      "step": 25637
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1783839205733375e-06,
+      "loss": 0.3178,
+      "step": 25638
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1748901300032185e-06,
+      "loss": 0.3856,
+      "step": 25639
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.171397487133023e-06,
+      "loss": 0.3811,
+      "step": 25640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.16790599200504e-06,
+      "loss": 0.4319,
+      "step": 25641
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.164415644661502e-06,
+      "loss": 0.3627,
+      "step": 25642
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.160926445144665e-06,
+      "loss": 0.3761,
+      "step": 25643
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.15743839349675e-06,
+      "loss": 0.3707,
+      "step": 25644
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.153951489759989e-06,
+      "loss": 0.3613,
+      "step": 25645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.150465733976583e-06,
+      "loss": 0.5453,
+      "step": 25646
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.146981126188721e-06,
+      "loss": 0.3332,
+      "step": 25647
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.143497666438568e-06,
+      "loss": 0.4227,
+      "step": 25648
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1400153547682906e-06,
+      "loss": 0.4722,
+      "step": 25649
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.136534191220044e-06,
+      "loss": 0.4392,
+      "step": 25650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.133054175835961e-06,
+      "loss": 0.2819,
+      "step": 25651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.129575308658152e-06,
+      "loss": 0.3968,
+      "step": 25652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.126097589728729e-06,
+      "loss": 0.4095,
+      "step": 25653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1226210190897685e-06,
+      "loss": 0.3792,
+      "step": 25654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1191455967833814e-06,
+      "loss": 0.4353,
+      "step": 25655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.115671322851601e-06,
+      "loss": 0.3777,
+      "step": 25656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.112198197336504e-06,
+      "loss": 0.3861,
+      "step": 25657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1087262202801025e-06,
+      "loss": 0.412,
+      "step": 25658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.105255391724417e-06,
+      "loss": 0.4023,
+      "step": 25659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.101785711711482e-06,
+      "loss": 0.4138,
+      "step": 25660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.098317180283285e-06,
+      "loss": 0.4023,
+      "step": 25661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.094849797481793e-06,
+      "loss": 0.3855,
+      "step": 25662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.091383563348983e-06,
+      "loss": 0.4011,
+      "step": 25663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.087918477926801e-06,
+      "loss": 0.3364,
+      "step": 25664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.084454541257177e-06,
+      "loss": 0.3755,
+      "step": 25665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.08099175338207e-06,
+      "loss": 0.4853,
+      "step": 25666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.077530114343365e-06,
+      "loss": 0.4569,
+      "step": 25667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.074069624182964e-06,
+      "loss": 0.4613,
+      "step": 25668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.070610282942756e-06,
+      "loss": 0.3746,
+      "step": 25669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.067152090664596e-06,
+      "loss": 0.4169,
+      "step": 25670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.063695047390349e-06,
+      "loss": 0.4459,
+      "step": 25671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.060239153161872e-06,
+      "loss": 0.3738,
+      "step": 25672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.056784408020965e-06,
+      "loss": 0.3216,
+      "step": 25673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0533308120094715e-06,
+      "loss": 0.4616,
+      "step": 25674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.049878365169147e-06,
+      "loss": 0.4268,
+      "step": 25675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0464270675418255e-06,
+      "loss": 0.4379,
+      "step": 25676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.042976919169251e-06,
+      "loss": 0.4648,
+      "step": 25677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.039527920093201e-06,
+      "loss": 0.3595,
+      "step": 25678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.036080070355409e-06,
+      "loss": 0.453,
+      "step": 25679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0326333699975966e-06,
+      "loss": 0.4727,
+      "step": 25680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.029187819061498e-06,
+      "loss": 0.3931,
+      "step": 25681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.025743417588802e-06,
+      "loss": 0.3544,
+      "step": 25682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.022300165621208e-06,
+      "loss": 0.3696,
+      "step": 25683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.018858063200371e-06,
+      "loss": 0.3484,
+      "step": 25684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0154171103679705e-06,
+      "loss": 0.3735,
+      "step": 25685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0119773071656606e-06,
+      "loss": 0.3859,
+      "step": 25686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0085386536350535e-06,
+      "loss": 0.3831,
+      "step": 25687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.005101149817781e-06,
+      "loss": 0.3723,
+      "step": 25688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.001664795755434e-06,
+      "loss": 0.3903,
+      "step": 25689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9982295914896335e-06,
+      "loss": 0.4956,
+      "step": 25690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.994795537061936e-06,
+      "loss": 0.4077,
+      "step": 25691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.991362632513908e-06,
+      "loss": 0.4061,
+      "step": 25692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.987930877887093e-06,
+      "loss": 0.3936,
+      "step": 25693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.984500273223025e-06,
+      "loss": 0.3704,
+      "step": 25694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.981070818563238e-06,
+      "loss": 0.4076,
+      "step": 25695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9776425139492435e-06,
+      "loss": 0.3931,
+      "step": 25696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.974215359422518e-06,
+      "loss": 0.4654,
+      "step": 25697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9707893550245635e-06,
+      "loss": 0.3997,
+      "step": 25698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.967364500796823e-06,
+      "loss": 0.39,
+      "step": 25699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.963940796780742e-06,
+      "loss": 0.4067,
+      "step": 25700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.960518243017797e-06,
+      "loss": 0.3945,
+      "step": 25701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.957096839549391e-06,
+      "loss": 0.402,
+      "step": 25702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.953676586416933e-06,
+      "loss": 0.3761,
+      "step": 25703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.950257483661813e-06,
+      "loss": 0.3559,
+      "step": 25704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.94683953132542e-06,
+      "loss": 0.3872,
+      "step": 25705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.943422729449121e-06,
+      "loss": 0.3776,
+      "step": 25706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.940007078074282e-06,
+      "loss": 0.376,
+      "step": 25707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.936592577242239e-06,
+      "loss": 0.3206,
+      "step": 25708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.933179226994311e-06,
+      "loss": 0.4129,
+      "step": 25709
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9297670273718014e-06,
+      "loss": 0.4225,
+      "step": 25710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9263559784160416e-06,
+      "loss": 0.3715,
+      "step": 25711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9229460801682896e-06,
+      "loss": 0.406,
+      "step": 25712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.919537332669832e-06,
+      "loss": 0.4354,
+      "step": 25713
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.916129735961905e-06,
+      "loss": 0.4049,
+      "step": 25714
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.912723290085764e-06,
+      "loss": 0.3692,
+      "step": 25715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.909317995082652e-06,
+      "loss": 0.4281,
+      "step": 25716
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.90591385099377e-06,
+      "loss": 0.5123,
+      "step": 25717
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.90251085786032e-06,
+      "loss": 0.4159,
+      "step": 25718
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.899109015723502e-06,
+      "loss": 0.4018,
+      "step": 25719
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8957083246244485e-06,
+      "loss": 0.4193,
+      "step": 25720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.892308784604383e-06,
+      "loss": 0.4698,
+      "step": 25721
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.888910395704405e-06,
+      "loss": 0.4055,
+      "step": 25722
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.88551315796566e-06,
+      "loss": 0.4432,
+      "step": 25723
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.882117071429259e-06,
+      "loss": 0.4174,
+      "step": 25724
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.878722136136316e-06,
+      "loss": 0.4124,
+      "step": 25725
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.875328352127917e-06,
+      "loss": 0.3926,
+      "step": 25726
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8719357194451425e-06,
+      "loss": 0.3778,
+      "step": 25727
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.868544238129047e-06,
+      "loss": 0.463,
+      "step": 25728
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.865153908220687e-06,
+      "loss": 0.3285,
+      "step": 25729
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.861764729761076e-06,
+      "loss": 0.3989,
+      "step": 25730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.858376702791267e-06,
+      "loss": 0.3855,
+      "step": 25731
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.854989827352241e-06,
+      "loss": 0.3459,
+      "step": 25732
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.851604103485008e-06,
+      "loss": 0.4452,
+      "step": 25733
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8482195312305355e-06,
+      "loss": 0.48,
+      "step": 25734
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8448361106297915e-06,
+      "loss": 0.4447,
+      "step": 25735
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.841453841723709e-06,
+      "loss": 0.4233,
+      "step": 25736
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.838072724553256e-06,
+      "loss": 0.4134,
+      "step": 25737
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.834692759159342e-06,
+      "loss": 0.4288,
+      "step": 25738
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.831313945582871e-06,
+      "loss": 0.4155,
+      "step": 25739
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.827936283864742e-06,
+      "loss": 0.4053,
+      "step": 25740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.824559774045822e-06,
+      "loss": 0.3958,
+      "step": 25741
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.821184416167013e-06,
+      "loss": 0.4914,
+      "step": 25742
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.817810210269136e-06,
+      "loss": 0.4129,
+      "step": 25743
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8144371563930476e-06,
+      "loss": 0.4086,
+      "step": 25744
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.811065254579561e-06,
+      "loss": 0.3852,
+      "step": 25745
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.807694504869475e-06,
+      "loss": 0.4545,
+      "step": 25746
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8043249073036255e-06,
+      "loss": 0.3964,
+      "step": 25747
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.800956461922779e-06,
+      "loss": 0.4299,
+      "step": 25748
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.797589168767691e-06,
+      "loss": 0.3787,
+      "step": 25749
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.79422302787913e-06,
+      "loss": 0.3937,
+      "step": 25750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7908580392978186e-06,
+      "loss": 0.3966,
+      "step": 25751
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.787494203064513e-06,
+      "loss": 0.4318,
+      "step": 25752
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7841315192199145e-06,
+      "loss": 0.3801,
+      "step": 25753
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.780769987804723e-06,
+      "loss": 0.3784,
+      "step": 25754
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.777409608859629e-06,
+      "loss": 0.4202,
+      "step": 25755
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.774050382425277e-06,
+      "loss": 0.4069,
+      "step": 25756
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.770692308542357e-06,
+      "loss": 0.424,
+      "step": 25757
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.767335387251503e-06,
+      "loss": 0.3936,
+      "step": 25758
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.763979618593351e-06,
+      "loss": 0.4326,
+      "step": 25759
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.76062500260851e-06,
+      "loss": 0.4296,
+      "step": 25760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.757271539337571e-06,
+      "loss": 0.4867,
+      "step": 25761
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.753919228821135e-06,
+      "loss": 0.4113,
+      "step": 25762
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.750568071099781e-06,
+      "loss": 0.354,
+      "step": 25763
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.747218066214066e-06,
+      "loss": 0.4769,
+      "step": 25764
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.743869214204522e-06,
+      "loss": 0.4019,
+      "step": 25765
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.740521515111707e-06,
+      "loss": 0.4262,
+      "step": 25766
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.737174968976099e-06,
+      "loss": 0.3841,
+      "step": 25767
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.733829575838245e-06,
+      "loss": 0.432,
+      "step": 25768
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.730485335738622e-06,
+      "loss": 0.4368,
+      "step": 25769
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.727142248717698e-06,
+      "loss": 0.4685,
+      "step": 25770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7238003148159406e-06,
+      "loss": 0.4877,
+      "step": 25771
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.720459534073784e-06,
+      "loss": 0.3924,
+      "step": 25772
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.717119906531686e-06,
+      "loss": 0.4213,
+      "step": 25773
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.713781432230069e-06,
+      "loss": 0.3994,
+      "step": 25774
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7104441112093225e-06,
+      "loss": 0.4518,
+      "step": 25775
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.707107943509848e-06,
+      "loss": 0.4545,
+      "step": 25776
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.703772929172001e-06,
+      "loss": 0.4212,
+      "step": 25777
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.700439068236184e-06,
+      "loss": 0.4748,
+      "step": 25778
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6971063607427315e-06,
+      "loss": 0.4282,
+      "step": 25779
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.693774806731976e-06,
+      "loss": 0.4254,
+      "step": 25780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.690444406244254e-06,
+      "loss": 0.3489,
+      "step": 25781
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.687115159319844e-06,
+      "loss": 0.4294,
+      "step": 25782
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.683787065999068e-06,
+      "loss": 0.4079,
+      "step": 25783
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.680460126322206e-06,
+      "loss": 0.3955,
+      "step": 25784
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.677134340329526e-06,
+      "loss": 0.4836,
+      "step": 25785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.673809708061261e-06,
+      "loss": 0.3769,
+      "step": 25786
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.670486229557658e-06,
+      "loss": 0.4172,
+      "step": 25787
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.667163904858962e-06,
+      "loss": 0.3903,
+      "step": 25788
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.663842734005375e-06,
+      "loss": 0.3501,
+      "step": 25789
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6605227170370744e-06,
+      "loss": 0.4537,
+      "step": 25790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.657203853994274e-06,
+      "loss": 0.4122,
+      "step": 25791
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.653886144917108e-06,
+      "loss": 0.4651,
+      "step": 25792
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6505695898457655e-06,
+      "loss": 0.4553,
+      "step": 25793
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6472541888203715e-06,
+      "loss": 0.4336,
+      "step": 25794
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.643939941881048e-06,
+      "loss": 0.388,
+      "step": 25795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.640626849067931e-06,
+      "loss": 0.4462,
+      "step": 25796
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.637314910421098e-06,
+      "loss": 0.4227,
+      "step": 25797
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.634004125980618e-06,
+      "loss": 0.3892,
+      "step": 25798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.630694495786614e-06,
+      "loss": 0.4221,
+      "step": 25799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.627386019879109e-06,
+      "loss": 0.3811,
+      "step": 25800
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 0.3988822400569916,
+      "eval_runtime": 844.4889,
+      "eval_samples_per_second": 28.491,
+      "eval_steps_per_second": 14.245,
+      "step": 25800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.624078698298151e-06,
+      "loss": 0.4283,
+      "step": 25801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.620772531083773e-06,
+      "loss": 0.5094,
+      "step": 25802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6174675182759756e-06,
+      "loss": 0.3585,
+      "step": 25803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.614163659914794e-06,
+      "loss": 0.3656,
+      "step": 25804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.610860956040186e-06,
+      "loss": 0.4004,
+      "step": 25805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.607559406692141e-06,
+      "loss": 0.4158,
+      "step": 25806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.604259011910617e-06,
+      "loss": 0.45,
+      "step": 25807
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.600959771735536e-06,
+      "loss": 0.3924,
+      "step": 25808
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.597661686206867e-06,
+      "loss": 0.4308,
+      "step": 25809
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.594364755364511e-06,
+      "loss": 0.3679,
+      "step": 25810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.59106897924837e-06,
+      "loss": 0.4541,
+      "step": 25811
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.587774357898333e-06,
+      "loss": 0.3635,
+      "step": 25812
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.584480891354259e-06,
+      "loss": 0.433,
+      "step": 25813
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.58118857965606e-06,
+      "loss": 0.366,
+      "step": 25814
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5778974228435466e-06,
+      "loss": 0.3689,
+      "step": 25815
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.574607420956556e-06,
+      "loss": 0.4246,
+      "step": 25816
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5713185740349e-06,
+      "loss": 0.3765,
+      "step": 25817
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.568030882118413e-06,
+      "loss": 0.4121,
+      "step": 25818
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.564744345246863e-06,
+      "loss": 0.3644,
+      "step": 25819
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.56145896346003e-06,
+      "loss": 0.4377,
+      "step": 25820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.558174736797694e-06,
+      "loss": 0.4194,
+      "step": 25821
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.554891665299566e-06,
+      "loss": 0.3673,
+      "step": 25822
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5516097490054365e-06,
+      "loss": 0.4467,
+      "step": 25823
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.548328987954997e-06,
+      "loss": 0.5119,
+      "step": 25824
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.54504938218796e-06,
+      "loss": 0.3887,
+      "step": 25825
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.541770931744027e-06,
+      "loss": 0.4787,
+      "step": 25826
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.538493636662855e-06,
+      "loss": 0.3622,
+      "step": 25827
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.535217496984135e-06,
+      "loss": 0.4418,
+      "step": 25828
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5319425127475115e-06,
+      "loss": 0.4213,
+      "step": 25829
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5286686839926205e-06,
+      "loss": 0.3714,
+      "step": 25830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.525396010759087e-06,
+      "loss": 0.423,
+      "step": 25831
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.52212449308651e-06,
+      "loss": 0.4182,
+      "step": 25832
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.518854131014505e-06,
+      "loss": 0.4313,
+      "step": 25833
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.515584924582661e-06,
+      "loss": 0.4058,
+      "step": 25834
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.512316873830513e-06,
+      "loss": 0.363,
+      "step": 25835
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.509049978797653e-06,
+      "loss": 0.4705,
+      "step": 25836
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.505784239523592e-06,
+      "loss": 0.4297,
+      "step": 25837
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.502519656047855e-06,
+      "loss": 0.4009,
+      "step": 25838
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.499256228409976e-06,
+      "loss": 0.429,
+      "step": 25839
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.495993956649447e-06,
+      "loss": 0.3689,
+      "step": 25840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.492732840805758e-06,
+      "loss": 0.3928,
+      "step": 25841
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.489472880918355e-06,
+      "loss": 0.4358,
+      "step": 25842
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.486214077026707e-06,
+      "loss": 0.4132,
+      "step": 25843
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.482956429170271e-06,
+      "loss": 0.4155,
+      "step": 25844
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.479699937388459e-06,
+      "loss": 0.4468,
+      "step": 25845
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4764446017206865e-06,
+      "loss": 0.3735,
+      "step": 25846
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4731904222063635e-06,
+      "loss": 0.4864,
+      "step": 25847
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.46993739888486e-06,
+      "loss": 0.3579,
+      "step": 25848
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.466685531795567e-06,
+      "loss": 0.3827,
+      "step": 25849
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.46343482097783e-06,
+      "loss": 0.3605,
+      "step": 25850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.460185266471006e-06,
+      "loss": 0.4251,
+      "step": 25851
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.456936868314421e-06,
+      "loss": 0.3874,
+      "step": 25852
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.453689626547363e-06,
+      "loss": 0.4571,
+      "step": 25853
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.450443541209182e-06,
+      "loss": 0.3935,
+      "step": 25854
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.447198612339143e-06,
+      "loss": 0.3896,
+      "step": 25855
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.443954839976517e-06,
+      "loss": 0.4211,
+      "step": 25856
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.44071222416057e-06,
+      "loss": 0.3878,
+      "step": 25857
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.437470764930529e-06,
+      "loss": 0.4052,
+      "step": 25858
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.434230462325673e-06,
+      "loss": 0.4157,
+      "step": 25859
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4309913163851805e-06,
+      "loss": 0.3365,
+      "step": 25860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4277533271482655e-06,
+      "loss": 0.3929,
+      "step": 25861
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.424516494654118e-06,
+      "loss": 0.3453,
+      "step": 25862
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.421280818941919e-06,
+      "loss": 0.4099,
+      "step": 25863
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.418046300050827e-06,
+      "loss": 0.3593,
+      "step": 25864
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4148129380200075e-06,
+      "loss": 0.3505,
+      "step": 25865
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.411580732888565e-06,
+      "loss": 0.4373,
+      "step": 25866
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.408349684695645e-06,
+      "loss": 0.3704,
+      "step": 25867
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.405119793480339e-06,
+      "loss": 0.4528,
+      "step": 25868
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4018910592817266e-06,
+      "loss": 0.3367,
+      "step": 25869
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.39866348213892e-06,
+      "loss": 0.3644,
+      "step": 25870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.395437062090979e-06,
+      "loss": 0.4133,
+      "step": 25871
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.392211799176926e-06,
+      "loss": 0.393,
+      "step": 25872
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.38898769343582e-06,
+      "loss": 0.3977,
+      "step": 25873
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.385764744906673e-06,
+      "loss": 0.3809,
+      "step": 25874
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.382542953628499e-06,
+      "loss": 0.4408,
+      "step": 25875
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3793223196403e-06,
+      "loss": 0.4366,
+      "step": 25876
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.376102842981045e-06,
+      "loss": 0.4918,
+      "step": 25877
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.372884523689702e-06,
+      "loss": 0.4294,
+      "step": 25878
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.369667361805208e-06,
+      "loss": 0.3444,
+      "step": 25879
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.36645135736653e-06,
+      "loss": 0.4005,
+      "step": 25880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.363236510412583e-06,
+      "loss": 0.334,
+      "step": 25881
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3600228209822795e-06,
+      "loss": 0.4259,
+      "step": 25882
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.35681028911451e-06,
+      "loss": 0.3745,
+      "step": 25883
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3535989148481445e-06,
+      "loss": 0.3086,
+      "step": 25884
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3503886982220744e-06,
+      "loss": 0.4299,
+      "step": 25885
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.347179639275145e-06,
+      "loss": 0.4174,
+      "step": 25886
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.343971738046193e-06,
+      "loss": 0.3858,
+      "step": 25887
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.340764994574054e-06,
+      "loss": 0.4099,
+      "step": 25888
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.337559408897518e-06,
+      "loss": 0.4681,
+      "step": 25889
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.334354981055411e-06,
+      "loss": 0.3921,
+      "step": 25890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.331151711086501e-06,
+      "loss": 0.4205,
+      "step": 25891
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.327949599029568e-06,
+      "loss": 0.385,
+      "step": 25892
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.32474864492336e-06,
+      "loss": 0.465,
+      "step": 25893
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.321548848806611e-06,
+      "loss": 0.3766,
+      "step": 25894
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.31835021071808e-06,
+      "loss": 0.3845,
+      "step": 25895
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3151527306964475e-06,
+      "loss": 0.3812,
+      "step": 25896
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.311956408780437e-06,
+      "loss": 0.3709,
+      "step": 25897
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.308761245008719e-06,
+      "loss": 0.4067,
+      "step": 25898
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.305567239419983e-06,
+      "loss": 0.4191,
+      "step": 25899
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.302374392052855e-06,
+      "loss": 0.4028,
+      "step": 25900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.299182702946014e-06,
+      "loss": 0.3373,
+      "step": 25901
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.295992172138075e-06,
+      "loss": 0.3436,
+      "step": 25902
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.292802799667661e-06,
+      "loss": 0.4482,
+      "step": 25903
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.289614585573365e-06,
+      "loss": 0.3764,
+      "step": 25904
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.286427529893755e-06,
+      "loss": 0.4425,
+      "step": 25905
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.283241632667456e-06,
+      "loss": 0.3964,
+      "step": 25906
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.280056893932993e-06,
+      "loss": 0.3432,
+      "step": 25907
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.276873313728924e-06,
+      "loss": 0.4751,
+      "step": 25908
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.273690892093773e-06,
+      "loss": 0.3723,
+      "step": 25909
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.270509629066044e-06,
+      "loss": 0.4098,
+      "step": 25910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.267329524684283e-06,
+      "loss": 0.3583,
+      "step": 25911
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.264150578986947e-06,
+      "loss": 0.4417,
+      "step": 25912
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.260972792012518e-06,
+      "loss": 0.3778,
+      "step": 25913
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.257796163799455e-06,
+      "loss": 0.3713,
+      "step": 25914
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.254620694386202e-06,
+      "loss": 0.3471,
+      "step": 25915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.251446383811219e-06,
+      "loss": 0.3524,
+      "step": 25916
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2482732321129094e-06,
+      "loss": 0.4886,
+      "step": 25917
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.245101239329663e-06,
+      "loss": 0.3937,
+      "step": 25918
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.241930405499905e-06,
+      "loss": 0.4155,
+      "step": 25919
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.238760730661962e-06,
+      "loss": 0.456,
+      "step": 25920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.235592214854256e-06,
+      "loss": 0.4884,
+      "step": 25921
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.232424858115103e-06,
+      "loss": 0.4724,
+      "step": 25922
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2292586604828484e-06,
+      "loss": 0.4136,
+      "step": 25923
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.226093621995808e-06,
+      "loss": 0.4431,
+      "step": 25924
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.222929742692283e-06,
+      "loss": 0.3971,
+      "step": 25925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.219767022610588e-06,
+      "loss": 0.4043,
+      "step": 25926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2166054617889915e-06,
+      "loss": 0.4209,
+      "step": 25927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.213445060265753e-06,
+      "loss": 0.3673,
+      "step": 25928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.21028581807914e-06,
+      "loss": 0.4339,
+      "step": 25929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.207127735267358e-06,
+      "loss": 0.4396,
+      "step": 25930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.203970811868663e-06,
+      "loss": 0.4066,
+      "step": 25931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.200815047921258e-06,
+      "loss": 0.4607,
+      "step": 25932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.197660443463325e-06,
+      "loss": 0.3973,
+      "step": 25933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.194506998533054e-06,
+      "loss": 0.3474,
+      "step": 25934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.191354713168616e-06,
+      "loss": 0.4549,
+      "step": 25935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.188203587408146e-06,
+      "loss": 0.3625,
+      "step": 25936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.185053621289814e-06,
+      "loss": 0.4074,
+      "step": 25937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1819048148517116e-06,
+      "loss": 0.3979,
+      "step": 25938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.178757168131975e-06,
+      "loss": 0.4986,
+      "step": 25939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1756106811686955e-06,
+      "loss": 0.4141,
+      "step": 25940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.172465353999944e-06,
+      "loss": 0.4343,
+      "step": 25941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.169321186663799e-06,
+      "loss": 0.4338,
+      "step": 25942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1661781791983215e-06,
+      "loss": 0.4228,
+      "step": 25943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.163036331641546e-06,
+      "loss": 0.4791,
+      "step": 25944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.159895644031486e-06,
+      "loss": 0.4242,
+      "step": 25945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.156756116406169e-06,
+      "loss": 0.4117,
+      "step": 25946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.153617748803595e-06,
+      "loss": 0.3884,
+      "step": 25947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1504805412617475e-06,
+      "loss": 0.3582,
+      "step": 25948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.147344493818605e-06,
+      "loss": 0.4465,
+      "step": 25949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.144209606512106e-06,
+      "loss": 0.3294,
+      "step": 25950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.141075879380196e-06,
+      "loss": 0.3871,
+      "step": 25951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.137943312460812e-06,
+      "loss": 0.3739,
+      "step": 25952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.134811905791869e-06,
+      "loss": 0.3806,
+      "step": 25953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.131681659411269e-06,
+      "loss": 0.36,
+      "step": 25954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.12855257335687e-06,
+      "loss": 0.4561,
+      "step": 25955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.125424647666576e-06,
+      "loss": 0.433,
+      "step": 25956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.122297882378245e-06,
+      "loss": 0.4628,
+      "step": 25957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1191722775297145e-06,
+      "loss": 0.4725,
+      "step": 25958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.11604783315881e-06,
+      "loss": 0.3748,
+      "step": 25959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.112924549303343e-06,
+      "loss": 0.4122,
+      "step": 25960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.10980242600113e-06,
+      "loss": 0.4115,
+      "step": 25961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.106681463289963e-06,
+      "loss": 0.3217,
+      "step": 25962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1035616612076e-06,
+      "loss": 0.4023,
+      "step": 25963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1004430197918e-06,
+      "loss": 0.4791,
+      "step": 25964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.09732553908031e-06,
+      "loss": 0.4217,
+      "step": 25965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0942092191108785e-06,
+      "loss": 0.4301,
+      "step": 25966
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.091094059921219e-06,
+      "loss": 0.432,
+      "step": 25967
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.087980061549024e-06,
+      "loss": 0.3878,
+      "step": 25968
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.084867224031985e-06,
+      "loss": 0.4707,
+      "step": 25969
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.081755547407784e-06,
+      "loss": 0.3599,
+      "step": 25970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.078645031714068e-06,
+      "loss": 0.3926,
+      "step": 25971
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0755356769885065e-06,
+      "loss": 0.4249,
+      "step": 25972
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.072427483268726e-06,
+      "loss": 0.3868,
+      "step": 25973
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.06932045059234e-06,
+      "loss": 0.3738,
+      "step": 25974
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.066214578996952e-06,
+      "loss": 0.4509,
+      "step": 25975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.063109868520154e-06,
+      "loss": 0.4361,
+      "step": 25976
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.060006319199539e-06,
+      "loss": 0.428,
+      "step": 25977
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0569039310726534e-06,
+      "loss": 0.3715,
+      "step": 25978
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.053802704177057e-06,
+      "loss": 0.3903,
+      "step": 25979
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.050702638550275e-06,
+      "loss": 0.4111,
+      "step": 25980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0476037342298214e-06,
+      "loss": 0.3599,
+      "step": 25981
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.044505991253222e-06,
+      "loss": 0.4403,
+      "step": 25982
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.041409409657982e-06,
+      "loss": 0.5344,
+      "step": 25983
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.038313989481546e-06,
+      "loss": 0.3836,
+      "step": 25984
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.035219730761397e-06,
+      "loss": 0.4459,
+      "step": 25985
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.032126633534972e-06,
+      "loss": 0.3789,
+      "step": 25986
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.029034697839729e-06,
+      "loss": 0.4175,
+      "step": 25987
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.025943923713082e-06,
+      "loss": 0.3641,
+      "step": 25988
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.022854311192437e-06,
+      "loss": 0.3772,
+      "step": 25989
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.019765860315195e-06,
+      "loss": 0.4971,
+      "step": 25990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0166785711187145e-06,
+      "loss": 0.379,
+      "step": 25991
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.013592443640391e-06,
+      "loss": 0.4076,
+      "step": 25992
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.010507477917558e-06,
+      "loss": 0.3966,
+      "step": 25993
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.007423673987576e-06,
+      "loss": 0.3989,
+      "step": 25994
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.004341031887737e-06,
+      "loss": 0.3938,
+      "step": 25995
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.001259551655368e-06,
+      "loss": 0.468,
+      "step": 25996
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.998179233327781e-06,
+      "loss": 0.5021,
+      "step": 25997
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.995100076942238e-06,
+      "loss": 0.4003,
+      "step": 25998
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.992022082536007e-06,
+      "loss": 0.3742,
+      "step": 25999
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9889452501463475e-06,
+      "loss": 0.4011,
+      "step": 26000
+    },
+    {
+      "epoch": 0.91,
+      "eval_loss": 0.39831990003585815,
+      "eval_runtime": 844.2736,
+      "eval_samples_per_second": 28.498,
+      "eval_steps_per_second": 14.249,
+      "step": 26000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.985869579810497e-06,
+      "loss": 0.4147,
+      "step": 26001
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9827950715656814e-06,
+      "loss": 0.4804,
+      "step": 26002
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.979721725449115e-06,
+      "loss": 0.4641,
+      "step": 26003
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.976649541498001e-06,
+      "loss": 0.4577,
+      "step": 26004
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.97357851974951e-06,
+      "loss": 0.4122,
+      "step": 26005
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.970508660240823e-06,
+      "loss": 0.4458,
+      "step": 26006
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.967439963009078e-06,
+      "loss": 0.3873,
+      "step": 26007
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.964372428091445e-06,
+      "loss": 0.3583,
+      "step": 26008
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.961306055525027e-06,
+      "loss": 0.3815,
+      "step": 26009
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.95824084534695e-06,
+      "loss": 0.367,
+      "step": 26010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9551767975943175e-06,
+      "loss": 0.428,
+      "step": 26011
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.952113912304178e-06,
+      "loss": 0.3556,
+      "step": 26012
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.949052189513658e-06,
+      "loss": 0.4278,
+      "step": 26013
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.945991629259782e-06,
+      "loss": 0.4446,
+      "step": 26014
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.942932231579599e-06,
+      "loss": 0.3794,
+      "step": 26015
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.939873996510135e-06,
+      "loss": 0.4296,
+      "step": 26016
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.936816924088393e-06,
+      "loss": 0.3298,
+      "step": 26017
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.933761014351401e-06,
+      "loss": 0.3919,
+      "step": 26018
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.930706267336137e-06,
+      "loss": 0.3971,
+      "step": 26019
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.927652683079575e-06,
+      "loss": 0.4276,
+      "step": 26020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.924600261618661e-06,
+      "loss": 0.3408,
+      "step": 26021
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9215490029903324e-06,
+      "loss": 0.3514,
+      "step": 26022
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.918498907231538e-06,
+      "loss": 0.3797,
+      "step": 26023
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.915449974379204e-06,
+      "loss": 0.4521,
+      "step": 26024
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.912402204470212e-06,
+      "loss": 0.4035,
+      "step": 26025
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.909355597541453e-06,
+      "loss": 0.4556,
+      "step": 26026
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9063101536297996e-06,
+      "loss": 0.4358,
+      "step": 26027
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.903265872772133e-06,
+      "loss": 0.3983,
+      "step": 26028
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.900222755005267e-06,
+      "loss": 0.4364,
+      "step": 26029
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.897180800366063e-06,
+      "loss": 0.4335,
+      "step": 26030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.894140008891323e-06,
+      "loss": 0.4356,
+      "step": 26031
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.89110038061784e-06,
+      "loss": 0.4357,
+      "step": 26032
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.88806191558243e-06,
+      "loss": 0.472,
+      "step": 26033
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.885024613821853e-06,
+      "loss": 0.4407,
+      "step": 26034
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8819884753728665e-06,
+      "loss": 0.4152,
+      "step": 26035
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.878953500272231e-06,
+      "loss": 0.4125,
+      "step": 26036
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8759196885566725e-06,
+      "loss": 0.4715,
+      "step": 26037
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8728870402628955e-06,
+      "loss": 0.42,
+      "step": 26038
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.869855555427637e-06,
+      "loss": 0.3774,
+      "step": 26039
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.866825234087568e-06,
+      "loss": 0.4362,
+      "step": 26040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.863796076279369e-06,
+      "loss": 0.3545,
+      "step": 26041
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.860768082039701e-06,
+      "loss": 0.3513,
+      "step": 26042
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.857741251405211e-06,
+      "loss": 0.3851,
+      "step": 26043
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.854715584412538e-06,
+      "loss": 0.4123,
+      "step": 26044
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8516910810983075e-06,
+      "loss": 0.4759,
+      "step": 26045
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.848667741499123e-06,
+      "loss": 0.3565,
+      "step": 26046
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8456455656515675e-06,
+      "loss": 0.4174,
+      "step": 26047
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8426245535922224e-06,
+      "loss": 0.4181,
+      "step": 26048
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.839604705357669e-06,
+      "loss": 0.4201,
+      "step": 26049
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.836586020984434e-06,
+      "loss": 0.3553,
+      "step": 26050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.833568500509066e-06,
+      "loss": 0.4151,
+      "step": 26051
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.830552143968092e-06,
+      "loss": 0.4428,
+      "step": 26052
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.827536951398003e-06,
+      "loss": 0.4339,
+      "step": 26053
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.824522922835305e-06,
+      "loss": 0.4087,
+      "step": 26054
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.82151005831648e-06,
+      "loss": 0.3912,
+      "step": 26055
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.818498357877998e-06,
+      "loss": 0.3133,
+      "step": 26056
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.815487821556285e-06,
+      "loss": 0.4035,
+      "step": 26057
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8124784493878017e-06,
+      "loss": 0.4325,
+      "step": 26058
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8094702414089632e-06,
+      "loss": 0.4796,
+      "step": 26059
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8064631976561847e-06,
+      "loss": 0.3256,
+      "step": 26060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8034573181658593e-06,
+      "loss": 0.3483,
+      "step": 26061
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.800452602974369e-06,
+      "loss": 0.4988,
+      "step": 26062
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7974490521180627e-06,
+      "loss": 0.4988,
+      "step": 26063
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.794446665633322e-06,
+      "loss": 0.3489,
+      "step": 26064
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.791445443556474e-06,
+      "loss": 0.4074,
+      "step": 26065
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7884453859238447e-06,
+      "loss": 0.3943,
+      "step": 26066
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.785446492771738e-06,
+      "loss": 0.3385,
+      "step": 26067
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7824487641364594e-06,
+      "loss": 0.417,
+      "step": 26068
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7794522000542785e-06,
+      "loss": 0.3667,
+      "step": 26069
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.776456800561479e-06,
+      "loss": 0.4558,
+      "step": 26070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.773462565694308e-06,
+      "loss": 0.5831,
+      "step": 26071
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.770469495489004e-06,
+      "loss": 0.4498,
+      "step": 26072
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.767477589981805e-06,
+      "loss": 0.3422,
+      "step": 26073
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7644868492088925e-06,
+      "loss": 0.4086,
+      "step": 26074
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7614972732065046e-06,
+      "loss": 0.4295,
+      "step": 26075
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.758508862010801e-06,
+      "loss": 0.3331,
+      "step": 26076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.755521615657953e-06,
+      "loss": 0.4031,
+      "step": 26077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.752535534184121e-06,
+      "loss": 0.3901,
+      "step": 26078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7495506176254415e-06,
+      "loss": 0.4875,
+      "step": 26079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.746566866018042e-06,
+      "loss": 0.37,
+      "step": 26080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.743584279398049e-06,
+      "loss": 0.3798,
+      "step": 26081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.740602857801556e-06,
+      "loss": 0.4524,
+      "step": 26082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7376226012646454e-06,
+      "loss": 0.3766,
+      "step": 26083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7346435098233655e-06,
+      "loss": 0.3795,
+      "step": 26084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7316655835138216e-06,
+      "loss": 0.3975,
+      "step": 26085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7286888223720286e-06,
+      "loss": 0.4445,
+      "step": 26086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7257132264340024e-06,
+      "loss": 0.4524,
+      "step": 26087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7227387957357694e-06,
+      "loss": 0.4309,
+      "step": 26088
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.719765530313335e-06,
+      "loss": 0.3532,
+      "step": 26089
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.716793430202692e-06,
+      "loss": 0.474,
+      "step": 26090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7138224954398006e-06,
+      "loss": 0.461,
+      "step": 26091
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7108527260606206e-06,
+      "loss": 0.4343,
+      "step": 26092
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7078841221010797e-06,
+      "loss": 0.4183,
+      "step": 26093
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7049166835971484e-06,
+      "loss": 0.4217,
+      "step": 26094
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.70195041058472e-06,
+      "loss": 0.4445,
+      "step": 26095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6989853030996892e-06,
+      "loss": 0.5053,
+      "step": 26096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6960213611779482e-06,
+      "loss": 0.3863,
+      "step": 26097
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.693058584855369e-06,
+      "loss": 0.4439,
+      "step": 26098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6900969741678225e-06,
+      "loss": 0.5021,
+      "step": 26099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.687136529151158e-06,
+      "loss": 0.5387,
+      "step": 26100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.684177249841181e-06,
+      "loss": 0.3621,
+      "step": 26101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.681219136273728e-06,
+      "loss": 0.384,
+      "step": 26102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6782621884846045e-06,
+      "loss": 0.4826,
+      "step": 26103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.675306406509571e-06,
+      "loss": 0.4398,
+      "step": 26104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.672351790384443e-06,
+      "loss": 0.3655,
+      "step": 26105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.669398340144958e-06,
+      "loss": 0.4146,
+      "step": 26106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6664460558268664e-06,
+      "loss": 0.4477,
+      "step": 26107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6634949374659057e-06,
+      "loss": 0.4618,
+      "step": 26108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.660544985097769e-06,
+      "loss": 0.5243,
+      "step": 26109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.657596198758195e-06,
+      "loss": 0.4588,
+      "step": 26110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6546485784828553e-06,
+      "loss": 0.3624,
+      "step": 26111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6517021243074435e-06,
+      "loss": 0.3849,
+      "step": 26112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6487568362675974e-06,
+      "loss": 0.3898,
+      "step": 26113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.645812714398966e-06,
+      "loss": 0.4437,
+      "step": 26114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6428697587371995e-06,
+      "loss": 0.431,
+      "step": 26115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6399279693179133e-06,
+      "loss": 0.386,
+      "step": 26116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6369873461767124e-06,
+      "loss": 0.4676,
+      "step": 26117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6340478893491904e-06,
+      "loss": 0.39,
+      "step": 26118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6311095988709076e-06,
+      "loss": 0.4357,
+      "step": 26119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6281724747774472e-06,
+      "loss": 0.4642,
+      "step": 26120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6252365171043467e-06,
+      "loss": 0.4177,
+      "step": 26121
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6223017258871563e-06,
+      "loss": 0.4004,
+      "step": 26122
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.619368101161369e-06,
+      "loss": 0.3093,
+      "step": 26123
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6164356429625125e-06,
+      "loss": 0.3864,
+      "step": 26124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.613504351326069e-06,
+      "loss": 0.4244,
+      "step": 26125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.610574226287533e-06,
+      "loss": 0.3949,
+      "step": 26126
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6076452678823648e-06,
+      "loss": 0.4403,
+      "step": 26127
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6047174761460025e-06,
+      "loss": 0.392,
+      "step": 26128
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6017908511138733e-06,
+      "loss": 0.4545,
+      "step": 26129
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5988653928214265e-06,
+      "loss": 0.3872,
+      "step": 26130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5959411013040455e-06,
+      "loss": 0.4462,
+      "step": 26131
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.593017976597146e-06,
+      "loss": 0.4082,
+      "step": 26132
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5900960187361e-06,
+      "loss": 0.4511,
+      "step": 26133
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5871752277562566e-06,
+      "loss": 0.4063,
+      "step": 26134
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5842556036929874e-06,
+      "loss": 0.4304,
+      "step": 26135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5813371465816315e-06,
+      "loss": 0.4647,
+      "step": 26136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5784198564574934e-06,
+      "loss": 0.4292,
+      "step": 26137
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5755037333558896e-06,
+      "loss": 0.4285,
+      "step": 26138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5725887773121246e-06,
+      "loss": 0.3977,
+      "step": 26139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.56967498836146e-06,
+      "loss": 0.3907,
+      "step": 26140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.566762366539178e-06,
+      "loss": 0.4714,
+      "step": 26141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5638509118805287e-06,
+      "loss": 0.4355,
+      "step": 26142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.56094062442075e-06,
+      "loss": 0.3326,
+      "step": 26143
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5580315041950696e-06,
+      "loss": 0.4086,
+      "step": 26144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5551235512386817e-06,
+      "loss": 0.4355,
+      "step": 26145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.552216765586791e-06,
+      "loss": 0.4854,
+      "step": 26146
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.549311147274592e-06,
+      "loss": 0.3884,
+      "step": 26147
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5464066963372455e-06,
+      "loss": 0.4498,
+      "step": 26148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.54350341280989e-06,
+      "loss": 0.4023,
+      "step": 26149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5406012967276746e-06,
+      "loss": 0.3837,
+      "step": 26150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5377003481257277e-06,
+      "loss": 0.5413,
+      "step": 26151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5348005670391648e-06,
+      "loss": 0.3628,
+      "step": 26152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5319019535030696e-06,
+      "loss": 0.469,
+      "step": 26153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.529004507552547e-06,
+      "loss": 0.4266,
+      "step": 26154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5261082292226245e-06,
+      "loss": 0.4168,
+      "step": 26155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5232131185484076e-06,
+      "loss": 0.4129,
+      "step": 26156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5203191755649013e-06,
+      "loss": 0.3956,
+      "step": 26157
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5174264003071554e-06,
+      "loss": 0.3657,
+      "step": 26158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5145347928101535e-06,
+      "loss": 0.3626,
+      "step": 26159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5116443531089117e-06,
+      "loss": 0.377,
+      "step": 26160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.508755081238424e-06,
+      "loss": 0.3779,
+      "step": 26161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5058669772336406e-06,
+      "loss": 0.3985,
+      "step": 26162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5029800411295334e-06,
+      "loss": 0.5006,
+      "step": 26163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5000942729610297e-06,
+      "loss": 0.3806,
+      "step": 26164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4972096727630468e-06,
+      "loss": 0.3374,
+      "step": 26165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4943262405705335e-06,
+      "loss": 0.4452,
+      "step": 26166
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4914439764183737e-06,
+      "loss": 0.4438,
+      "step": 26167
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4885628803414394e-06,
+      "loss": 0.3329,
+      "step": 26168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.485682952374614e-06,
+      "loss": 0.4622,
+      "step": 26169
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4828041925527465e-06,
+      "loss": 0.4776,
+      "step": 26170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4799266009106767e-06,
+      "loss": 0.4152,
+      "step": 26171
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.477050177483254e-06,
+      "loss": 0.3817,
+      "step": 26172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4741749223052732e-06,
+      "loss": 0.402,
+      "step": 26173
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4713008354115393e-06,
+      "loss": 0.5078,
+      "step": 26174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.468427916836847e-06,
+      "loss": 0.3748,
+      "step": 26175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465556166615935e-06,
+      "loss": 0.453,
+      "step": 26176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4626855847836094e-06,
+      "loss": 0.3959,
+      "step": 26177
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4598161713745857e-06,
+      "loss": 0.4463,
+      "step": 26178
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4569479264235926e-06,
+      "loss": 0.3955,
+      "step": 26179
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4540808499653575e-06,
+      "loss": 0.451,
+      "step": 26180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.451214942034564e-06,
+      "loss": 0.4073,
+      "step": 26181
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4483502026659287e-06,
+      "loss": 0.4196,
+      "step": 26182
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4454866318940905e-06,
+      "loss": 0.4471,
+      "step": 26183
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4426242297537327e-06,
+      "loss": 0.3917,
+      "step": 26184
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4397629962794943e-06,
+      "loss": 0.4301,
+      "step": 26185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.436902931505981e-06,
+      "loss": 0.4166,
+      "step": 26186
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.434044035467854e-06,
+      "loss": 0.5154,
+      "step": 26187
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4311863081996852e-06,
+      "loss": 0.3765,
+      "step": 26188
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.428329749736059e-06,
+      "loss": 0.3728,
+      "step": 26189
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4254743601115803e-06,
+      "loss": 0.4004,
+      "step": 26190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.422620139360766e-06,
+      "loss": 0.3473,
+      "step": 26191
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4197670875181885e-06,
+      "loss": 0.3927,
+      "step": 26192
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4169152046183874e-06,
+      "loss": 0.392,
+      "step": 26193
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4140644906958564e-06,
+      "loss": 0.4422,
+      "step": 26194
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4112149457851127e-06,
+      "loss": 0.4137,
+      "step": 26195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.40836656992064e-06,
+      "loss": 0.346,
+      "step": 26196
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.405519363136911e-06,
+      "loss": 0.3962,
+      "step": 26197
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4026733254684085e-06,
+      "loss": 0.4038,
+      "step": 26198
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3998284569495497e-06,
+      "loss": 0.4503,
+      "step": 26199
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.396984757614774e-06,
+      "loss": 0.4219,
+      "step": 26200
+    },
+    {
+      "epoch": 0.92,
+      "eval_loss": 0.3981442153453827,
+      "eval_runtime": 848.677,
+      "eval_samples_per_second": 28.35,
+      "eval_steps_per_second": 14.175,
+      "step": 26200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.394142227498498e-06,
+      "loss": 0.3596,
+      "step": 26201
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3913008666351498e-06,
+      "loss": 0.4678,
+      "step": 26202
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3884606750590907e-06,
+      "loss": 0.4364,
+      "step": 26203
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3856216528047156e-06,
+      "loss": 0.4171,
+      "step": 26204
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.382783799906375e-06,
+      "loss": 0.4881,
+      "step": 26205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3799471163984187e-06,
+      "loss": 0.3505,
+      "step": 26206
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3771116023151638e-06,
+      "loss": 0.4845,
+      "step": 26207
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3742772576909608e-06,
+      "loss": 0.4074,
+      "step": 26208
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3714440825601044e-06,
+      "loss": 0.3783,
+      "step": 26209
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3686120769568787e-06,
+      "loss": 0.3684,
+      "step": 26210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3657812409155665e-06,
+      "loss": 0.4917,
+      "step": 26211
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3629515744704075e-06,
+      "loss": 0.393,
+      "step": 26212
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3601230776556856e-06,
+      "loss": 0.4227,
+      "step": 26213
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3572957505056178e-06,
+      "loss": 0.3841,
+      "step": 26214
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3544695930544214e-06,
+      "loss": 0.4059,
+      "step": 26215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3516446053363015e-06,
+      "loss": 0.5394,
+      "step": 26216
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.348820787385454e-06,
+      "loss": 0.4173,
+      "step": 26217
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3459981392360617e-06,
+      "loss": 0.4391,
+      "step": 26218
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3431766609222982e-06,
+      "loss": 0.4391,
+      "step": 26219
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.340356352478291e-06,
+      "loss": 0.3789,
+      "step": 26220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3375372139381576e-06,
+      "loss": 0.4105,
+      "step": 26221
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3347192453360708e-06,
+      "loss": 0.4083,
+      "step": 26222
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.331902446706092e-06,
+      "loss": 0.427,
+      "step": 26223
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3290868180823497e-06,
+      "loss": 0.4321,
+      "step": 26224
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3262723594988944e-06,
+      "loss": 0.3861,
+      "step": 26225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3234590709897872e-06,
+      "loss": 0.387,
+      "step": 26226
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3206469525891128e-06,
+      "loss": 0.3828,
+      "step": 26227
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3178360043308765e-06,
+      "loss": 0.4125,
+      "step": 26228
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3150262262491184e-06,
+      "loss": 0.446,
+      "step": 26229
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.312217618377822e-06,
+      "loss": 0.3569,
+      "step": 26230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.309410180751005e-06,
+      "loss": 0.4119,
+      "step": 26231
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3066039134026396e-06,
+      "loss": 0.3682,
+      "step": 26232
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3037988163666877e-06,
+      "loss": 0.3676,
+      "step": 26233
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3009948896771e-06,
+      "loss": 0.3595,
+      "step": 26234
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2981921333678166e-06,
+      "loss": 0.4451,
+      "step": 26235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.295390547472754e-06,
+      "loss": 0.387,
+      "step": 26236
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.29259013202583e-06,
+      "loss": 0.4715,
+      "step": 26237
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2897908870609397e-06,
+      "loss": 0.4028,
+      "step": 26238
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2869928126119553e-06,
+      "loss": 0.3819,
+      "step": 26239
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2841959087127505e-06,
+      "loss": 0.3572,
+      "step": 26240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.281400175397165e-06,
+      "loss": 0.3987,
+      "step": 26241
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.278605612699037e-06,
+      "loss": 0.4286,
+      "step": 26242
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.275812220652208e-06,
+      "loss": 0.4188,
+      "step": 26243
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2730199992904832e-06,
+      "loss": 0.4289,
+      "step": 26244
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.270228948647647e-06,
+      "loss": 0.4599,
+      "step": 26245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.267439068757483e-06,
+      "loss": 0.4366,
+      "step": 26246
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2646503596537424e-06,
+      "loss": 0.4323,
+      "step": 26247
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2618628213702095e-06,
+      "loss": 0.3285,
+      "step": 26248
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2590764539406126e-06,
+      "loss": 0.3498,
+      "step": 26249
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.256291257398669e-06,
+      "loss": 0.4318,
+      "step": 26250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2535072317780857e-06,
+      "loss": 0.4784,
+      "step": 26251
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2507243771125574e-06,
+      "loss": 0.3798,
+      "step": 26252
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2479426934357904e-06,
+      "loss": 0.3589,
+      "step": 26253
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2451621807814246e-06,
+      "loss": 0.4269,
+      "step": 26254
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2423828391831334e-06,
+      "loss": 0.4229,
+      "step": 26255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.239604668674545e-06,
+      "loss": 0.4757,
+      "step": 26256
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2368276692892664e-06,
+      "loss": 0.3834,
+      "step": 26257
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2340518410609476e-06,
+      "loss": 0.4281,
+      "step": 26258
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2312771840231627e-06,
+      "loss": 0.4286,
+      "step": 26259
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.228503698209495e-06,
+      "loss": 0.4411,
+      "step": 26260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2257313836535184e-06,
+      "loss": 0.3909,
+      "step": 26261
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2229602403887725e-06,
+      "loss": 0.4592,
+      "step": 26262
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2201902684488304e-06,
+      "loss": 0.4765,
+      "step": 26263
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2174214678671876e-06,
+      "loss": 0.4332,
+      "step": 26264
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2146538386773616e-06,
+      "loss": 0.3966,
+      "step": 26265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.211887380912859e-06,
+      "loss": 0.4219,
+      "step": 26266
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.209122094607142e-06,
+      "loss": 0.3514,
+      "step": 26267
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.206357979793706e-06,
+      "loss": 0.471,
+      "step": 26268
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2035950365060017e-06,
+      "loss": 0.398,
+      "step": 26269
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2008332647774584e-06,
+      "loss": 0.3822,
+      "step": 26270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.198072664641505e-06,
+      "loss": 0.4402,
+      "step": 26271
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1953132361315585e-06,
+      "loss": 0.3276,
+      "step": 26272
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.192554979281004e-06,
+      "loss": 0.394,
+      "step": 26273
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1897978941232477e-06,
+      "loss": 0.3993,
+      "step": 26274
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.187041980691652e-06,
+      "loss": 0.3425,
+      "step": 26275
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.184287239019568e-06,
+      "loss": 0.4288,
+      "step": 26276
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.181533669140346e-06,
+      "loss": 0.4027,
+      "step": 26277
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1787812710872832e-06,
+      "loss": 0.4685,
+      "step": 26278
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1760300448937408e-06,
+      "loss": 0.4731,
+      "step": 26279
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.173279990592981e-06,
+      "loss": 0.4296,
+      "step": 26280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1705311082182996e-06,
+      "loss": 0.3967,
+      "step": 26281
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.16778339780297e-06,
+      "loss": 0.4233,
+      "step": 26282
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.165036859380244e-06,
+      "loss": 0.4275,
+      "step": 26283
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.162291492983371e-06,
+      "loss": 0.3875,
+      "step": 26284
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.159547298645571e-06,
+      "loss": 0.4119,
+      "step": 26285
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1568042764000607e-06,
+      "loss": 0.4086,
+      "step": 26286
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1540624262800467e-06,
+      "loss": 0.4672,
+      "step": 26287
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.151321748318692e-06,
+      "loss": 0.4235,
+      "step": 26288
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.148582242549203e-06,
+      "loss": 0.3664,
+      "step": 26289
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1458439090047196e-06,
+      "loss": 0.4444,
+      "step": 26290
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1431067477183717e-06,
+      "loss": 0.4129,
+      "step": 26291
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.14037075872331e-06,
+      "loss": 0.4711,
+      "step": 26292
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.137635942052619e-06,
+      "loss": 0.3409,
+      "step": 26293
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1349022977394394e-06,
+      "loss": 0.4589,
+      "step": 26294
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1321698258168332e-06,
+      "loss": 0.4052,
+      "step": 26295
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.129438526317874e-06,
+      "loss": 0.4359,
+      "step": 26296
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1267083992756242e-06,
+      "loss": 0.4282,
+      "step": 26297
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.123979444723113e-06,
+      "loss": 0.3308,
+      "step": 26298
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.121251662693392e-06,
+      "loss": 0.4075,
+      "step": 26299
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1185250532194675e-06,
+      "loss": 0.3879,
+      "step": 26300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.115799616334336e-06,
+      "loss": 0.5008,
+      "step": 26301
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.113075352070993e-06,
+      "loss": 0.4064,
+      "step": 26302
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.110352260462401e-06,
+      "loss": 0.3994,
+      "step": 26303
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.107630341541512e-06,
+      "loss": 0.3525,
+      "step": 26304
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1049095953412987e-06,
+      "loss": 0.3035,
+      "step": 26305
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1021900218946686e-06,
+      "loss": 0.4631,
+      "step": 26306
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.099471621234551e-06,
+      "loss": 0.4961,
+      "step": 26307
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0967543933938303e-06,
+      "loss": 0.2975,
+      "step": 26308
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.094038338405403e-06,
+      "loss": 0.4343,
+      "step": 26309
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.091323456302142e-06,
+      "loss": 0.352,
+      "step": 26310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0886097471169217e-06,
+      "loss": 0.3402,
+      "step": 26311
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0858972108825603e-06,
+      "loss": 0.4576,
+      "step": 26312
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0831858476319086e-06,
+      "loss": 0.3621,
+      "step": 26313
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0804756573977524e-06,
+      "loss": 0.4046,
+      "step": 26314
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0777666402129425e-06,
+      "loss": 0.3696,
+      "step": 26315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0750587961102308e-06,
+      "loss": 0.4361,
+      "step": 26316
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.072352125122413e-06,
+      "loss": 0.3857,
+      "step": 26317
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.06964662728223e-06,
+      "loss": 0.3932,
+      "step": 26318
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.066942302622422e-06,
+      "loss": 0.4067,
+      "step": 26319
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0642391511757518e-06,
+      "loss": 0.4032,
+      "step": 26320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0615371729749154e-06,
+      "loss": 0.4407,
+      "step": 26321
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.05883636805262e-06,
+      "loss": 0.321,
+      "step": 26322
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0561367364415503e-06,
+      "loss": 0.3981,
+      "step": 26323
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.053438278174381e-06,
+      "loss": 0.4171,
+      "step": 26324
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0507409932837736e-06,
+      "loss": 0.3445,
+      "step": 26325
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.048044881802381e-06,
+      "loss": 0.4433,
+      "step": 26326
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.045349943762832e-06,
+      "loss": 0.372,
+      "step": 26327
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.042656179197745e-06,
+      "loss": 0.4534,
+      "step": 26328
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0399635881397047e-06,
+      "loss": 0.4441,
+      "step": 26329
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.03727217062133e-06,
+      "loss": 0.3781,
+      "step": 26330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.034581926675173e-06,
+      "loss": 0.4543,
+      "step": 26331
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.031892856333818e-06,
+      "loss": 0.3768,
+      "step": 26332
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0292049596297834e-06,
+      "loss": 0.5815,
+      "step": 26333
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.026518236595621e-06,
+      "loss": 0.4136,
+      "step": 26334
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0238326872638387e-06,
+      "loss": 0.396,
+      "step": 26335
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0211483116669548e-06,
+      "loss": 0.3545,
+      "step": 26336
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.018465109837454e-06,
+      "loss": 0.4008,
+      "step": 26337
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.015783081807799e-06,
+      "loss": 0.4282,
+      "step": 26338
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.013102227610476e-06,
+      "loss": 0.4527,
+      "step": 26339
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0104225472778913e-06,
+      "loss": 0.4636,
+      "step": 26340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0077440408425196e-06,
+      "loss": 0.4081,
+      "step": 26341
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0050667083367678e-06,
+      "loss": 0.4126,
+      "step": 26342
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0023905497930327e-06,
+      "loss": 0.4876,
+      "step": 26343
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9997155652437103e-06,
+      "loss": 0.3797,
+      "step": 26344
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9970417547211527e-06,
+      "loss": 0.4215,
+      "step": 26345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.994369118257767e-06,
+      "loss": 0.4117,
+      "step": 26346
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9916976558858835e-06,
+      "loss": 0.4086,
+      "step": 26347
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.98902736763782e-06,
+      "loss": 0.3689,
+      "step": 26348
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9863582535459068e-06,
+      "loss": 0.388,
+      "step": 26349
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.98369031364244e-06,
+      "loss": 0.4972,
+      "step": 26350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.981023547959738e-06,
+      "loss": 0.4311,
+      "step": 26351
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9783579565300534e-06,
+      "loss": 0.42,
+      "step": 26352
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9756935393856487e-06,
+      "loss": 0.4602,
+      "step": 26353
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9730302965587876e-06,
+      "loss": 0.4311,
+      "step": 26354
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.970368228081688e-06,
+      "loss": 0.3921,
+      "step": 26355
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.96770733398658e-06,
+      "loss": 0.3512,
+      "step": 26356
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9650476143056604e-06,
+      "loss": 0.3938,
+      "step": 26357
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.962389069071125e-06,
+      "loss": 0.4977,
+      "step": 26358
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.959731698315149e-06,
+      "loss": 0.4214,
+      "step": 26359
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.957075502069906e-06,
+      "loss": 0.4147,
+      "step": 26360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9544204803675367e-06,
+      "loss": 0.449,
+      "step": 26361
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9517666332401718e-06,
+      "loss": 0.5277,
+      "step": 26362
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9491139607199404e-06,
+      "loss": 0.3468,
+      "step": 26363
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.946462462838939e-06,
+      "loss": 0.4496,
+      "step": 26364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.943812139629265e-06,
+      "loss": 0.3872,
+      "step": 26365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.941162991123003e-06,
+      "loss": 0.3378,
+      "step": 26366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9385150173522057e-06,
+      "loss": 0.433,
+      "step": 26367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.935868218348925e-06,
+      "loss": 0.4057,
+      "step": 26368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.933222594145191e-06,
+      "loss": 0.4438,
+      "step": 26369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.930578144773044e-06,
+      "loss": 0.3637,
+      "step": 26370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9279348702644704e-06,
+      "loss": 0.4184,
+      "step": 26371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9252927706514776e-06,
+      "loss": 0.398,
+      "step": 26372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.922651845966029e-06,
+      "loss": 0.3357,
+      "step": 26373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.920012096240099e-06,
+      "loss": 0.4205,
+      "step": 26374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9173735215056175e-06,
+      "loss": 0.4226,
+      "step": 26375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.914736121794559e-06,
+      "loss": 0.4929,
+      "step": 26376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.912099897138809e-06,
+      "loss": 0.4145,
+      "step": 26377
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.909464847570298e-06,
+      "loss": 0.381,
+      "step": 26378
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9068309731209e-06,
+      "loss": 0.3589,
+      "step": 26379
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9041982738225004e-06,
+      "loss": 0.3921,
+      "step": 26380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9015667497069742e-06,
+      "loss": 0.3502,
+      "step": 26381
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8989364008061515e-06,
+      "loss": 0.4711,
+      "step": 26382
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8963072271518952e-06,
+      "loss": 0.3606,
+      "step": 26383
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.893679228776003e-06,
+      "loss": 0.3716,
+      "step": 26384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8910524057102818e-06,
+      "loss": 0.5263,
+      "step": 26385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.888426757986551e-06,
+      "loss": 0.5222,
+      "step": 26386
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.885802285636574e-06,
+      "loss": 0.4098,
+      "step": 26387
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8831789886921036e-06,
+      "loss": 0.4098,
+      "step": 26388
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8805568671849136e-06,
+      "loss": 0.3949,
+      "step": 26389
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8779359211467126e-06,
+      "loss": 0.2961,
+      "step": 26390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8753161506092417e-06,
+      "loss": 0.3963,
+      "step": 26391
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.872697555604209e-06,
+      "loss": 0.4285,
+      "step": 26392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.870080136163311e-06,
+      "loss": 0.4436,
+      "step": 26393
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.867463892318223e-06,
+      "loss": 0.3989,
+      "step": 26394
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8648488241005853e-06,
+      "loss": 0.4428,
+      "step": 26395
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.862234931542096e-06,
+      "loss": 0.4548,
+      "step": 26396
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8596222146743623e-06,
+      "loss": 0.4354,
+      "step": 26397
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.857010673529015e-06,
+      "loss": 0.392,
+      "step": 26398
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.85440030813765e-06,
+      "loss": 0.39,
+      "step": 26399
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8517911185318657e-06,
+      "loss": 0.3764,
+      "step": 26400
+    },
+    {
+      "epoch": 0.92,
+      "eval_loss": 0.39770928025245667,
+      "eval_runtime": 853.6387,
+      "eval_samples_per_second": 28.185,
+      "eval_steps_per_second": 14.093,
+      "step": 26400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.849183104743258e-06,
+      "loss": 0.4153,
+      "step": 26401
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8465762668033913e-06,
+      "loss": 0.4171,
+      "step": 26402
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.843970604743795e-06,
+      "loss": 0.393,
+      "step": 26403
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.841366118596034e-06,
+      "loss": 0.3888,
+      "step": 26404
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.838762808391604e-06,
+      "loss": 0.4674,
+      "step": 26405
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.836160674162014e-06,
+      "loss": 0.4612,
+      "step": 26406
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8335597159387827e-06,
+      "loss": 0.4035,
+      "step": 26407
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8309599337533853e-06,
+      "loss": 0.4699,
+      "step": 26408
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8283613276372744e-06,
+      "loss": 0.413,
+      "step": 26409
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8257638976219135e-06,
+      "loss": 0.4315,
+      "step": 26410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8231676437387224e-06,
+      "loss": 0.4049,
+      "step": 26411
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.820572566019153e-06,
+      "loss": 0.3853,
+      "step": 26412
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8179786644945915e-06,
+      "loss": 0.3588,
+      "step": 26413
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8153859391964355e-06,
+      "loss": 0.39,
+      "step": 26414
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8127943901560707e-06,
+      "loss": 0.3889,
+      "step": 26415
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8102040174048604e-06,
+      "loss": 0.34,
+      "step": 26416
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.807614820974169e-06,
+      "loss": 0.3747,
+      "step": 26417
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8050268008953164e-06,
+      "loss": 0.4578,
+      "step": 26418
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8024399571996317e-06,
+      "loss": 0.4581,
+      "step": 26419
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7998542899184355e-06,
+      "loss": 0.4113,
+      "step": 26420
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.797269799083002e-06,
+      "loss": 0.3742,
+      "step": 26421
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.794686484724629e-06,
+      "loss": 0.5044,
+      "step": 26422
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.792104346874591e-06,
+      "loss": 0.4194,
+      "step": 26423
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.789523385564119e-06,
+      "loss": 0.3302,
+      "step": 26424
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.786943600824454e-06,
+      "loss": 0.4326,
+      "step": 26425
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7843649926868277e-06,
+      "loss": 0.3674,
+      "step": 26426
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7817875611824584e-06,
+      "loss": 0.395,
+      "step": 26427
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7792113063425216e-06,
+      "loss": 0.3967,
+      "step": 26428
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.776636228198215e-06,
+      "loss": 0.4488,
+      "step": 26429
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.774062326780702e-06,
+      "loss": 0.3294,
+      "step": 26430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7714896021211133e-06,
+      "loss": 0.4637,
+      "step": 26431
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7689180542506243e-06,
+      "loss": 0.4017,
+      "step": 26432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.766347683200343e-06,
+      "loss": 0.4347,
+      "step": 26433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7637784890013785e-06,
+      "loss": 0.4663,
+      "step": 26434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7612104716848273e-06,
+      "loss": 0.3963,
+      "step": 26435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.758643631281754e-06,
+      "loss": 0.4709,
+      "step": 26436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7560779678232563e-06,
+      "loss": 0.4056,
+      "step": 26437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.753513481340364e-06,
+      "loss": 0.4744,
+      "step": 26438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7509501718641305e-06,
+      "loss": 0.451,
+      "step": 26439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7483880394255758e-06,
+      "loss": 0.3601,
+      "step": 26440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7458270840557187e-06,
+      "loss": 0.396,
+      "step": 26441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.743267305785524e-06,
+      "loss": 0.4031,
+      "step": 26442
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.740708704646011e-06,
+      "loss": 0.3812,
+      "step": 26443
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.738151280668122e-06,
+      "loss": 0.447,
+      "step": 26444
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7355950338828317e-06,
+      "loss": 0.4019,
+      "step": 26445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7330399643210603e-06,
+      "loss": 0.55,
+      "step": 26446
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7304860720137273e-06,
+      "loss": 0.3671,
+      "step": 26447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7279333569917744e-06,
+      "loss": 0.4228,
+      "step": 26448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.725381819286077e-06,
+      "loss": 0.4229,
+      "step": 26449
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.722831458927522e-06,
+      "loss": 0.416,
+      "step": 26450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7202822759469614e-06,
+      "loss": 0.3872,
+      "step": 26451
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.717734270375272e-06,
+      "loss": 0.4523,
+      "step": 26452
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7151874422432834e-06,
+      "loss": 0.4192,
+      "step": 26453
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7126417915818157e-06,
+      "loss": 0.3888,
+      "step": 26454
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7100973184216893e-06,
+      "loss": 0.4115,
+      "step": 26455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7075540227937012e-06,
+      "loss": 0.3793,
+      "step": 26456
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.705011904728616e-06,
+      "loss": 0.4033,
+      "step": 26457
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.70247096425722e-06,
+      "loss": 0.4569,
+      "step": 26458
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.699931201410266e-06,
+      "loss": 0.4626,
+      "step": 26459
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.697392616218486e-06,
+      "loss": 0.4383,
+      "step": 26460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.69485520871261e-06,
+      "loss": 0.3735,
+      "step": 26461
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6923189789233473e-06,
+      "loss": 0.3765,
+      "step": 26462
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.689783926881395e-06,
+      "loss": 0.417,
+      "step": 26463
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6872500526174406e-06,
+      "loss": 0.444,
+      "step": 26464
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6847173561621363e-06,
+      "loss": 0.5035,
+      "step": 26465
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6821858375461585e-06,
+      "loss": 0.3591,
+      "step": 26466
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6796554968001262e-06,
+      "loss": 0.363,
+      "step": 26467
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6771263339546827e-06,
+      "loss": 0.3904,
+      "step": 26468
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6745983490404357e-06,
+      "loss": 0.3994,
+      "step": 26469
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6720715420879725e-06,
+      "loss": 0.4369,
+      "step": 26470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.66954591312788e-06,
+      "loss": 0.3968,
+      "step": 26471
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.667021462190733e-06,
+      "loss": 0.4443,
+      "step": 26472
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.664498189307074e-06,
+      "loss": 0.4771,
+      "step": 26473
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6619760945074456e-06,
+      "loss": 0.3299,
+      "step": 26474
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.65945517782239e-06,
+      "loss": 0.4108,
+      "step": 26475
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.656935439282404e-06,
+      "loss": 0.4313,
+      "step": 26476
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6544168789179868e-06,
+      "loss": 0.3838,
+      "step": 26477
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6518994967596135e-06,
+      "loss": 0.4936,
+      "step": 26478
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6493832928377706e-06,
+      "loss": 0.4159,
+      "step": 26479
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6468682671829005e-06,
+      "loss": 0.458,
+      "step": 26480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6443544198254565e-06,
+      "loss": 0.3535,
+      "step": 26481
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.641841750795837e-06,
+      "loss": 0.3813,
+      "step": 26482
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.639330260124473e-06,
+      "loss": 0.4261,
+      "step": 26483
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6368199478417732e-06,
+      "loss": 0.3322,
+      "step": 26484
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6343108139780915e-06,
+      "loss": 0.4384,
+      "step": 26485
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6318028585638253e-06,
+      "loss": 0.3215,
+      "step": 26486
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6292960816293176e-06,
+      "loss": 0.4806,
+      "step": 26487
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6267904832048883e-06,
+      "loss": 0.3677,
+      "step": 26488
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6242860633209132e-06,
+      "loss": 0.4052,
+      "step": 26489
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.621782822007657e-06,
+      "loss": 0.3407,
+      "step": 26490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6192807592954394e-06,
+      "loss": 0.3721,
+      "step": 26491
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6167798752145255e-06,
+      "loss": 0.3991,
+      "step": 26492
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6142801697952022e-06,
+      "loss": 0.378,
+      "step": 26493
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6117816430677234e-06,
+      "loss": 0.3737,
+      "step": 26494
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6092842950623196e-06,
+      "loss": 0.4745,
+      "step": 26495
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6067881258092343e-06,
+      "loss": 0.3786,
+      "step": 26496
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6042931353386535e-06,
+      "loss": 0.3814,
+      "step": 26497
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.601799323680798e-06,
+      "loss": 0.3514,
+      "step": 26498
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5993066908658436e-06,
+      "loss": 0.3651,
+      "step": 26499
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.596815236923955e-06,
+      "loss": 0.3482,
+      "step": 26500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5943249618852973e-06,
+      "loss": 0.354,
+      "step": 26501
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.59183586577999e-06,
+      "loss": 0.4731,
+      "step": 26502
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.589347948638188e-06,
+      "loss": 0.357,
+      "step": 26503
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.586861210489988e-06,
+      "loss": 0.4243,
+      "step": 26504
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5843756513654783e-06,
+      "loss": 0.473,
+      "step": 26505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5818912712947675e-06,
+      "loss": 0.3732,
+      "step": 26506
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5794080703079093e-06,
+      "loss": 0.4101,
+      "step": 26507
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5769260484349466e-06,
+      "loss": 0.4552,
+      "step": 26508
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.574445205705944e-06,
+      "loss": 0.327,
+      "step": 26509
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5719655421509115e-06,
+      "loss": 0.5254,
+      "step": 26510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5694870577998796e-06,
+      "loss": 0.3251,
+      "step": 26511
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.567009752682825e-06,
+      "loss": 0.4424,
+      "step": 26512
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.564533626829735e-06,
+      "loss": 0.4182,
+      "step": 26513
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5620586802705847e-06,
+      "loss": 0.3623,
+      "step": 26514
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.55958491303534e-06,
+      "loss": 0.3277,
+      "step": 26515
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.557112325153932e-06,
+      "loss": 0.3637,
+      "step": 26516
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5546409166562814e-06,
+      "loss": 0.4747,
+      "step": 26517
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5521706875722863e-06,
+      "loss": 0.3546,
+      "step": 26518
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5497016379318894e-06,
+      "loss": 0.3281,
+      "step": 26519
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.547233767764934e-06,
+      "loss": 0.4681,
+      "step": 26520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5447670771013175e-06,
+      "loss": 0.4127,
+      "step": 26521
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.542301565970873e-06,
+      "loss": 0.4241,
+      "step": 26522
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5398372344034418e-06,
+      "loss": 0.3834,
+      "step": 26523
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.537374082428867e-06,
+      "loss": 0.4821,
+      "step": 26524
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.534912110076948e-06,
+      "loss": 0.3842,
+      "step": 26525
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.532451317377493e-06,
+      "loss": 0.4505,
+      "step": 26526
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5299917043602794e-06,
+      "loss": 0.3953,
+      "step": 26527
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5275332710550716e-06,
+      "loss": 0.4044,
+      "step": 26528
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5250760174916454e-06,
+      "loss": 0.4291,
+      "step": 26529
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5226199436997113e-06,
+      "loss": 0.4204,
+      "step": 26530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5201650497090223e-06,
+      "loss": 0.4246,
+      "step": 26531
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5177113355492778e-06,
+      "loss": 0.4762,
+      "step": 26532
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5152588012501752e-06,
+      "loss": 0.356,
+      "step": 26533
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.512807446841403e-06,
+      "loss": 0.4145,
+      "step": 26534
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.510357272352626e-06,
+      "loss": 0.3962,
+      "step": 26535
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5079082778135087e-06,
+      "loss": 0.5014,
+      "step": 26536
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5054604632536836e-06,
+      "loss": 0.4011,
+      "step": 26537
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5030138287027826e-06,
+      "loss": 0.4234,
+      "step": 26538
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5005683741904152e-06,
+      "loss": 0.4576,
+      "step": 26539
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.49812409974618e-06,
+      "loss": 0.4172,
+      "step": 26540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4956810053996527e-06,
+      "loss": 0.3966,
+      "step": 26541
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4932390911804216e-06,
+      "loss": 0.489,
+      "step": 26542
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.490798357118029e-06,
+      "loss": 0.3628,
+      "step": 26543
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.488358803242008e-06,
+      "loss": 0.3937,
+      "step": 26544
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4859204295819117e-06,
+      "loss": 0.3493,
+      "step": 26545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.483483236167228e-06,
+      "loss": 0.3785,
+      "step": 26546
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4810472230274662e-06,
+      "loss": 0.3722,
+      "step": 26547
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4786123901921144e-06,
+      "loss": 0.4341,
+      "step": 26548
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4761787376906156e-06,
+      "loss": 0.4145,
+      "step": 26549
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4737462655524566e-06,
+      "loss": 0.3836,
+      "step": 26550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4713149738070707e-06,
+      "loss": 0.4039,
+      "step": 26551
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.468884862483878e-06,
+      "loss": 0.4483,
+      "step": 26552
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4664559316122883e-06,
+      "loss": 0.4167,
+      "step": 26553
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4640281812217004e-06,
+      "loss": 0.4488,
+      "step": 26554
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.461601611341513e-06,
+      "loss": 0.3821,
+      "step": 26555
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.459176222001092e-06,
+      "loss": 0.4139,
+      "step": 26556
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4567520132297795e-06,
+      "loss": 0.3964,
+      "step": 26557
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4543289850569197e-06,
+      "loss": 0.4321,
+      "step": 26558
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.451907137511833e-06,
+      "loss": 0.4349,
+      "step": 26559
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.449486470623863e-06,
+      "loss": 0.4506,
+      "step": 26560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.447066984422275e-06,
+      "loss": 0.4331,
+      "step": 26561
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4446486789363674e-06,
+      "loss": 0.397,
+      "step": 26562
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4422315541954067e-06,
+      "loss": 0.3915,
+      "step": 26563
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.439815610228635e-06,
+      "loss": 0.335,
+      "step": 26564
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4374008470653185e-06,
+      "loss": 0.3754,
+      "step": 26565
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.434987264734678e-06,
+      "loss": 0.4316,
+      "step": 26566
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.432574863265913e-06,
+      "loss": 0.4956,
+      "step": 26567
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4301636426882326e-06,
+      "loss": 0.3822,
+      "step": 26568
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.427753603030802e-06,
+      "loss": 0.3822,
+      "step": 26569
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4253447443228106e-06,
+      "loss": 0.4602,
+      "step": 26570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4229370665934225e-06,
+      "loss": 0.4367,
+      "step": 26571
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4205305698717483e-06,
+      "loss": 0.4075,
+      "step": 26572
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4181252541869424e-06,
+      "loss": 0.3973,
+      "step": 26573
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.415721119568104e-06,
+      "loss": 0.3687,
+      "step": 26574
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4133181660443094e-06,
+      "loss": 0.4202,
+      "step": 26575
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.410916393644691e-06,
+      "loss": 0.4632,
+      "step": 26576
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.408515802398281e-06,
+      "loss": 0.4234,
+      "step": 26577
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.406116392334157e-06,
+      "loss": 0.4514,
+      "step": 26578
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4037181634813386e-06,
+      "loss": 0.5041,
+      "step": 26579
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.40132111586886e-06,
+      "loss": 0.4226,
+      "step": 26580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3989252495257407e-06,
+      "loss": 0.4087,
+      "step": 26581
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.396530564480981e-06,
+      "loss": 0.3743,
+      "step": 26582
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3941370607635573e-06,
+      "loss": 0.4558,
+      "step": 26583
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.391744738402435e-06,
+      "loss": 0.3823,
+      "step": 26584
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3893535974265692e-06,
+      "loss": 0.3318,
+      "step": 26585
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.386963637864903e-06,
+      "loss": 0.3608,
+      "step": 26586
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.38457485974638e-06,
+      "loss": 0.39,
+      "step": 26587
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.382187263099889e-06,
+      "loss": 0.4969,
+      "step": 26588
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3798008479543388e-06,
+      "loss": 0.4235,
+      "step": 26589
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3774156143385963e-06,
+      "loss": 0.4662,
+      "step": 26590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.37503156228156e-06,
+      "loss": 0.3823,
+      "step": 26591
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.372648691812063e-06,
+      "loss": 0.3971,
+      "step": 26592
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.370267002958959e-06,
+      "loss": 0.4267,
+      "step": 26593
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3678864957510595e-06,
+      "loss": 0.3683,
+      "step": 26594
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3655071702171848e-06,
+      "loss": 0.4279,
+      "step": 26595
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.363129026386146e-06,
+      "loss": 0.3894,
+      "step": 26596
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3607520642866974e-06,
+      "loss": 0.4093,
+      "step": 26597
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3583762839476386e-06,
+      "loss": 0.3959,
+      "step": 26598
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.356001685397702e-06,
+      "loss": 0.4873,
+      "step": 26599
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.353628268665631e-06,
+      "loss": 0.4046,
+      "step": 26600
+    },
+    {
+      "epoch": 0.93,
+      "eval_loss": 0.39736002683639526,
+      "eval_runtime": 846.0117,
+      "eval_samples_per_second": 28.439,
+      "eval_steps_per_second": 14.22,
+      "step": 26600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3512560337801693e-06,
+      "loss": 0.3718,
+      "step": 26601
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3488849807700163e-06,
+      "loss": 0.3609,
+      "step": 26602
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3465151096638714e-06,
+      "loss": 0.4513,
+      "step": 26603
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3441464204904117e-06,
+      "loss": 0.3825,
+      "step": 26604
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.341778913278303e-06,
+      "loss": 0.3611,
+      "step": 26605
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.339412588056211e-06,
+      "loss": 0.3506,
+      "step": 26606
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.33704744485278e-06,
+      "loss": 0.4149,
+      "step": 26607
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3346834836966314e-06,
+      "loss": 0.4398,
+      "step": 26608
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.332320704616375e-06,
+      "loss": 0.414,
+      "step": 26609
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.329959107640611e-06,
+      "loss": 0.4997,
+      "step": 26610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3275986927979053e-06,
+      "loss": 0.3789,
+      "step": 26611
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.325239460116846e-06,
+      "loss": 0.3999,
+      "step": 26612
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.322881409625999e-06,
+      "loss": 0.38,
+      "step": 26613
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3205245413538747e-06,
+      "loss": 0.3922,
+      "step": 26614
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.318168855329017e-06,
+      "loss": 0.3235,
+      "step": 26615
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3158143515799258e-06,
+      "loss": 0.4574,
+      "step": 26616
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3134610301351113e-06,
+      "loss": 0.4735,
+      "step": 26617
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3111088910230507e-06,
+      "loss": 0.4451,
+      "step": 26618
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.308757934272221e-06,
+      "loss": 0.359,
+      "step": 26619
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3064081599110666e-06,
+      "loss": 0.4623,
+      "step": 26620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.30405956796802e-06,
+      "loss": 0.4554,
+      "step": 26621
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3017121584715252e-06,
+      "loss": 0.458,
+      "step": 26622
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2993659314499928e-06,
+      "loss": 0.4118,
+      "step": 26623
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.297020886931811e-06,
+      "loss": 0.4296,
+      "step": 26624
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2946770249453463e-06,
+      "loss": 0.4322,
+      "step": 26625
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2923343455190093e-06,
+      "loss": 0.4008,
+      "step": 26626
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2899928486811106e-06,
+      "loss": 0.3958,
+      "step": 26627
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2876525344600275e-06,
+      "loss": 0.4307,
+      "step": 26628
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2853134028840594e-06,
+      "loss": 0.4527,
+      "step": 26629
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2829754539815174e-06,
+      "loss": 0.4007,
+      "step": 26630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2806386877807116e-06,
+      "loss": 0.3767,
+      "step": 26631
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.278303104309931e-06,
+      "loss": 0.3745,
+      "step": 26632
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.27596870359742e-06,
+      "loss": 0.4037,
+      "step": 26633
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.273635485671455e-06,
+      "loss": 0.4382,
+      "step": 26634
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2713034505602583e-06,
+      "loss": 0.3276,
+      "step": 26635
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.268972598292074e-06,
+      "loss": 0.3536,
+      "step": 26636
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.266642928895091e-06,
+      "loss": 0.4157,
+      "step": 26637
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2643144423975305e-06,
+      "loss": 0.4266,
+      "step": 26638
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2619871388275595e-06,
+      "loss": 0.4084,
+      "step": 26639
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.259661018213333e-06,
+      "loss": 0.4138,
+      "step": 26640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.257336080583039e-06,
+      "loss": 0.4651,
+      "step": 26641
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2550123259647893e-06,
+      "loss": 0.3666,
+      "step": 26642
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2526897543867277e-06,
+      "loss": 0.3743,
+      "step": 26643
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2503683658769535e-06,
+      "loss": 0.3866,
+      "step": 26644
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2480481604635563e-06,
+      "loss": 0.4155,
+      "step": 26645
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.245729138174635e-06,
+      "loss": 0.3649,
+      "step": 26646
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.243411299038245e-06,
+      "loss": 0.4902,
+      "step": 26647
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.241094643082453e-06,
+      "loss": 0.3654,
+      "step": 26648
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.238779170335281e-06,
+      "loss": 0.3845,
+      "step": 26649
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2364648808247735e-06,
+      "loss": 0.4385,
+      "step": 26650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2341517745789185e-06,
+      "loss": 0.3831,
+      "step": 26651
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2318398516257276e-06,
+      "loss": 0.3595,
+      "step": 26652
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2295291119931896e-06,
+      "loss": 0.4161,
+      "step": 26653
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.227219555709259e-06,
+      "loss": 0.4488,
+      "step": 26654
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2249111828018922e-06,
+      "loss": 0.3583,
+      "step": 26655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2226039932990108e-06,
+      "loss": 0.3918,
+      "step": 26656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2202979872285813e-06,
+      "loss": 0.4177,
+      "step": 26657
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2179931646184814e-06,
+      "loss": 0.3334,
+      "step": 26658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2156895254966114e-06,
+      "loss": 0.4255,
+      "step": 26659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.213387069890871e-06,
+      "loss": 0.4274,
+      "step": 26660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.211085797829093e-06,
+      "loss": 0.3328,
+      "step": 26661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2087857093391563e-06,
+      "loss": 0.375,
+      "step": 26662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.206486804448904e-06,
+      "loss": 0.3737,
+      "step": 26663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2041890831861477e-06,
+      "loss": 0.3701,
+      "step": 26664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.20189254557871e-06,
+      "loss": 0.3878,
+      "step": 26665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.199597191654357e-06,
+      "loss": 0.5248,
+      "step": 26666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1973030214408997e-06,
+      "loss": 0.4126,
+      "step": 26667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.195010034966105e-06,
+      "loss": 0.362,
+      "step": 26668
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1927182322577067e-06,
+      "loss": 0.4137,
+      "step": 26669
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.19042761334346e-06,
+      "loss": 0.4464,
+      "step": 26670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.188138178251076e-06,
+      "loss": 0.3937,
+      "step": 26671
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1858499270082766e-06,
+      "loss": 0.3157,
+      "step": 26672
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1835628596427517e-06,
+      "loss": 0.462,
+      "step": 26673
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1812769761821783e-06,
+      "loss": 0.429,
+      "step": 26674
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.178992276654235e-06,
+      "loss": 0.3916,
+      "step": 26675
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1767087610865657e-06,
+      "loss": 0.399,
+      "step": 26676
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.174426429506793e-06,
+      "loss": 0.404,
+      "step": 26677
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1721452819425725e-06,
+      "loss": 0.4003,
+      "step": 26678
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1698653184214933e-06,
+      "loss": 0.3712,
+      "step": 26679
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1675865389711558e-06,
+      "loss": 0.4531,
+      "step": 26680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1653089436191487e-06,
+      "loss": 0.3393,
+      "step": 26681
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1630325323930167e-06,
+      "loss": 0.3987,
+      "step": 26682
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.160757305320338e-06,
+      "loss": 0.4442,
+      "step": 26683
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1584832624286345e-06,
+      "loss": 0.3994,
+      "step": 26684
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1562104037454287e-06,
+      "loss": 0.3929,
+      "step": 26685
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1539387292982327e-06,
+      "loss": 0.4077,
+      "step": 26686
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.151668239114546e-06,
+      "loss": 0.4717,
+      "step": 26687
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1493989332218468e-06,
+      "loss": 0.3732,
+      "step": 26688
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.147130811647602e-06,
+      "loss": 0.4113,
+      "step": 26689
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1448638744192674e-06,
+      "loss": 0.4122,
+      "step": 26690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1425981215642653e-06,
+      "loss": 0.3526,
+      "step": 26691
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1403335531100297e-06,
+      "loss": 0.4069,
+      "step": 26692
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1380701690839723e-06,
+      "loss": 0.4853,
+      "step": 26693
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1358079695134815e-06,
+      "loss": 0.49,
+      "step": 26694
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.133546954425947e-06,
+      "loss": 0.4099,
+      "step": 26695
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1312871238487243e-06,
+      "loss": 0.4032,
+      "step": 26696
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1290284778091584e-06,
+      "loss": 0.5276,
+      "step": 26697
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1267710163345945e-06,
+      "loss": 0.3536,
+      "step": 26698
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1245147394523655e-06,
+      "loss": 0.4075,
+      "step": 26699
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1222596471897725e-06,
+      "loss": 0.4163,
+      "step": 26700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1200057395741157e-06,
+      "loss": 0.4204,
+      "step": 26701
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1177530166326507e-06,
+      "loss": 0.3539,
+      "step": 26702
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1155014783926675e-06,
+      "loss": 0.4381,
+      "step": 26703
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.11325112488141e-06,
+      "loss": 0.4169,
+      "step": 26704
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.111001956126113e-06,
+      "loss": 0.4724,
+      "step": 26705
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1087539721539986e-06,
+      "loss": 0.4173,
+      "step": 26706
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.106507172992278e-06,
+      "loss": 0.4153,
+      "step": 26707
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1042615586681413e-06,
+      "loss": 0.3883,
+      "step": 26708
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1020171292087777e-06,
+      "loss": 0.4244,
+      "step": 26709
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0997738846413428e-06,
+      "loss": 0.4433,
+      "step": 26710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.097531824992982e-06,
+      "loss": 0.4004,
+      "step": 26711
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0952909502908513e-06,
+      "loss": 0.4318,
+      "step": 26712
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.09305126056204e-06,
+      "loss": 0.3733,
+      "step": 26713
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.090812755833693e-06,
+      "loss": 0.3931,
+      "step": 26714
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.088575436132889e-06,
+      "loss": 0.4804,
+      "step": 26715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.086339301486706e-06,
+      "loss": 0.3694,
+      "step": 26716
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0841043519222002e-06,
+      "loss": 0.4828,
+      "step": 26717
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.081870587466428e-06,
+      "loss": 0.4121,
+      "step": 26718
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.079638008146423e-06,
+      "loss": 0.4026,
+      "step": 26719
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.077406613989219e-06,
+      "loss": 0.342,
+      "step": 26720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.075176405021817e-06,
+      "loss": 0.4415,
+      "step": 26721
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0729473812712063e-06,
+      "loss": 0.3808,
+      "step": 26722
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0707195427643544e-06,
+      "loss": 0.4575,
+      "step": 26723
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.068492889528262e-06,
+      "loss": 0.4188,
+      "step": 26724
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.06626742158984e-06,
+      "loss": 0.4005,
+      "step": 26725
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0640431389760463e-06,
+      "loss": 0.4435,
+      "step": 26726
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0618200417137913e-06,
+      "loss": 0.4254,
+      "step": 26727
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.059598129829976e-06,
+      "loss": 0.4307,
+      "step": 26728
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0573774033515126e-06,
+      "loss": 0.4026,
+      "step": 26729
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.055157862305268e-06,
+      "loss": 0.3846,
+      "step": 26730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.052939506718121e-06,
+      "loss": 0.4527,
+      "step": 26731
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0507223366168948e-06,
+      "loss": 0.3889,
+      "step": 26732
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.048506352028423e-06,
+      "loss": 0.3859,
+      "step": 26733
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0462915529795623e-06,
+      "loss": 0.4164,
+      "step": 26734
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.044077939497091e-06,
+      "loss": 0.4322,
+      "step": 26735
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.04186551160781e-06,
+      "loss": 0.4521,
+      "step": 26736
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.039654269338498e-06,
+      "loss": 0.5636,
+      "step": 26737
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0374442127159e-06,
+      "loss": 0.3841,
+      "step": 26738
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0352353417667946e-06,
+      "loss": 0.399,
+      "step": 26739
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.033027656517894e-06,
+      "loss": 0.3855,
+      "step": 26740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0308211569959322e-06,
+      "loss": 0.4162,
+      "step": 26741
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0286158432276104e-06,
+      "loss": 0.3993,
+      "step": 26742
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.026411715239618e-06,
+      "loss": 0.3681,
+      "step": 26743
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0242087730586333e-06,
+      "loss": 0.3442,
+      "step": 26744
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0220070167113136e-06,
+      "loss": 0.3663,
+      "step": 26745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.019806446224326e-06,
+      "loss": 0.3279,
+      "step": 26746
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0176070616242824e-06,
+      "loss": 0.3945,
+      "step": 26747
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0154088629378175e-06,
+      "loss": 0.456,
+      "step": 26748
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.013211850191532e-06,
+      "loss": 0.4394,
+      "step": 26749
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0110160234120156e-06,
+      "loss": 0.3256,
+      "step": 26750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0088213826258472e-06,
+      "loss": 0.3855,
+      "step": 26751
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0066279278595946e-06,
+      "loss": 0.3802,
+      "step": 26752
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.004435659139792e-06,
+      "loss": 0.5087,
+      "step": 26753
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.002244576492973e-06,
+      "loss": 0.349,
+      "step": 26754
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0000546799456845e-06,
+      "loss": 0.3834,
+      "step": 26755
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.997865969524404e-06,
+      "loss": 0.4098,
+      "step": 26756
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.995678445255633e-06,
+      "loss": 0.4392,
+      "step": 26757
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9934921071658396e-06,
+      "loss": 0.425,
+      "step": 26758
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9913069552815023e-06,
+      "loss": 0.3574,
+      "step": 26759
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9891229896290444e-06,
+      "loss": 0.3661,
+      "step": 26760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.986940210234922e-06,
+      "loss": 0.4168,
+      "step": 26761
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9847586171255482e-06,
+      "loss": 0.4018,
+      "step": 26762
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.982578210327302e-06,
+      "loss": 0.3818,
+      "step": 26763
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9803989898666165e-06,
+      "loss": 0.3304,
+      "step": 26764
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9782209557698385e-06,
+      "loss": 0.4217,
+      "step": 26765
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9760441080633463e-06,
+      "loss": 0.47,
+      "step": 26766
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.973868446773475e-06,
+      "loss": 0.3395,
+      "step": 26767
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.971693971926547e-06,
+      "loss": 0.3801,
+      "step": 26768
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.969520683548909e-06,
+      "loss": 0.4786,
+      "step": 26769
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.96734858166685e-06,
+      "loss": 0.3702,
+      "step": 26770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.96517766630665e-06,
+      "loss": 0.4124,
+      "step": 26771
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9630079374946096e-06,
+      "loss": 0.3967,
+      "step": 26772
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9608393952569527e-06,
+      "loss": 0.4102,
+      "step": 26773
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9586720396199687e-06,
+      "loss": 0.3674,
+      "step": 26774
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.956505870609859e-06,
+      "loss": 0.4162,
+      "step": 26775
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.954340888252859e-06,
+      "loss": 0.4314,
+      "step": 26776
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9521770925751583e-06,
+      "loss": 0.3341,
+      "step": 26777
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9500144836029575e-06,
+      "loss": 0.3765,
+      "step": 26778
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.947853061362415e-06,
+      "loss": 0.5038,
+      "step": 26779
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.945692825879708e-06,
+      "loss": 0.462,
+      "step": 26780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9435337771809725e-06,
+      "loss": 0.4049,
+      "step": 26781
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.941375915292354e-06,
+      "loss": 0.3645,
+      "step": 26782
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9392192402399643e-06,
+      "loss": 0.4203,
+      "step": 26783
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9370637520498837e-06,
+      "loss": 0.3937,
+      "step": 26784
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9349094507482347e-06,
+      "loss": 0.4933,
+      "step": 26785
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9327563363610744e-06,
+      "loss": 0.3832,
+      "step": 26786
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9306044089144603e-06,
+      "loss": 0.4336,
+      "step": 26787
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.928453668434449e-06,
+      "loss": 0.3379,
+      "step": 26788
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9263041149470527e-06,
+      "loss": 0.4327,
+      "step": 26789
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.924155748478307e-06,
+      "loss": 0.3563,
+      "step": 26790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9220085690542123e-06,
+      "loss": 0.3978,
+      "step": 26791
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9198625767007485e-06,
+      "loss": 0.4124,
+      "step": 26792
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9177177714438945e-06,
+      "loss": 0.4488,
+      "step": 26793
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.915574153309596e-06,
+      "loss": 0.5099,
+      "step": 26794
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.913431722323822e-06,
+      "loss": 0.4152,
+      "step": 26795
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9112904785124954e-06,
+      "loss": 0.4366,
+      "step": 26796
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.909150421901518e-06,
+      "loss": 0.4381,
+      "step": 26797
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.907011552516813e-06,
+      "loss": 0.3459,
+      "step": 26798
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9048738703842383e-06,
+      "loss": 0.5306,
+      "step": 26799
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9027373755296952e-06,
+      "loss": 0.4342,
+      "step": 26800
+    },
+    {
+      "epoch": 0.94,
+      "eval_loss": 0.3971897065639496,
+      "eval_runtime": 853.8203,
+      "eval_samples_per_second": 28.179,
+      "eval_steps_per_second": 14.09,
+      "step": 26800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9006020679790293e-06,
+      "loss": 0.4637,
+      "step": 26801
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.898467947758098e-06,
+      "loss": 0.4236,
+      "step": 26802
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.896335014892714e-06,
+      "loss": 0.3987,
+      "step": 26803
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8942032694086897e-06,
+      "loss": 0.3362,
+      "step": 26804
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8920727113318494e-06,
+      "loss": 0.4442,
+      "step": 26805
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8899433406879608e-06,
+      "loss": 0.4574,
+      "step": 26806
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8878151575028036e-06,
+      "loss": 0.421,
+      "step": 26807
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.885688161802146e-06,
+      "loss": 0.3953,
+      "step": 26808
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.883562353611712e-06,
+      "loss": 0.4263,
+      "step": 26809
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8814377329572364e-06,
+      "loss": 0.459,
+      "step": 26810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8793142998644319e-06,
+      "loss": 0.3759,
+      "step": 26811
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8771920543590227e-06,
+      "loss": 0.3888,
+      "step": 26812
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8750709964666657e-06,
+      "loss": 0.3997,
+      "step": 26813
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8729511262130517e-06,
+      "loss": 0.4283,
+      "step": 26814
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8708324436238156e-06,
+      "loss": 0.3541,
+      "step": 26815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.868714948724626e-06,
+      "loss": 0.4791,
+      "step": 26816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8665986415411063e-06,
+      "loss": 0.3631,
+      "step": 26817
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8644835220988587e-06,
+      "loss": 0.4688,
+      "step": 26818
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8623695904234961e-06,
+      "loss": 0.467,
+      "step": 26819
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8602568465405868e-06,
+      "loss": 0.4419,
+      "step": 26820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8581452904757212e-06,
+      "loss": 0.3622,
+      "step": 26821
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8560349222544459e-06,
+      "loss": 0.4367,
+      "step": 26822
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.853925741902307e-06,
+      "loss": 0.3974,
+      "step": 26823
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8518177494448396e-06,
+      "loss": 0.3346,
+      "step": 26824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8497109449075345e-06,
+      "loss": 0.4048,
+      "step": 26825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8476053283159155e-06,
+      "loss": 0.3834,
+      "step": 26826
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8455008996954627e-06,
+      "loss": 0.3765,
+      "step": 26827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8433976590716328e-06,
+      "loss": 0.3252,
+      "step": 26828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.841295606469895e-06,
+      "loss": 0.388,
+      "step": 26829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8391947419156841e-06,
+      "loss": 0.4653,
+      "step": 26830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8370950654344355e-06,
+      "loss": 0.3053,
+      "step": 26831
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.834996577051562e-06,
+      "loss": 0.3695,
+      "step": 26832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8328992767924546e-06,
+      "loss": 0.4076,
+      "step": 26833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.830803164682493e-06,
+      "loss": 0.3335,
+      "step": 26834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8287082407470568e-06,
+      "loss": 0.4693,
+      "step": 26835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8266145050115036e-06,
+      "loss": 0.3855,
+      "step": 26836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8245219575011796e-06,
+      "loss": 0.3886,
+      "step": 26837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.822430598241398e-06,
+      "loss": 0.3891,
+      "step": 26838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.820340427257472e-06,
+      "loss": 0.403,
+      "step": 26839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8182514445746922e-06,
+      "loss": 0.3799,
+      "step": 26840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.816163650218372e-06,
+      "loss": 0.4427,
+      "step": 26841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8140770442137577e-06,
+      "loss": 0.4821,
+      "step": 26842
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.811991626586118e-06,
+      "loss": 0.4217,
+      "step": 26843
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8099073973606773e-06,
+      "loss": 0.459,
+      "step": 26844
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8078243565626596e-06,
+      "loss": 0.4041,
+      "step": 26845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8057425042172893e-06,
+      "loss": 0.4839,
+      "step": 26846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8036618403497685e-06,
+      "loss": 0.3338,
+      "step": 26847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.801582364985266e-06,
+      "loss": 0.4409,
+      "step": 26848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7995040781489614e-06,
+      "loss": 0.4743,
+      "step": 26849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7974269798660015e-06,
+      "loss": 0.4259,
+      "step": 26850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7953510701615217e-06,
+      "loss": 0.4746,
+      "step": 26851
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7932763490606574e-06,
+      "loss": 0.3977,
+      "step": 26852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.791202816588522e-06,
+      "loss": 0.4784,
+      "step": 26853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7891304727702064e-06,
+      "loss": 0.5273,
+      "step": 26854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7870593176307904e-06,
+      "loss": 0.3658,
+      "step": 26855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.784989351195332e-06,
+      "loss": 0.3889,
+      "step": 26856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.782920573488911e-06,
+      "loss": 0.3118,
+      "step": 26857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7808529845365629e-06,
+      "loss": 0.3691,
+      "step": 26858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.77878658436329e-06,
+      "loss": 0.4474,
+      "step": 26859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7767213729941167e-06,
+      "loss": 0.4414,
+      "step": 26860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7746573504540232e-06,
+      "loss": 0.4668,
+      "step": 26861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7725945167680226e-06,
+      "loss": 0.4423,
+      "step": 26862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7705328719610615e-06,
+      "loss": 0.3574,
+      "step": 26863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.768472416058098e-06,
+      "loss": 0.4273,
+      "step": 26864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7664131490840675e-06,
+      "loss": 0.4913,
+      "step": 26865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7643550710638834e-06,
+      "loss": 0.3985,
+      "step": 26866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7622981820224705e-06,
+      "loss": 0.401,
+      "step": 26867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7602424819847308e-06,
+      "loss": 0.3498,
+      "step": 26868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7581879709755333e-06,
+      "loss": 0.4666,
+      "step": 26869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.756134649019736e-06,
+      "loss": 0.4113,
+      "step": 26870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7540825161421969e-06,
+      "loss": 0.4025,
+      "step": 26871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7520315723677627e-06,
+      "loss": 0.4085,
+      "step": 26872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7499818177212578e-06,
+      "loss": 0.4021,
+      "step": 26873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.747933252227485e-06,
+      "loss": 0.409,
+      "step": 26874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.745885875911235e-06,
+      "loss": 0.4326,
+      "step": 26875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7438396887972885e-06,
+      "loss": 0.5001,
+      "step": 26876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7417946909104033e-06,
+      "loss": 0.3904,
+      "step": 26877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7397508822753594e-06,
+      "loss": 0.3815,
+      "step": 26878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7377082629168706e-06,
+      "loss": 0.3651,
+      "step": 26879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7356668328596615e-06,
+      "loss": 0.4828,
+      "step": 26880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7336265921284457e-06,
+      "loss": 0.4319,
+      "step": 26881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7315875407479032e-06,
+      "loss": 0.3513,
+      "step": 26882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7295496787427368e-06,
+      "loss": 0.469,
+      "step": 26883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7275130061375932e-06,
+      "loss": 0.4348,
+      "step": 26884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7254775229571307e-06,
+      "loss": 0.4146,
+      "step": 26885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.723443229225985e-06,
+      "loss": 0.4639,
+      "step": 26886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7214101249687698e-06,
+      "loss": 0.4052,
+      "step": 26887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7193782102100986e-06,
+      "loss": 0.5223,
+      "step": 26888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.717347484974574e-06,
+      "loss": 0.402,
+      "step": 26889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7153179492867654e-06,
+      "loss": 0.381,
+      "step": 26890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.713289603171231e-06,
+      "loss": 0.3762,
+      "step": 26891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7112624466525173e-06,
+      "loss": 0.3934,
+      "step": 26892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.709236479755194e-06,
+      "loss": 0.4631,
+      "step": 26893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7072117025037415e-06,
+      "loss": 0.3926,
+      "step": 26894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7051881149226845e-06,
+      "loss": 0.4059,
+      "step": 26895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7031657170364924e-06,
+      "loss": 0.458,
+      "step": 26896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7011445088696898e-06,
+      "loss": 0.3665,
+      "step": 26897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.699124490446702e-06,
+      "loss": 0.4257,
+      "step": 26898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6971056617919867e-06,
+      "loss": 0.4652,
+      "step": 26899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6950880229299804e-06,
+      "loss": 0.3736,
+      "step": 26900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6930715738851078e-06,
+      "loss": 0.3702,
+      "step": 26901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6910563146817714e-06,
+      "loss": 0.4008,
+      "step": 26902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6890422453443522e-06,
+      "loss": 0.4197,
+      "step": 26903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6870293658972525e-06,
+      "loss": 0.453,
+      "step": 26904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6850176763648085e-06,
+      "loss": 0.3736,
+      "step": 26905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6830071767713673e-06,
+      "loss": 0.4024,
+      "step": 26906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6809978671412873e-06,
+      "loss": 0.4214,
+      "step": 26907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6789897474988824e-06,
+      "loss": 0.4236,
+      "step": 26908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6769828178684333e-06,
+      "loss": 0.3446,
+      "step": 26909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6749770782742647e-06,
+      "loss": 0.4129,
+      "step": 26910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.672972528740624e-06,
+      "loss": 0.3829,
+      "step": 26911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6709691692917807e-06,
+      "loss": 0.4532,
+      "step": 26912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6689669999519819e-06,
+      "loss": 0.4552,
+      "step": 26913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.666966020745475e-06,
+      "loss": 0.3713,
+      "step": 26914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6649662316964633e-06,
+      "loss": 0.4264,
+      "step": 26915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6629676328291598e-06,
+      "loss": 0.3772,
+      "step": 26916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.660970224167746e-06,
+      "loss": 0.4567,
+      "step": 26917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6589740057363911e-06,
+      "loss": 0.3948,
+      "step": 26918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6569789775592758e-06,
+      "loss": 0.3937,
+      "step": 26919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6549851396605477e-06,
+      "loss": 0.3966,
+      "step": 26920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6529924920643093e-06,
+      "loss": 0.4272,
+      "step": 26921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.651001034794697e-06,
+      "loss": 0.382,
+      "step": 26922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6490107678758249e-06,
+      "loss": 0.3774,
+      "step": 26923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6470216913317626e-06,
+      "loss": 0.3466,
+      "step": 26924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.645033805186591e-06,
+      "loss": 0.4552,
+      "step": 26925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6430471094643795e-06,
+      "loss": 0.4397,
+      "step": 26926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6410616041891536e-06,
+      "loss": 0.4445,
+      "step": 26927
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6390772893849716e-06,
+      "loss": 0.4009,
+      "step": 26928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6370941650758254e-06,
+      "loss": 0.4212,
+      "step": 26929
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6351122312857402e-06,
+      "loss": 0.4385,
+      "step": 26930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.633131488038686e-06,
+      "loss": 0.3647,
+      "step": 26931
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6311519353586324e-06,
+      "loss": 0.3242,
+      "step": 26932
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.629173573269549e-06,
+      "loss": 0.3748,
+      "step": 26933
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6271964017953835e-06,
+      "loss": 0.4187,
+      "step": 26934
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.625220420960072e-06,
+      "loss": 0.4411,
+      "step": 26935
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6232456307875066e-06,
+      "loss": 0.4055,
+      "step": 26936
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6212720313016127e-06,
+      "loss": 0.427,
+      "step": 26937
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6192996225262603e-06,
+      "loss": 0.3943,
+      "step": 26938
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6173284044853298e-06,
+      "loss": 0.4396,
+      "step": 26939
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6153583772026915e-06,
+      "loss": 0.3641,
+      "step": 26940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6133895407021705e-06,
+      "loss": 0.4134,
+      "step": 26941
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611421895007592e-06,
+      "loss": 0.4489,
+      "step": 26942
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609455440142793e-06,
+      "loss": 0.3901,
+      "step": 26943
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6074901761315652e-06,
+      "loss": 0.4045,
+      "step": 26944
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6055261029976898e-06,
+      "loss": 0.4112,
+      "step": 26945
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6035632207649366e-06,
+      "loss": 0.3901,
+      "step": 26946
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6016015294570753e-06,
+      "loss": 0.3721,
+      "step": 26947
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5996410290978314e-06,
+      "loss": 0.3255,
+      "step": 26948
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.597681719710953e-06,
+      "loss": 0.4446,
+      "step": 26949
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5957236013201427e-06,
+      "loss": 0.3904,
+      "step": 26950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5937666739491042e-06,
+      "loss": 0.3854,
+      "step": 26951
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5918109376215184e-06,
+      "loss": 0.4504,
+      "step": 26952
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5898563923610554e-06,
+      "loss": 0.3336,
+      "step": 26953
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.587903038191374e-06,
+      "loss": 0.4351,
+      "step": 26954
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.585950875136133e-06,
+      "loss": 0.4047,
+      "step": 26955
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5839999032189357e-06,
+      "loss": 0.405,
+      "step": 26956
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5820501224634077e-06,
+      "loss": 0.3564,
+      "step": 26957
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5801015328931301e-06,
+      "loss": 0.3185,
+      "step": 26958
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5781541345317174e-06,
+      "loss": 0.4117,
+      "step": 26959
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5762079274027175e-06,
+      "loss": 0.3693,
+      "step": 26960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5742629115296892e-06,
+      "loss": 0.4043,
+      "step": 26961
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5723190869361914e-06,
+      "loss": 0.4505,
+      "step": 26962
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5703764536457166e-06,
+      "loss": 0.466,
+      "step": 26963
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5684350116818013e-06,
+      "loss": 0.397,
+      "step": 26964
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5664947610679493e-06,
+      "loss": 0.3823,
+      "step": 26965
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5645557018276303e-06,
+      "loss": 0.4353,
+      "step": 26966
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5626178339843144e-06,
+      "loss": 0.4552,
+      "step": 26967
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.560681157561461e-06,
+      "loss": 0.4546,
+      "step": 26968
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5587456725825068e-06,
+      "loss": 0.3698,
+      "step": 26969
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5568113790708771e-06,
+      "loss": 0.38,
+      "step": 26970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.554878277049987e-06,
+      "loss": 0.3512,
+      "step": 26971
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5529463665432398e-06,
+      "loss": 0.3687,
+      "step": 26972
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5510156475739946e-06,
+      "loss": 0.5283,
+      "step": 26973
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5490861201656436e-06,
+      "loss": 0.4317,
+      "step": 26974
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5471577843415353e-06,
+      "loss": 0.3459,
+      "step": 26975
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.545230640124995e-06,
+      "loss": 0.5018,
+      "step": 26976
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.543304687539371e-06,
+      "loss": 0.4503,
+      "step": 26977
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5413799266079554e-06,
+      "loss": 0.4054,
+      "step": 26978
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.539456357354041e-06,
+      "loss": 0.478,
+      "step": 26979
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5375339798009203e-06,
+      "loss": 0.4164,
+      "step": 26980
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5356127939718635e-06,
+      "loss": 0.4137,
+      "step": 26981
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5336927998901184e-06,
+      "loss": 0.4696,
+      "step": 26982
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5317739975789226e-06,
+      "loss": 0.4095,
+      "step": 26983
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5298563870615012e-06,
+      "loss": 0.4747,
+      "step": 26984
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5279399683610585e-06,
+      "loss": 0.3894,
+      "step": 26985
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5260247415007978e-06,
+      "loss": 0.4318,
+      "step": 26986
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5241107065039006e-06,
+      "loss": 0.4177,
+      "step": 26987
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5221978633935152e-06,
+      "loss": 0.4277,
+      "step": 26988
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5202862121928118e-06,
+      "loss": 0.387,
+      "step": 26989
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5183757529249276e-06,
+      "loss": 0.39,
+      "step": 26990
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5164664856129773e-06,
+      "loss": 0.436,
+      "step": 26991
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5145584102800646e-06,
+      "loss": 0.4162,
+      "step": 26992
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5126515269493046e-06,
+      "loss": 0.4131,
+      "step": 26993
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5107458356437564e-06,
+      "loss": 0.4404,
+      "step": 26994
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5088413363864906e-06,
+      "loss": 0.4509,
+      "step": 26995
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5069380292005663e-06,
+      "loss": 0.397,
+      "step": 26996
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5050359141090098e-06,
+      "loss": 0.3995,
+      "step": 26997
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5031349911348468e-06,
+      "loss": 0.434,
+      "step": 26998
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5012352603010816e-06,
+      "loss": 0.5158,
+      "step": 26999
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4993367216307176e-06,
+      "loss": 0.4209,
+      "step": 27000
+    },
+    {
+      "epoch": 0.94,
+      "eval_loss": 0.3968821167945862,
+      "eval_runtime": 844.9311,
+      "eval_samples_per_second": 28.476,
+      "eval_steps_per_second": 14.238,
+      "step": 27000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4974393751467252e-06,
+      "loss": 0.413,
+      "step": 27001
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4955432208720642e-06,
+      "loss": 0.422,
+      "step": 27002
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.493648258829694e-06,
+      "loss": 0.3675,
+      "step": 27003
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4917544890425406e-06,
+      "loss": 0.4044,
+      "step": 27004
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.489861911533541e-06,
+      "loss": 0.4135,
+      "step": 27005
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4879705263255772e-06,
+      "loss": 0.4393,
+      "step": 27006
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.486080333441564e-06,
+      "loss": 0.3751,
+      "step": 27007
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4841913329043722e-06,
+      "loss": 0.4493,
+      "step": 27008
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.48230352473685e-06,
+      "loss": 0.4853,
+      "step": 27009
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.480416908961868e-06,
+      "loss": 0.3949,
+      "step": 27010
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4785314856022526e-06,
+      "loss": 0.3719,
+      "step": 27011
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4766472546808297e-06,
+      "loss": 0.4611,
+      "step": 27012
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4747642162203923e-06,
+      "loss": 0.4369,
+      "step": 27013
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4728823702437334e-06,
+      "loss": 0.3331,
+      "step": 27014
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4710017167736234e-06,
+      "loss": 0.3737,
+      "step": 27015
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4691222558328443e-06,
+      "loss": 0.3333,
+      "step": 27016
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4672439874441336e-06,
+      "loss": 0.3875,
+      "step": 27017
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4653669116302282e-06,
+      "loss": 0.4035,
+      "step": 27018
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4634910284138437e-06,
+      "loss": 0.4043,
+      "step": 27019
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4616163378176729e-06,
+      "loss": 0.4024,
+      "step": 27020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4597428398644308e-06,
+      "loss": 0.3676,
+      "step": 27021
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4578705345767662e-06,
+      "loss": 0.4522,
+      "step": 27022
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4559994219773609e-06,
+      "loss": 0.3963,
+      "step": 27023
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4541295020888633e-06,
+      "loss": 0.3881,
+      "step": 27024
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4522607749338778e-06,
+      "loss": 0.3737,
+      "step": 27025
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4503932405350529e-06,
+      "loss": 0.3904,
+      "step": 27026
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4485268989149814e-06,
+      "loss": 0.3684,
+      "step": 27027
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4466617500962454e-06,
+      "loss": 0.415,
+      "step": 27028
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4447977941014269e-06,
+      "loss": 0.3902,
+      "step": 27029
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4429350309530853e-06,
+      "loss": 0.3841,
+      "step": 27030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4410734606737585e-06,
+      "loss": 0.4142,
+      "step": 27031
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4392130832859952e-06,
+      "loss": 0.3511,
+      "step": 27032
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4373538988122881e-06,
+      "loss": 0.3928,
+      "step": 27033
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4354959072751529e-06,
+      "loss": 0.4286,
+      "step": 27034
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4336391086970824e-06,
+      "loss": 0.4038,
+      "step": 27035
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4317835031005479e-06,
+      "loss": 0.4515,
+      "step": 27036
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4299290905080088e-06,
+      "loss": 0.3262,
+      "step": 27037
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4280758709418918e-06,
+      "loss": 0.4824,
+      "step": 27038
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4262238444246457e-06,
+      "loss": 0.4419,
+      "step": 27039
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4243730109786858e-06,
+      "loss": 0.3839,
+      "step": 27040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4225233706264052e-06,
+      "loss": 0.4052,
+      "step": 27041
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4206749233902084e-06,
+      "loss": 0.4279,
+      "step": 27042
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.418827669292444e-06,
+      "loss": 0.3585,
+      "step": 27043
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4169816083554721e-06,
+      "loss": 0.4005,
+      "step": 27044
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4151367406016525e-06,
+      "loss": 0.4125,
+      "step": 27045
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4132930660533118e-06,
+      "loss": 0.3829,
+      "step": 27046
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.411450584732754e-06,
+      "loss": 0.3714,
+      "step": 27047
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4096092966622843e-06,
+      "loss": 0.5235,
+      "step": 27048
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4077692018641842e-06,
+      "loss": 0.3987,
+      "step": 27049
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4059303003607248e-06,
+      "loss": 0.4304,
+      "step": 27050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4040925921741776e-06,
+      "loss": 0.3647,
+      "step": 27051
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.402256077326769e-06,
+      "loss": 0.3982,
+      "step": 27052
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4004207558407367e-06,
+      "loss": 0.447,
+      "step": 27053
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3985866277382965e-06,
+      "loss": 0.3906,
+      "step": 27054
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3967536930416191e-06,
+      "loss": 0.4055,
+      "step": 27055
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3949219517729316e-06,
+      "loss": 0.4584,
+      "step": 27056
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3930914039543718e-06,
+      "loss": 0.4496,
+      "step": 27057
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3912620496081218e-06,
+      "loss": 0.4065,
+      "step": 27058
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3894338887562974e-06,
+      "loss": 0.3565,
+      "step": 27059
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3876069214210252e-06,
+      "loss": 0.4169,
+      "step": 27060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3857811476244431e-06,
+      "loss": 0.4189,
+      "step": 27061
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3839565673886334e-06,
+      "loss": 0.4613,
+      "step": 27062
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3821331807356786e-06,
+      "loss": 0.417,
+      "step": 27063
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3803109876876608e-06,
+      "loss": 0.4044,
+      "step": 27064
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.378489988266607e-06,
+      "loss": 0.376,
+      "step": 27065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3766701824945772e-06,
+      "loss": 0.3169,
+      "step": 27066
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3748515703935982e-06,
+      "loss": 0.3775,
+      "step": 27067
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.373034151985686e-06,
+      "loss": 0.4154,
+      "step": 27068
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3712179272928227e-06,
+      "loss": 0.3925,
+      "step": 27069
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3694028963369799e-06,
+      "loss": 0.4283,
+      "step": 27070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.367589059140162e-06,
+      "loss": 0.3371,
+      "step": 27071
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3657764157242958e-06,
+      "loss": 0.5294,
+      "step": 27072
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3639649661113308e-06,
+      "loss": 0.402,
+      "step": 27073
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3621547103231936e-06,
+      "loss": 0.338,
+      "step": 27074
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3603456483817668e-06,
+      "loss": 0.4179,
+      "step": 27075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3585377803089883e-06,
+      "loss": 0.4505,
+      "step": 27076
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3567311061267075e-06,
+      "loss": 0.3784,
+      "step": 27077
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3549256258568177e-06,
+      "loss": 0.4063,
+      "step": 27078
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3531213395211461e-06,
+      "loss": 0.3714,
+      "step": 27079
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.351318247141542e-06,
+      "loss": 0.4097,
+      "step": 27080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.349516348739821e-06,
+      "loss": 0.4176,
+      "step": 27081
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.347715644337799e-06,
+      "loss": 0.4423,
+      "step": 27082
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3459161339572811e-06,
+      "loss": 0.4107,
+      "step": 27083
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3441178176200276e-06,
+      "loss": 0.3887,
+      "step": 27084
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3423206953478206e-06,
+      "loss": 0.4067,
+      "step": 27085
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3405247671623877e-06,
+      "loss": 0.3583,
+      "step": 27086
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3387300330855001e-06,
+      "loss": 0.4892,
+      "step": 27087
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3369364931388518e-06,
+      "loss": 0.3718,
+      "step": 27088
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3351441473441695e-06,
+      "loss": 0.409,
+      "step": 27089
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.333352995723125e-06,
+      "loss": 0.4789,
+      "step": 27090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.331563038297412e-06,
+      "loss": 0.4342,
+      "step": 27091
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.329774275088702e-06,
+      "loss": 0.2999,
+      "step": 27092
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3279867061186224e-06,
+      "loss": 0.4006,
+      "step": 27093
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3262003314088333e-06,
+      "loss": 0.441,
+      "step": 27094
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3244151509809398e-06,
+      "loss": 0.4933,
+      "step": 27095
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.322631164856547e-06,
+      "loss": 0.4179,
+      "step": 27096
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3208483730572596e-06,
+      "loss": 0.4464,
+      "step": 27097
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3190667756046492e-06,
+      "loss": 0.4521,
+      "step": 27098
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3172863725202879e-06,
+      "loss": 0.4688,
+      "step": 27099
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.315507163825702e-06,
+      "loss": 0.4071,
+      "step": 27100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.313729149542442e-06,
+      "loss": 0.3975,
+      "step": 27101
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3119523296920232e-06,
+      "loss": 0.3843,
+      "step": 27102
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3101767042959624e-06,
+      "loss": 0.4339,
+      "step": 27103
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3084022733757306e-06,
+      "loss": 0.4046,
+      "step": 27104
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3066290369528223e-06,
+      "loss": 0.372,
+      "step": 27105
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3048569950486755e-06,
+      "loss": 0.3987,
+      "step": 27106
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3030861476847733e-06,
+      "loss": 0.3876,
+      "step": 27107
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3013164948825207e-06,
+      "loss": 0.4541,
+      "step": 27108
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2995480366633562e-06,
+      "loss": 0.4692,
+      "step": 27109
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2977807730486624e-06,
+      "loss": 0.4435,
+      "step": 27110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2960147040598447e-06,
+      "loss": 0.4972,
+      "step": 27111
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2942498297182749e-06,
+      "loss": 0.39,
+      "step": 27112
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2924861500453024e-06,
+      "loss": 0.4528,
+      "step": 27113
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2907236650622988e-06,
+      "loss": 0.3415,
+      "step": 27114
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.28896237479057e-06,
+      "loss": 0.4431,
+      "step": 27115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2872022792514538e-06,
+      "loss": 0.3967,
+      "step": 27116
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2854433784662334e-06,
+      "loss": 0.4156,
+      "step": 27117
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.283685672456214e-06,
+      "loss": 0.4001,
+      "step": 27118
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2819291612426675e-06,
+      "loss": 0.361,
+      "step": 27119
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2801738448468547e-06,
+      "loss": 0.3786,
+      "step": 27120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.278419723290003e-06,
+      "loss": 0.4494,
+      "step": 27121
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.276666796593362e-06,
+      "loss": 0.3772,
+      "step": 27122
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.274915064778137e-06,
+      "loss": 0.4233,
+      "step": 27123
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2731645278655445e-06,
+      "loss": 0.3178,
+      "step": 27124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2714151858767565e-06,
+      "loss": 0.41,
+      "step": 27125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2696670388329556e-06,
+      "loss": 0.3757,
+      "step": 27126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.267920086755281e-06,
+      "loss": 0.3794,
+      "step": 27127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2661743296649041e-06,
+      "loss": 0.3437,
+      "step": 27128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.264429767582953e-06,
+      "loss": 0.3674,
+      "step": 27129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2626864005305217e-06,
+      "loss": 0.3524,
+      "step": 27130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2609442285287154e-06,
+      "loss": 0.4362,
+      "step": 27131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2592032515986285e-06,
+      "loss": 0.3715,
+      "step": 27132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2574634697613219e-06,
+      "loss": 0.424,
+      "step": 27133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2557248830378787e-06,
+      "loss": 0.3699,
+      "step": 27134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2539874914493154e-06,
+      "loss": 0.4117,
+      "step": 27135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.252251295016671e-06,
+      "loss": 0.472,
+      "step": 27136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2505162937609393e-06,
+      "loss": 0.4278,
+      "step": 27137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2487824877031594e-06,
+      "loss": 0.4068,
+      "step": 27138
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2470498768642813e-06,
+      "loss": 0.305,
+      "step": 27139
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2453184612652879e-06,
+      "loss": 0.3672,
+      "step": 27140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2435882409271404e-06,
+      "loss": 0.4377,
+      "step": 27141
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2418592158707665e-06,
+      "loss": 0.4077,
+      "step": 27142
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.240131386117116e-06,
+      "loss": 0.4363,
+      "step": 27143
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2384047516870723e-06,
+      "loss": 0.5078,
+      "step": 27144
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.236679312601563e-06,
+      "loss": 0.3977,
+      "step": 27145
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2349550688814493e-06,
+      "loss": 0.4058,
+      "step": 27146
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2332320205476034e-06,
+      "loss": 0.3617,
+      "step": 27147
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.231510167620875e-06,
+      "loss": 0.3854,
+      "step": 27148
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2297895101221258e-06,
+      "loss": 0.4818,
+      "step": 27149
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2280700480721718e-06,
+      "loss": 0.4556,
+      "step": 27150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.226351781491819e-06,
+      "loss": 0.438,
+      "step": 27151
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2246347104018729e-06,
+      "loss": 0.3476,
+      "step": 27152
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2229188348230946e-06,
+      "loss": 0.4028,
+      "step": 27153
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2212041547762677e-06,
+      "loss": 0.4163,
+      "step": 27154
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.219490670282153e-06,
+      "loss": 0.3941,
+      "step": 27155
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2177783813614896e-06,
+      "loss": 0.4276,
+      "step": 27156
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2160672880349833e-06,
+      "loss": 0.4258,
+      "step": 27157
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2143573903233508e-06,
+      "loss": 0.3875,
+      "step": 27158
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2126486882472865e-06,
+      "loss": 0.3821,
+      "step": 27159
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2109411818274852e-06,
+      "loss": 0.3675,
+      "step": 27160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.209234871084608e-06,
+      "loss": 0.3635,
+      "step": 27161
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2075297560392939e-06,
+      "loss": 0.3422,
+      "step": 27162
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2058258367121932e-06,
+      "loss": 0.3493,
+      "step": 27163
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.204123113123934e-06,
+      "loss": 0.4889,
+      "step": 27164
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2024215852950993e-06,
+      "loss": 0.5068,
+      "step": 27165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2007212532463175e-06,
+      "loss": 0.3296,
+      "step": 27166
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1990221169981276e-06,
+      "loss": 0.3669,
+      "step": 27167
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1973241765711352e-06,
+      "loss": 0.4525,
+      "step": 27168
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1956274319858796e-06,
+      "loss": 0.3844,
+      "step": 27169
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1939318832628776e-06,
+      "loss": 0.4012,
+      "step": 27170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1922375304226686e-06,
+      "loss": 0.3566,
+      "step": 27171
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.190544373485758e-06,
+      "loss": 0.4707,
+      "step": 27172
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1888524124726297e-06,
+      "loss": 0.4147,
+      "step": 27173
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1871616474037784e-06,
+      "loss": 0.5104,
+      "step": 27174
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1854720782996653e-06,
+      "loss": 0.4001,
+      "step": 27175
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1837837051807188e-06,
+      "loss": 0.3941,
+      "step": 27176
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.182096528067389e-06,
+      "loss": 0.5019,
+      "step": 27177
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.180410546980104e-06,
+      "loss": 0.3698,
+      "step": 27178
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.178725761939259e-06,
+      "loss": 0.3259,
+      "step": 27179
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.177042172965248e-06,
+      "loss": 0.4149,
+      "step": 27180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1753597800784556e-06,
+      "loss": 0.3257,
+      "step": 27181
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1736785832992203e-06,
+      "loss": 0.4483,
+      "step": 27182
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.171998582647915e-06,
+      "loss": 0.3594,
+      "step": 27183
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1703197781448683e-06,
+      "loss": 0.4048,
+      "step": 27184
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1686421698103965e-06,
+      "loss": 0.4272,
+      "step": 27185
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1669657576647953e-06,
+      "loss": 0.4665,
+      "step": 27186
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.16529054172837e-06,
+      "loss": 0.4097,
+      "step": 27187
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1636165220213825e-06,
+      "loss": 0.4652,
+      "step": 27188
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1619436985641053e-06,
+      "loss": 0.3809,
+      "step": 27189
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.160272071376789e-06,
+      "loss": 0.3659,
+      "step": 27190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1586016404796506e-06,
+      "loss": 0.4542,
+      "step": 27191
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1569324058929186e-06,
+      "loss": 0.3894,
+      "step": 27192
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.155264367636788e-06,
+      "loss": 0.3359,
+      "step": 27193
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1535975257314535e-06,
+      "loss": 0.5168,
+      "step": 27194
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.151931880197088e-06,
+      "loss": 0.4198,
+      "step": 27195
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.150267431053853e-06,
+      "loss": 0.4147,
+      "step": 27196
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1486041783218992e-06,
+      "loss": 0.4489,
+      "step": 27197
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1469421220213438e-06,
+      "loss": 0.3523,
+      "step": 27198
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1452812621723042e-06,
+      "loss": 0.4608,
+      "step": 27199
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1436215987948974e-06,
+      "loss": 0.4549,
+      "step": 27200
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 0.39672136306762695,
+      "eval_runtime": 854.184,
+      "eval_samples_per_second": 28.167,
+      "eval_steps_per_second": 14.084,
+      "step": 27200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1419631319092073e-06,
+      "loss": 0.4339,
+      "step": 27201
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1403058615352957e-06,
+      "loss": 0.409,
+      "step": 27202
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1386497876932245e-06,
+      "loss": 0.3843,
+      "step": 27203
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1369949104030442e-06,
+      "loss": 0.3224,
+      "step": 27204
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.135341229684772e-06,
+      "loss": 0.3727,
+      "step": 27205
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1336887455584477e-06,
+      "loss": 0.3572,
+      "step": 27206
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.132037458044044e-06,
+      "loss": 0.3683,
+      "step": 27207
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1303873671615561e-06,
+      "loss": 0.3799,
+      "step": 27208
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1287384729309569e-06,
+      "loss": 0.3955,
+      "step": 27209
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1270907753722082e-06,
+      "loss": 0.4257,
+      "step": 27210
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1254442745052606e-06,
+      "loss": 0.3911,
+      "step": 27211
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1237989703500207e-06,
+      "loss": 0.5291,
+      "step": 27212
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1221548629264056e-06,
+      "loss": 0.4235,
+      "step": 27213
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1205119522543217e-06,
+      "loss": 0.3898,
+      "step": 27214
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1188702383536642e-06,
+      "loss": 0.486,
+      "step": 27215
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1172297212442839e-06,
+      "loss": 0.5193,
+      "step": 27216
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.115590400946054e-06,
+      "loss": 0.357,
+      "step": 27217
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1139522774788026e-06,
+      "loss": 0.4537,
+      "step": 27218
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1123153508623474e-06,
+      "loss": 0.3997,
+      "step": 27219
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1106796211165283e-06,
+      "loss": 0.4353,
+      "step": 27220
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.109045088261118e-06,
+      "loss": 0.4372,
+      "step": 27221
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1074117523159232e-06,
+      "loss": 0.4535,
+      "step": 27222
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1057796133006947e-06,
+      "loss": 0.3482,
+      "step": 27223
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1041486712351834e-06,
+      "loss": 0.4059,
+      "step": 27224
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.10251892613914e-06,
+      "loss": 0.3693,
+      "step": 27225
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1008903780322933e-06,
+      "loss": 0.4074,
+      "step": 27226
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0992630269343495e-06,
+      "loss": 0.4281,
+      "step": 27227
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0976368728650045e-06,
+      "loss": 0.3617,
+      "step": 27228
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0960119158439308e-06,
+      "loss": 0.3461,
+      "step": 27229
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.094388155890802e-06,
+      "loss": 0.4901,
+      "step": 27230
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.092765593025291e-06,
+      "loss": 0.3834,
+      "step": 27231
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0911442272670047e-06,
+      "loss": 0.4117,
+      "step": 27232
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0895240586355825e-06,
+      "loss": 0.3527,
+      "step": 27233
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0879050871506314e-06,
+      "loss": 0.3999,
+      "step": 27234
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0862873128317463e-06,
+      "loss": 0.404,
+      "step": 27235
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0846707356985119e-06,
+      "loss": 0.4523,
+      "step": 27236
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0830553557705014e-06,
+      "loss": 0.405,
+      "step": 27237
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0814411730672435e-06,
+      "loss": 0.4568,
+      "step": 27238
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0798281876082784e-06,
+      "loss": 0.4406,
+      "step": 27239
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0782163994131566e-06,
+      "loss": 0.4321,
+      "step": 27240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0766058085013631e-06,
+      "loss": 0.3506,
+      "step": 27241
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.074996414892393e-06,
+      "loss": 0.3974,
+      "step": 27242
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0733882186057198e-06,
+      "loss": 0.4174,
+      "step": 27243
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0717812196608168e-06,
+      "loss": 0.352,
+      "step": 27244
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0701754180771462e-06,
+      "loss": 0.3871,
+      "step": 27245
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0685708138741256e-06,
+      "loss": 0.383,
+      "step": 27246
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0669674070711844e-06,
+      "loss": 0.3904,
+      "step": 27247
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0653651976877178e-06,
+      "loss": 0.3339,
+      "step": 27248
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0637641857431324e-06,
+      "loss": 0.4302,
+      "step": 27249
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0621643712567907e-06,
+      "loss": 0.3717,
+      "step": 27250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0605657542480662e-06,
+      "loss": 0.3605,
+      "step": 27251
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.05896833473631e-06,
+      "loss": 0.3934,
+      "step": 27252
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0573721127408509e-06,
+      "loss": 0.4555,
+      "step": 27253
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.055777088281018e-06,
+      "loss": 0.3623,
+      "step": 27254
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.054183261376085e-06,
+      "loss": 0.4324,
+      "step": 27255
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0525906320453916e-06,
+      "loss": 0.4924,
+      "step": 27256
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.050999200308167e-06,
+      "loss": 0.4024,
+      "step": 27257
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.049408966183707e-06,
+      "loss": 0.4581,
+      "step": 27258
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0478199296912406e-06,
+      "loss": 0.3537,
+      "step": 27259
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.046232090850008e-06,
+      "loss": 0.3797,
+      "step": 27260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0446454496792157e-06,
+      "loss": 0.4789,
+      "step": 27261
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0430600061980933e-06,
+      "loss": 0.3799,
+      "step": 27262
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0414757604258029e-06,
+      "loss": 0.3955,
+      "step": 27263
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0398927123815295e-06,
+      "loss": 0.4155,
+      "step": 27264
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0383108620844239e-06,
+      "loss": 0.3984,
+      "step": 27265
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.036730209553649e-06,
+      "loss": 0.3641,
+      "step": 27266
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0351507548083228e-06,
+      "loss": 0.4752,
+      "step": 27267
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0335724978675744e-06,
+      "loss": 0.3802,
+      "step": 27268
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0319954387504994e-06,
+      "loss": 0.4004,
+      "step": 27269
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0304195774761715e-06,
+      "loss": 0.434,
+      "step": 27270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0288449140636868e-06,
+      "loss": 0.3575,
+      "step": 27271
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0272714485320966e-06,
+      "loss": 0.3916,
+      "step": 27272
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0256991809004413e-06,
+      "loss": 0.459,
+      "step": 27273
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0241281111877498e-06,
+      "loss": 0.4348,
+      "step": 27274
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0225582394130406e-06,
+      "loss": 0.3466,
+      "step": 27275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0209895655953094e-06,
+      "loss": 0.4458,
+      "step": 27276
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0194220897535522e-06,
+      "loss": 0.3677,
+      "step": 27277
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0178558119067315e-06,
+      "loss": 0.3737,
+      "step": 27278
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0162907320737992e-06,
+      "loss": 0.4246,
+      "step": 27279
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0147268502737172e-06,
+      "loss": 0.4007,
+      "step": 27280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0131641665253933e-06,
+      "loss": 0.4464,
+      "step": 27281
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0116026808477563e-06,
+      "loss": 0.3762,
+      "step": 27282
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0100423932597024e-06,
+      "loss": 0.347,
+      "step": 27283
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0084833037801168e-06,
+      "loss": 0.3831,
+      "step": 27284
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0069254124278614e-06,
+      "loss": 0.3777,
+      "step": 27285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0053687192217887e-06,
+      "loss": 0.3495,
+      "step": 27286
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0038132241807608e-06,
+      "loss": 0.4199,
+      "step": 27287
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.002258927323585e-06,
+      "loss": 0.3959,
+      "step": 27288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0007058286690796e-06,
+      "loss": 0.4313,
+      "step": 27289
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.991539282360518e-07,
+      "loss": 0.3981,
+      "step": 27290
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.976032260432644e-07,
+      "loss": 0.3528,
+      "step": 27291
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.960537221095022e-07,
+      "loss": 0.3343,
+      "step": 27292
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.945054164535172e-07,
+      "loss": 0.3543,
+      "step": 27293
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.929583090940386e-07,
+      "loss": 0.3754,
+      "step": 27294
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.91412400049807e-07,
+      "loss": 0.4704,
+      "step": 27295
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.898676893395186e-07,
+      "loss": 0.3997,
+      "step": 27296
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.883241769818918e-07,
+      "loss": 0.366,
+      "step": 27297
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.867818629955783e-07,
+      "loss": 0.3693,
+      "step": 27298
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.852407473992632e-07,
+      "loss": 0.4165,
+      "step": 27299
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.83700830211598e-07,
+      "loss": 0.6024,
+      "step": 27300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.821621114512125e-07,
+      "loss": 0.4648,
+      "step": 27301
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.806245911367473e-07,
+      "loss": 0.4239,
+      "step": 27302
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.790882692867986e-07,
+      "loss": 0.4538,
+      "step": 27303
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.775531459199516e-07,
+      "loss": 0.4231,
+      "step": 27304
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.760192210548024e-07,
+      "loss": 0.3957,
+      "step": 27305
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.74486494709914e-07,
+      "loss": 0.444,
+      "step": 27306
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.729549669038385e-07,
+      "loss": 0.434,
+      "step": 27307
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.71424637655116e-07,
+      "loss": 0.3663,
+      "step": 27308
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.698955069822546e-07,
+      "loss": 0.4235,
+      "step": 27309
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.683675749037613e-07,
+      "loss": 0.3859,
+      "step": 27310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.668408414381435e-07,
+      "loss": 0.3718,
+      "step": 27311
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.653153066038756e-07,
+      "loss": 0.3913,
+      "step": 27312
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.637909704194203e-07,
+      "loss": 0.416,
+      "step": 27313
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.622678329032298e-07,
+      "loss": 0.4027,
+      "step": 27314
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.607458940737336e-07,
+      "loss": 0.4393,
+      "step": 27315
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.592251539493502e-07,
+      "loss": 0.3805,
+      "step": 27316
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.577056125484873e-07,
+      "loss": 0.4337,
+      "step": 27317
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.561872698895412e-07,
+      "loss": 0.394,
+      "step": 27318
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.546701259908974e-07,
+      "loss": 0.4541,
+      "step": 27319
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.531541808708966e-07,
+      "loss": 0.4681,
+      "step": 27320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.516394345478907e-07,
+      "loss": 0.397,
+      "step": 27321
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.501258870402318e-07,
+      "loss": 0.3764,
+      "step": 27322
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.486135383662276e-07,
+      "loss": 0.4698,
+      "step": 27323
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.471023885441854e-07,
+      "loss": 0.4253,
+      "step": 27324
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.455924375923908e-07,
+      "loss": 0.3691,
+      "step": 27325
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.440836855291069e-07,
+      "loss": 0.3892,
+      "step": 27326
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.42576132372619e-07,
+      "loss": 0.4523,
+      "step": 27327
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.410697781411681e-07,
+      "loss": 0.3782,
+      "step": 27328
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.39564622852973e-07,
+      "loss": 0.3801,
+      "step": 27329
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.380606665262637e-07,
+      "loss": 0.3614,
+      "step": 27330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.365579091792365e-07,
+      "loss": 0.463,
+      "step": 27331
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.350563508300769e-07,
+      "loss": 0.4008,
+      "step": 27332
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.335559914969705e-07,
+      "loss": 0.3678,
+      "step": 27333
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.320568311980693e-07,
+      "loss": 0.479,
+      "step": 27334
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.305588699515144e-07,
+      "loss": 0.4108,
+      "step": 27335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.290621077754358e-07,
+      "loss": 0.4312,
+      "step": 27336
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.275665446879633e-07,
+      "loss": 0.4631,
+      "step": 27337
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.260721807071937e-07,
+      "loss": 0.4899,
+      "step": 27338
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.245790158512013e-07,
+      "loss": 0.341,
+      "step": 27339
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.230870501380606e-07,
+      "loss": 0.3893,
+      "step": 27340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.215962835858349e-07,
+      "loss": 0.4086,
+      "step": 27341
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.201067162125654e-07,
+      "loss": 0.4044,
+      "step": 27342
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.186183480362931e-07,
+      "loss": 0.3926,
+      "step": 27343
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.171311790750148e-07,
+      "loss": 0.3854,
+      "step": 27344
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.156452093467383e-07,
+      "loss": 0.3297,
+      "step": 27345
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.141604388694492e-07,
+      "loss": 0.4429,
+      "step": 27346
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.126768676611108e-07,
+      "loss": 0.4153,
+      "step": 27347
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.111944957396979e-07,
+      "loss": 0.3531,
+      "step": 27348
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.097133231231292e-07,
+      "loss": 0.4395,
+      "step": 27349
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.08233349829346e-07,
+      "loss": 0.4267,
+      "step": 27350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.067545758762563e-07,
+      "loss": 0.4053,
+      "step": 27351
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.052770012817458e-07,
+      "loss": 0.431,
+      "step": 27352
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.038006260637222e-07,
+      "loss": 0.4295,
+      "step": 27353
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.02325450240038e-07,
+      "loss": 0.3514,
+      "step": 27354
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.008514738285678e-07,
+      "loss": 0.4118,
+      "step": 27355
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.993786968471196e-07,
+      "loss": 0.4044,
+      "step": 27356
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.979071193135458e-07,
+      "loss": 0.5135,
+      "step": 27357
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.964367412456431e-07,
+      "loss": 0.4489,
+      "step": 27358
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.949675626612086e-07,
+      "loss": 0.4204,
+      "step": 27359
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.934995835780391e-07,
+      "loss": 0.4724,
+      "step": 27360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.92032804013887e-07,
+      "loss": 0.3431,
+      "step": 27361
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.905672239865049e-07,
+      "loss": 0.3876,
+      "step": 27362
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.891028435136339e-07,
+      "loss": 0.3829,
+      "step": 27363
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.876396626130157e-07,
+      "loss": 0.4708,
+      "step": 27364
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.861776813023248e-07,
+      "loss": 0.4174,
+      "step": 27365
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.847168995992916e-07,
+      "loss": 0.4221,
+      "step": 27366
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.832573175215575e-07,
+      "loss": 0.4053,
+      "step": 27367
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.817989350868305e-07,
+      "loss": 0.4023,
+      "step": 27368
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.803417523127411e-07,
+      "loss": 0.3939,
+      "step": 27369
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.788857692169194e-07,
+      "loss": 0.332,
+      "step": 27370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.77430985816996e-07,
+      "loss": 0.4607,
+      "step": 27371
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.759774021305788e-07,
+      "loss": 0.4642,
+      "step": 27372
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.74525018175254e-07,
+      "loss": 0.446,
+      "step": 27373
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.730738339686184e-07,
+      "loss": 0.3728,
+      "step": 27374
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.716238495282136e-07,
+      "loss": 0.4653,
+      "step": 27375
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.701750648716145e-07,
+      "loss": 0.4361,
+      "step": 27376
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.687274800163181e-07,
+      "loss": 0.4122,
+      "step": 27377
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.672810949798882e-07,
+      "loss": 0.4115,
+      "step": 27378
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.658359097797997e-07,
+      "loss": 0.3779,
+      "step": 27379
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.643919244335608e-07,
+      "loss": 0.4567,
+      "step": 27380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.629491389586464e-07,
+      "loss": 0.4315,
+      "step": 27381
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.615075533725092e-07,
+      "loss": 0.4391,
+      "step": 27382
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.600671676926019e-07,
+      "loss": 0.3948,
+      "step": 27383
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.586279819363663e-07,
+      "loss": 0.4531,
+      "step": 27384
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.571899961212215e-07,
+      "loss": 0.3728,
+      "step": 27385
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.557532102645538e-07,
+      "loss": 0.3495,
+      "step": 27386
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.543176243837714e-07,
+      "loss": 0.4124,
+      "step": 27387
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.52883238496227e-07,
+      "loss": 0.4332,
+      "step": 27388
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.514500526193181e-07,
+      "loss": 0.4305,
+      "step": 27389
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.500180667703639e-07,
+      "loss": 0.4041,
+      "step": 27390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.48587280966695e-07,
+      "loss": 0.4821,
+      "step": 27391
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.471576952256421e-07,
+      "loss": 0.448,
+      "step": 27392
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.457293095645025e-07,
+      "loss": 0.4138,
+      "step": 27393
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.443021240005622e-07,
+      "loss": 0.4522,
+      "step": 27394
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.428761385510964e-07,
+      "loss": 0.3653,
+      "step": 27395
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.41451353233369e-07,
+      "loss": 0.4394,
+      "step": 27396
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.400277680646218e-07,
+      "loss": 0.4055,
+      "step": 27397
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.386053830620743e-07,
+      "loss": 0.3765,
+      "step": 27398
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.371841982429574e-07,
+      "loss": 0.3887,
+      "step": 27399
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.35764213624457e-07,
+      "loss": 0.3316,
+      "step": 27400
+    },
+    {
+      "epoch": 0.96,
+      "eval_loss": 0.3967449367046356,
+      "eval_runtime": 844.3903,
+      "eval_samples_per_second": 28.494,
+      "eval_steps_per_second": 14.247,
+      "step": 27400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.343454292237818e-07,
+      "loss": 0.4224,
+      "step": 27401
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.329278450580846e-07,
+      "loss": 0.3613,
+      "step": 27402
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.315114611445296e-07,
+      "loss": 0.4495,
+      "step": 27403
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.300962775002586e-07,
+      "loss": 0.4127,
+      "step": 27404
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.286822941424022e-07,
+      "loss": 0.4376,
+      "step": 27405
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.27269511088069e-07,
+      "loss": 0.4526,
+      "step": 27406
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.258579283543677e-07,
+      "loss": 0.4702,
+      "step": 27407
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.244475459583623e-07,
+      "loss": 0.4463,
+      "step": 27408
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.230383639171502e-07,
+      "loss": 0.3756,
+      "step": 27409
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.216303822477844e-07,
+      "loss": 0.3895,
+      "step": 27410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.202236009672848e-07,
+      "loss": 0.3922,
+      "step": 27411
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.188180200926932e-07,
+      "loss": 0.3939,
+      "step": 27412
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.174136396410182e-07,
+      "loss": 0.456,
+      "step": 27413
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.160104596292572e-07,
+      "loss": 0.4786,
+      "step": 27414
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.14608480074397e-07,
+      "loss": 0.3665,
+      "step": 27415
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.132077009934125e-07,
+      "loss": 0.3793,
+      "step": 27416
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.118081224032348e-07,
+      "loss": 0.4077,
+      "step": 27417
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.10409744320828e-07,
+      "loss": 0.3664,
+      "step": 27418
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.090125667630899e-07,
+      "loss": 0.4286,
+      "step": 27419
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.076165897469623e-07,
+      "loss": 0.3052,
+      "step": 27420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.062218132893318e-07,
+      "loss": 0.4513,
+      "step": 27421
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.048282374070626e-07,
+      "loss": 0.421,
+      "step": 27422
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.034358621170413e-07,
+      "loss": 0.3786,
+      "step": 27423
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.020446874361099e-07,
+      "loss": 0.3718,
+      "step": 27424
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.006547133811105e-07,
+      "loss": 0.3809,
+      "step": 27425
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.992659399688629e-07,
+      "loss": 0.3798,
+      "step": 27426
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.97878367216176e-07,
+      "loss": 0.3566,
+      "step": 27427
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.964919951398586e-07,
+      "loss": 0.4732,
+      "step": 27428
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.95106823756664e-07,
+      "loss": 0.3696,
+      "step": 27429
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.937228530833784e-07,
+      "loss": 0.3672,
+      "step": 27430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.923400831367444e-07,
+      "loss": 0.3821,
+      "step": 27431
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.909585139335041e-07,
+      "loss": 0.4687,
+      "step": 27432
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.89578145490355e-07,
+      "loss": 0.4136,
+      "step": 27433
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.881989778240395e-07,
+      "loss": 0.3768,
+      "step": 27434
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.86821010951233e-07,
+      "loss": 0.5,
+      "step": 27435
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.854442448886113e-07,
+      "loss": 0.4048,
+      "step": 27436
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.840686796528496e-07,
+      "loss": 0.3723,
+      "step": 27437
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.826943152605682e-07,
+      "loss": 0.3446,
+      "step": 27438
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.813211517284424e-07,
+      "loss": 0.3714,
+      "step": 27439
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.799491890730592e-07,
+      "loss": 0.4317,
+      "step": 27440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.785784273110386e-07,
+      "loss": 0.3721,
+      "step": 27441
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.772088664589783e-07,
+      "loss": 0.4792,
+      "step": 27442
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.758405065334317e-07,
+      "loss": 0.4571,
+      "step": 27443
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.744733475509746e-07,
+      "loss": 0.4293,
+      "step": 27444
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.731073895281604e-07,
+      "loss": 0.4594,
+      "step": 27445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.717426324815091e-07,
+      "loss": 0.4377,
+      "step": 27446
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.70379076427552e-07,
+      "loss": 0.4109,
+      "step": 27447
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.690167213827759e-07,
+      "loss": 0.424,
+      "step": 27448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.676555673636787e-07,
+      "loss": 0.4474,
+      "step": 27449
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.662956143867361e-07,
+      "loss": 0.3753,
+      "step": 27450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.649368624684128e-07,
+      "loss": 0.4469,
+      "step": 27451
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.635793116251511e-07,
+      "loss": 0.4286,
+      "step": 27452
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.622229618733712e-07,
+      "loss": 0.4295,
+      "step": 27453
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.608678132294933e-07,
+      "loss": 0.5136,
+      "step": 27454
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.595138657099377e-07,
+      "loss": 0.4458,
+      "step": 27455
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.58161119331069e-07,
+      "loss": 0.4068,
+      "step": 27456
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.56809574109274e-07,
+      "loss": 0.4445,
+      "step": 27457
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.554592300608953e-07,
+      "loss": 0.427,
+      "step": 27458
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.541100872022977e-07,
+      "loss": 0.3955,
+      "step": 27459
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.527621455497902e-07,
+      "loss": 0.4375,
+      "step": 27460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.514154051197042e-07,
+      "loss": 0.3852,
+      "step": 27461
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.500698659283267e-07,
+      "loss": 0.3638,
+      "step": 27462
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.487255279919558e-07,
+      "loss": 0.3638,
+      "step": 27463
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.473823913268451e-07,
+      "loss": 0.4573,
+      "step": 27464
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.460404559492595e-07,
+      "loss": 0.41,
+      "step": 27465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.446997218754526e-07,
+      "loss": 0.5155,
+      "step": 27466
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.433601891216336e-07,
+      "loss": 0.4307,
+      "step": 27467
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.420218577040339e-07,
+      "loss": 0.4207,
+      "step": 27468
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.406847276388295e-07,
+      "loss": 0.4158,
+      "step": 27469
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.393487989422187e-07,
+      "loss": 0.3957,
+      "step": 27470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.380140716303774e-07,
+      "loss": 0.3891,
+      "step": 27471
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.36680545719437e-07,
+      "loss": 0.5355,
+      "step": 27472
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.353482212255624e-07,
+      "loss": 0.4728,
+      "step": 27473
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.340170981648631e-07,
+      "loss": 0.3809,
+      "step": 27474
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.326871765534482e-07,
+      "loss": 0.4148,
+      "step": 27475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.313584564074272e-07,
+      "loss": 0.4292,
+      "step": 27476
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.300309377428871e-07,
+      "loss": 0.4454,
+      "step": 27477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.287046205758707e-07,
+      "loss": 0.4494,
+      "step": 27478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.27379504922443e-07,
+      "loss": 0.3986,
+      "step": 27479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.260555907986466e-07,
+      "loss": 0.3705,
+      "step": 27480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.247328782205131e-07,
+      "loss": 0.5592,
+      "step": 27481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.234113672040299e-07,
+      "loss": 0.455,
+      "step": 27482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.220910577652063e-07,
+      "loss": 0.4451,
+      "step": 27483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.207719499200183e-07,
+      "loss": 0.5089,
+      "step": 27484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.194540436844199e-07,
+      "loss": 0.3996,
+      "step": 27485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.181373390743873e-07,
+      "loss": 0.4583,
+      "step": 27486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.16821836105841e-07,
+      "loss": 0.4662,
+      "step": 27487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.155075347947127e-07,
+      "loss": 0.4086,
+      "step": 27488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.141944351569008e-07,
+      "loss": 0.4051,
+      "step": 27489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.128825372082926e-07,
+      "loss": 0.3853,
+      "step": 27490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.115718409647753e-07,
+      "loss": 0.401,
+      "step": 27491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.102623464422142e-07,
+      "loss": 0.4789,
+      "step": 27492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.089540536564631e-07,
+      "loss": 0.4288,
+      "step": 27493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.076469626233429e-07,
+      "loss": 0.4164,
+      "step": 27494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.063410733586739e-07,
+      "loss": 0.4003,
+      "step": 27495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.050363858782771e-07,
+      "loss": 0.4829,
+      "step": 27496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.037329001979398e-07,
+      "loss": 0.408,
+      "step": 27497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.024306163334271e-07,
+      "loss": 0.4392,
+      "step": 27498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.011295343005042e-07,
+      "loss": 0.4002,
+      "step": 27499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.998296541149252e-07,
+      "loss": 0.4978,
+      "step": 27500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.985309757924107e-07,
+      "loss": 0.4256,
+      "step": 27501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.972334993487039e-07,
+      "loss": 0.3799,
+      "step": 27502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.959372247994811e-07,
+      "loss": 0.4554,
+      "step": 27503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.946421521604518e-07,
+      "loss": 0.5293,
+      "step": 27504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.933482814472703e-07,
+      "loss": 0.4115,
+      "step": 27505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.920556126756128e-07,
+      "loss": 0.3552,
+      "step": 27506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.907641458611336e-07,
+      "loss": 0.4487,
+      "step": 27507
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.894738810194312e-07,
+      "loss": 0.335,
+      "step": 27508
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.881848181661598e-07,
+      "loss": 0.4648,
+      "step": 27509
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.868969573168849e-07,
+      "loss": 0.3968,
+      "step": 27510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.85610298487227e-07,
+      "loss": 0.3875,
+      "step": 27511
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.843248416927294e-07,
+      "loss": 0.4682,
+      "step": 27512
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.830405869489797e-07,
+      "loss": 0.4578,
+      "step": 27513
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.817575342714988e-07,
+      "loss": 0.3794,
+      "step": 27514
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.804756836758297e-07,
+      "loss": 0.3002,
+      "step": 27515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.791950351774712e-07,
+      "loss": 0.4306,
+      "step": 27516
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.779155887919442e-07,
+      "loss": 0.3998,
+      "step": 27517
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.766373445347141e-07,
+      "loss": 0.3903,
+      "step": 27518
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.753603024212685e-07,
+      "loss": 0.4169,
+      "step": 27519
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.740844624670617e-07,
+      "loss": 0.428,
+      "step": 27520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.728098246875147e-07,
+      "loss": 0.3246,
+      "step": 27521
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.715363890980819e-07,
+      "loss": 0.3832,
+      "step": 27522
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.702641557141731e-07,
+      "loss": 0.378,
+      "step": 27523
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.689931245511649e-07,
+      "loss": 0.4888,
+      "step": 27524
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.677232956244672e-07,
+      "loss": 0.3861,
+      "step": 27525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.664546689494233e-07,
+      "loss": 0.4718,
+      "step": 27526
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.651872445414098e-07,
+      "loss": 0.4143,
+      "step": 27527
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.639210224157588e-07,
+      "loss": 0.3927,
+      "step": 27528
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.626560025877915e-07,
+      "loss": 0.4206,
+      "step": 27529
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.613921850728178e-07,
+      "loss": 0.3694,
+      "step": 27530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.601295698861476e-07,
+      "loss": 0.3719,
+      "step": 27531
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.588681570430466e-07,
+      "loss": 0.396,
+      "step": 27532
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.576079465588025e-07,
+      "loss": 0.4503,
+      "step": 27533
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.563489384486477e-07,
+      "loss": 0.2961,
+      "step": 27534
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.550911327278253e-07,
+      "loss": 0.4656,
+      "step": 27535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.538345294115567e-07,
+      "loss": 0.5002,
+      "step": 27536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.525791285150628e-07,
+      "loss": 0.4712,
+      "step": 27537
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.513249300535318e-07,
+      "loss": 0.3905,
+      "step": 27538
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.500719340421402e-07,
+      "loss": 0.3556,
+      "step": 27539
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.488201404960648e-07,
+      "loss": 0.3944,
+      "step": 27540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.475695494304379e-07,
+      "loss": 0.4406,
+      "step": 27541
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.463201608604031e-07,
+      "loss": 0.451,
+      "step": 27542
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.450719748010925e-07,
+      "loss": 0.3975,
+      "step": 27543
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.438249912676053e-07,
+      "loss": 0.488,
+      "step": 27544
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.425792102750406e-07,
+      "loss": 0.3628,
+      "step": 27545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.413346318384528e-07,
+      "loss": 0.4534,
+      "step": 27546
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.400912559729411e-07,
+      "loss": 0.4776,
+      "step": 27547
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.388490826935267e-07,
+      "loss": 0.3767,
+      "step": 27548
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.376081120152533e-07,
+      "loss": 0.4404,
+      "step": 27549
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.36368343953142e-07,
+      "loss": 0.4013,
+      "step": 27550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.351297785222032e-07,
+      "loss": 0.422,
+      "step": 27551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.338924157374137e-07,
+      "loss": 0.3966,
+      "step": 27552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.326562556137616e-07,
+      "loss": 0.4012,
+      "step": 27553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.314212981662126e-07,
+      "loss": 0.3986,
+      "step": 27554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.301875434096993e-07,
+      "loss": 0.4561,
+      "step": 27555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.289549913591542e-07,
+      "loss": 0.332,
+      "step": 27556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.277236420295096e-07,
+      "loss": 0.4039,
+      "step": 27557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.264934954356538e-07,
+      "loss": 0.3471,
+      "step": 27558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.252645515924971e-07,
+      "loss": 0.3502,
+      "step": 27559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.240368105148942e-07,
+      "loss": 0.4265,
+      "step": 27560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.22810272217711e-07,
+      "loss": 0.3674,
+      "step": 27561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.2158493671578e-07,
+      "loss": 0.4365,
+      "step": 27562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.203608040239561e-07,
+      "loss": 0.4529,
+      "step": 27563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.191378741570497e-07,
+      "loss": 0.3896,
+      "step": 27564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.17916147129849e-07,
+      "loss": 0.3892,
+      "step": 27565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.166956229571419e-07,
+      "loss": 0.4704,
+      "step": 27566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.154763016537168e-07,
+      "loss": 0.396,
+      "step": 27567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.142581832343175e-07,
+      "loss": 0.3977,
+      "step": 27568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.130412677136987e-07,
+      "loss": 0.318,
+      "step": 27569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.11825555106571e-07,
+      "loss": 0.3565,
+      "step": 27570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.106110454276559e-07,
+      "loss": 0.443,
+      "step": 27571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.093977386916749e-07,
+      "loss": 0.5002,
+      "step": 27572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.08185634913283e-07,
+      "loss": 0.3874,
+      "step": 27573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.069747341071685e-07,
+      "loss": 0.4072,
+      "step": 27574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.057650362879752e-07,
+      "loss": 0.3946,
+      "step": 27575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.045565414703469e-07,
+      "loss": 0.355,
+      "step": 27576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.033492496689164e-07,
+      "loss": 0.48,
+      "step": 27577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.021431608983052e-07,
+      "loss": 0.3965,
+      "step": 27578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.009382751730907e-07,
+      "loss": 0.4034,
+      "step": 27579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.997345925078612e-07,
+      "loss": 0.4024,
+      "step": 27580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.985321129171939e-07,
+      "loss": 0.3902,
+      "step": 27581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.973308364156327e-07,
+      "loss": 0.4105,
+      "step": 27582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.961307630177326e-07,
+      "loss": 0.4869,
+      "step": 27583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.949318927380043e-07,
+      "loss": 0.4207,
+      "step": 27584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.937342255909584e-07,
+      "loss": 0.3969,
+      "step": 27585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.925377615910943e-07,
+      "loss": 0.4683,
+      "step": 27586
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.913425007528894e-07,
+      "loss": 0.409,
+      "step": 27587
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.90148443090821e-07,
+      "loss": 0.395,
+      "step": 27588
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.889555886193332e-07,
+      "loss": 0.4324,
+      "step": 27589
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.877639373528699e-07,
+      "loss": 0.5236,
+      "step": 27590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.865734893058417e-07,
+      "loss": 0.327,
+      "step": 27591
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.853842444926594e-07,
+      "loss": 0.4279,
+      "step": 27592
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.841962029277337e-07,
+      "loss": 0.4231,
+      "step": 27593
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.830093646254197e-07,
+      "loss": 0.4292,
+      "step": 27594
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.81823729600095e-07,
+      "loss": 0.4076,
+      "step": 27595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.806392978661146e-07,
+      "loss": 0.4677,
+      "step": 27596
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.794560694378115e-07,
+      "loss": 0.3794,
+      "step": 27597
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.782740443294965e-07,
+      "loss": 0.404,
+      "step": 27598
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.770932225554804e-07,
+      "loss": 0.4626,
+      "step": 27599
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.759136041300628e-07,
+      "loss": 0.4648,
+      "step": 27600
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 0.39674118161201477,
+      "eval_runtime": 844.1679,
+      "eval_samples_per_second": 28.501,
+      "eval_steps_per_second": 14.251,
+      "step": 27600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.747351890675212e-07,
+      "loss": 0.3918,
+      "step": 27601
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.735579773820998e-07,
+      "loss": 0.3348,
+      "step": 27602
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.72381969088065e-07,
+      "loss": 0.3672,
+      "step": 27603
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.712071641996497e-07,
+      "loss": 0.4094,
+      "step": 27604
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.700335627310649e-07,
+      "loss": 0.4405,
+      "step": 27605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.688611646965213e-07,
+      "loss": 0.2845,
+      "step": 27606
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.676899701101968e-07,
+      "loss": 0.4127,
+      "step": 27607
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.665199789862907e-07,
+      "loss": 0.4472,
+      "step": 27608
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.653511913389364e-07,
+      "loss": 0.3814,
+      "step": 27609
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.641836071823003e-07,
+      "loss": 0.396,
+      "step": 27610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.630172265305155e-07,
+      "loss": 0.5084,
+      "step": 27611
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.618520493976708e-07,
+      "loss": 0.4283,
+      "step": 27612
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.606880757978994e-07,
+      "loss": 0.3464,
+      "step": 27613
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.595253057452788e-07,
+      "loss": 0.4456,
+      "step": 27614
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.583637392538865e-07,
+      "loss": 0.4078,
+      "step": 27615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.572033763377671e-07,
+      "loss": 0.5032,
+      "step": 27616
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.560442170109759e-07,
+      "loss": 0.3712,
+      "step": 27617
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.548862612875461e-07,
+      "loss": 0.4035,
+      "step": 27618
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.537295091814775e-07,
+      "loss": 0.5179,
+      "step": 27619
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.525739607067926e-07,
+      "loss": 0.3568,
+      "step": 27620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.514196158774687e-07,
+      "loss": 0.3911,
+      "step": 27621
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.502664747074615e-07,
+      "loss": 0.3785,
+      "step": 27622
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.491145372107486e-07,
+      "loss": 0.4027,
+      "step": 27623
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.479638034012635e-07,
+      "loss": 0.3967,
+      "step": 27624
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.468142732929282e-07,
+      "loss": 0.4286,
+      "step": 27625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.456659468996761e-07,
+      "loss": 0.3827,
+      "step": 27626
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.445188242353739e-07,
+      "loss": 0.453,
+      "step": 27627
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.433729053139325e-07,
+      "loss": 0.3426,
+      "step": 27628
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.422281901492188e-07,
+      "loss": 0.3762,
+      "step": 27629
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.410846787550771e-07,
+      "loss": 0.3527,
+      "step": 27630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.399423711453522e-07,
+      "loss": 0.4198,
+      "step": 27631
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.388012673338661e-07,
+      "loss": 0.3635,
+      "step": 27632
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3766136733443e-07,
+      "loss": 0.3581,
+      "step": 27633
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.365226711608551e-07,
+      "loss": 0.533,
+      "step": 27634
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.353851788269082e-07,
+      "loss": 0.3588,
+      "step": 27635
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.342488903463671e-07,
+      "loss": 0.4287,
+      "step": 27636
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.331138057329655e-07,
+      "loss": 0.4914,
+      "step": 27637
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.319799250004586e-07,
+      "loss": 0.3828,
+      "step": 27638
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.308472481625693e-07,
+      "loss": 0.4712,
+      "step": 27639
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.297157752330084e-07,
+      "loss": 0.3278,
+      "step": 27640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.285855062254541e-07,
+      "loss": 0.4135,
+      "step": 27641
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.274564411536065e-07,
+      "loss": 0.359,
+      "step": 27642
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.263285800311213e-07,
+      "loss": 0.4672,
+      "step": 27643
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.252019228716542e-07,
+      "loss": 0.3896,
+      "step": 27644
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.240764696888278e-07,
+      "loss": 0.3677,
+      "step": 27645
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.229522204962867e-07,
+      "loss": 0.4203,
+      "step": 27646
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.218291753076199e-07,
+      "loss": 0.4026,
+      "step": 27647
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.207073341364166e-07,
+      "loss": 0.3722,
+      "step": 27648
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.195866969962771e-07,
+      "loss": 0.4824,
+      "step": 27649
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.18467263900746e-07,
+      "loss": 0.4414,
+      "step": 27650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.173490348633792e-07,
+      "loss": 0.4548,
+      "step": 27651
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.162320098977103e-07,
+      "loss": 0.3884,
+      "step": 27652
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.151161890172618e-07,
+      "loss": 0.4156,
+      "step": 27653
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.14001572235523e-07,
+      "loss": 0.522,
+      "step": 27654
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.128881595660051e-07,
+      "loss": 0.3796,
+      "step": 27655
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.117759510221753e-07,
+      "loss": 0.4859,
+      "step": 27656
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.106649466175006e-07,
+      "loss": 0.4175,
+      "step": 27657
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.095551463654147e-07,
+      "loss": 0.4273,
+      "step": 27658
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.084465502793512e-07,
+      "loss": 0.3949,
+      "step": 27659
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.073391583727549e-07,
+      "loss": 0.4314,
+      "step": 27660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.06232970658993e-07,
+      "loss": 0.3364,
+      "step": 27661
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.051279871514769e-07,
+      "loss": 0.4284,
+      "step": 27662
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.040242078635737e-07,
+      "loss": 0.4822,
+      "step": 27663
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.029216328086284e-07,
+      "loss": 0.4444,
+      "step": 27664
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.01820262000019e-07,
+      "loss": 0.4603,
+      "step": 27665
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.007200954510461e-07,
+      "loss": 0.3957,
+      "step": 27666
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.996211331750322e-07,
+      "loss": 0.4414,
+      "step": 27667
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.985233751852892e-07,
+      "loss": 0.3921,
+      "step": 27668
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.97426821495095e-07,
+      "loss": 0.411,
+      "step": 27669
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.963314721177281e-07,
+      "loss": 0.4242,
+      "step": 27670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.952373270664335e-07,
+      "loss": 0.4662,
+      "step": 27671
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.94144386354467e-07,
+      "loss": 0.3269,
+      "step": 27672
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.930526499950516e-07,
+      "loss": 0.4443,
+      "step": 27673
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.919621180013878e-07,
+      "loss": 0.3948,
+      "step": 27674
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.908727903866983e-07,
+      "loss": 0.4016,
+      "step": 27675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.897846671641615e-07,
+      "loss": 0.4764,
+      "step": 27676
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.886977483469335e-07,
+      "loss": 0.3337,
+      "step": 27677
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.876120339481815e-07,
+      "loss": 0.3741,
+      "step": 27678
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.865275239810396e-07,
+      "loss": 0.427,
+      "step": 27679
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.854442184586416e-07,
+      "loss": 0.4388,
+      "step": 27680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.843621173940993e-07,
+      "loss": 0.3891,
+      "step": 27681
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.832812208005022e-07,
+      "loss": 0.3905,
+      "step": 27682
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.822015286909399e-07,
+      "loss": 0.395,
+      "step": 27683
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.811230410784684e-07,
+      "loss": 0.4152,
+      "step": 27684
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.800457579761552e-07,
+      "loss": 0.3965,
+      "step": 27685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.789696793970344e-07,
+      "loss": 0.4439,
+      "step": 27686
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.778948053541177e-07,
+      "loss": 0.4241,
+      "step": 27687
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.768211358604391e-07,
+      "loss": 0.4343,
+      "step": 27688
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7574867092897714e-07,
+      "loss": 0.4271,
+      "step": 27689
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.746774105727103e-07,
+      "loss": 0.427,
+      "step": 27690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7360735480461714e-07,
+      "loss": 0.4416,
+      "step": 27691
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7253850363763173e-07,
+      "loss": 0.463,
+      "step": 27692
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.714708570847104e-07,
+      "loss": 0.3747,
+      "step": 27693
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.70404415158765e-07,
+      "loss": 0.403,
+      "step": 27694
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.693391778726852e-07,
+      "loss": 0.3451,
+      "step": 27695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.682751452394052e-07,
+      "loss": 0.3535,
+      "step": 27696
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6721231727177015e-07,
+      "loss": 0.3474,
+      "step": 27697
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6615069398264763e-07,
+      "loss": 0.5262,
+      "step": 27698
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6509027538490515e-07,
+      "loss": 0.3831,
+      "step": 27699
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6403106149135454e-07,
+      "loss": 0.4907,
+      "step": 27700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.629730523148301e-07,
+      "loss": 0.3389,
+      "step": 27701
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.619162478681327e-07,
+      "loss": 0.3715,
+      "step": 27702
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6086064816405205e-07,
+      "loss": 0.3705,
+      "step": 27703
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5980625321536686e-07,
+      "loss": 0.3899,
+      "step": 27704
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5875306303484465e-07,
+      "loss": 0.3957,
+      "step": 27705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.577010776352197e-07,
+      "loss": 0.3466,
+      "step": 27706
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5665029702923743e-07,
+      "loss": 0.4856,
+      "step": 27707
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5560072122959876e-07,
+      "loss": 0.3312,
+      "step": 27708
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5455235024902674e-07,
+      "loss": 0.4432,
+      "step": 27709
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.535051841002003e-07,
+      "loss": 0.3807,
+      "step": 27710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5245922279578687e-07,
+      "loss": 0.4585,
+      "step": 27711
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5141446634846543e-07,
+      "loss": 0.3496,
+      "step": 27712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.50370914770859e-07,
+      "loss": 0.4191,
+      "step": 27713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.493285680756132e-07,
+      "loss": 0.3661,
+      "step": 27714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4828742627532894e-07,
+      "loss": 0.4171,
+      "step": 27715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.472474893826295e-07,
+      "loss": 0.3834,
+      "step": 27716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.462087574100937e-07,
+      "loss": 0.4628,
+      "step": 27717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.451712303702782e-07,
+      "loss": 0.4521,
+      "step": 27718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4413490827576175e-07,
+      "loss": 0.4174,
+      "step": 27719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4309979113906775e-07,
+      "loss": 0.3686,
+      "step": 27720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4206587897273054e-07,
+      "loss": 0.3106,
+      "step": 27721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.410331717892735e-07,
+      "loss": 0.3496,
+      "step": 27722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.400016696011977e-07,
+      "loss": 0.3544,
+      "step": 27723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3897137242097095e-07,
+      "loss": 0.4101,
+      "step": 27724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.379422802610722e-07,
+      "loss": 0.4523,
+      "step": 27725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.369143931339581e-07,
+      "loss": 0.492,
+      "step": 27726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.358877110520743e-07,
+      "loss": 0.4033,
+      "step": 27727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.34862234027833e-07,
+      "loss": 0.3853,
+      "step": 27728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3383796207365766e-07,
+      "loss": 0.3365,
+      "step": 27729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.328148952019384e-07,
+      "loss": 0.3889,
+      "step": 27730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.317930334250653e-07,
+      "loss": 0.362,
+      "step": 27731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3077237675540614e-07,
+      "loss": 0.4761,
+      "step": 27732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.297529252053067e-07,
+      "loss": 0.3671,
+      "step": 27733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.287346787871127e-07,
+      "loss": 0.4164,
+      "step": 27734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2771763751313645e-07,
+      "loss": 0.4126,
+      "step": 27735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.267018013957014e-07,
+      "loss": 0.3957,
+      "step": 27736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.256871704470866e-07,
+      "loss": 0.3644,
+      "step": 27737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.246737446795934e-07,
+      "loss": 0.4412,
+      "step": 27738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.236615241054787e-07,
+      "loss": 0.4546,
+      "step": 27739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.226505087369881e-07,
+      "loss": 0.3884,
+      "step": 27740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2164069858636745e-07,
+      "loss": 0.3849,
+      "step": 27741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2063209366582925e-07,
+      "loss": 0.3195,
+      "step": 27742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1962469398758586e-07,
+      "loss": 0.3426,
+      "step": 27743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1861849956382757e-07,
+      "loss": 0.3824,
+      "step": 27744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.176135104067336e-07,
+      "loss": 0.3768,
+      "step": 27745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1660972652848295e-07,
+      "loss": 0.454,
+      "step": 27746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.156071479411994e-07,
+      "loss": 0.3689,
+      "step": 27747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1460577465702864e-07,
+      "loss": 0.4805,
+      "step": 27748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1360560668809447e-07,
+      "loss": 0.4196,
+      "step": 27749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.126066440464982e-07,
+      "loss": 0.4005,
+      "step": 27750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.116088867443302e-07,
+      "loss": 0.3788,
+      "step": 27751
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1061233479366965e-07,
+      "loss": 0.3827,
+      "step": 27752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0961698820657367e-07,
+      "loss": 0.3974,
+      "step": 27753
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0862284699508813e-07,
+      "loss": 0.4263,
+      "step": 27754
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0762991117125894e-07,
+      "loss": 0.3858,
+      "step": 27755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0663818074708757e-07,
+      "loss": 0.5026,
+      "step": 27756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.056476557345756e-07,
+      "loss": 0.3539,
+      "step": 27757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.046583361457357e-07,
+      "loss": 0.4318,
+      "step": 27758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.036702219925137e-07,
+      "loss": 0.5248,
+      "step": 27759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.02683313286889e-07,
+      "loss": 0.3747,
+      "step": 27760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.016976100407965e-07,
+      "loss": 0.4743,
+      "step": 27761
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0071311226617116e-07,
+      "loss": 0.4319,
+      "step": 27762
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.997298199749255e-07,
+      "loss": 0.3662,
+      "step": 27763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.987477331789613e-07,
+      "loss": 0.3884,
+      "step": 27764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9776685189015784e-07,
+      "loss": 0.4468,
+      "step": 27765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9678717612039453e-07,
+      "loss": 0.4329,
+      "step": 27766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.958087058815396e-07,
+      "loss": 0.3779,
+      "step": 27767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9483144118541703e-07,
+      "loss": 0.4076,
+      "step": 27768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9385538204387285e-07,
+      "loss": 0.4134,
+      "step": 27769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.928805284687087e-07,
+      "loss": 0.3879,
+      "step": 27770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.919068804717152e-07,
+      "loss": 0.4589,
+      "step": 27771
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9093443806469397e-07,
+      "loss": 0.3906,
+      "step": 27772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8996320125940235e-07,
+      "loss": 0.3619,
+      "step": 27773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.889931700676086e-07,
+      "loss": 0.3968,
+      "step": 27774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8802434450103675e-07,
+      "loss": 0.4635,
+      "step": 27775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8705672457143294e-07,
+      "loss": 0.3737,
+      "step": 27776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8609031029048783e-07,
+      "loss": 0.3653,
+      "step": 27777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.851251016699142e-07,
+      "loss": 0.447,
+      "step": 27778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.841610987213806e-07,
+      "loss": 0.4542,
+      "step": 27779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.831983014565776e-07,
+      "loss": 0.4434,
+      "step": 27780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8223670988712936e-07,
+      "loss": 0.3472,
+      "step": 27781
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8127632402468195e-07,
+      "loss": 0.4632,
+      "step": 27782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.803171438808817e-07,
+      "loss": 0.3874,
+      "step": 27783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.793591694673082e-07,
+      "loss": 0.3543,
+      "step": 27784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7840240079557445e-07,
+      "loss": 0.4731,
+      "step": 27785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7744683787726e-07,
+      "loss": 0.4027,
+      "step": 27786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7649248072392227e-07,
+      "loss": 0.4013,
+      "step": 27787
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.755393293471077e-07,
+      "loss": 0.3452,
+      "step": 27788
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7458738375837354e-07,
+      "loss": 0.3747,
+      "step": 27789
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7363664396922185e-07,
+      "loss": 0.3835,
+      "step": 27790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7268710999116553e-07,
+      "loss": 0.3922,
+      "step": 27791
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7173878183569545e-07,
+      "loss": 0.3777,
+      "step": 27792
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.707916595142913e-07,
+      "loss": 0.4476,
+      "step": 27793
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.698457430384217e-07,
+      "loss": 0.4308,
+      "step": 27794
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.68901032419533e-07,
+      "loss": 0.4106,
+      "step": 27795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.679575276690606e-07,
+      "loss": 0.4788,
+      "step": 27796
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6701522879840634e-07,
+      "loss": 0.4468,
+      "step": 27797
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6607413581899453e-07,
+      "loss": 0.3309,
+      "step": 27798
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.65134248742216e-07,
+      "loss": 0.427,
+      "step": 27799
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.641955675794395e-07,
+      "loss": 0.4657,
+      "step": 27800
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 0.3964925706386566,
+      "eval_runtime": 846.09,
+      "eval_samples_per_second": 28.437,
+      "eval_steps_per_second": 14.218,
+      "step": 27800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.632580923420226e-07,
+      "loss": 0.4975,
+      "step": 27801
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.623218230413228e-07,
+      "loss": 0.3964,
+      "step": 27802
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6138675968866445e-07,
+      "loss": 0.3876,
+      "step": 27803
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6045290229536067e-07,
+      "loss": 0.3867,
+      "step": 27804
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5952025087272466e-07,
+      "loss": 0.4221,
+      "step": 27805
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.585888054320474e-07,
+      "loss": 0.4014,
+      "step": 27806
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5765856598458657e-07,
+      "loss": 0.3566,
+      "step": 27807
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.567295325416109e-07,
+      "loss": 0.4409,
+      "step": 27808
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5580170511435584e-07,
+      "loss": 0.404,
+      "step": 27809
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5487508371407907e-07,
+      "loss": 0.3526,
+      "step": 27810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5394966835196054e-07,
+      "loss": 0.3942,
+      "step": 27811
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5302545903921347e-07,
+      "loss": 0.401,
+      "step": 27812
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5210245578704003e-07,
+      "loss": 0.3891,
+      "step": 27813
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5118065860658687e-07,
+      "loss": 0.4161,
+      "step": 27814
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5026006750903397e-07,
+      "loss": 0.4753,
+      "step": 27815
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4934068250549456e-07,
+      "loss": 0.359,
+      "step": 27816
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4842250360712645e-07,
+      "loss": 0.4252,
+      "step": 27817
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.475055308250208e-07,
+      "loss": 0.4563,
+      "step": 27818
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4658976417029086e-07,
+      "loss": 0.3961,
+      "step": 27819
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.456752036540167e-07,
+      "loss": 0.5171,
+      "step": 27820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.447618492872673e-07,
+      "loss": 0.4863,
+      "step": 27821
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.438497010810893e-07,
+      "loss": 0.4365,
+      "step": 27822
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.429387590465294e-07,
+      "loss": 0.4284,
+      "step": 27823
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.420290231946122e-07,
+      "loss": 0.4548,
+      "step": 27824
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.41120493536351e-07,
+      "loss": 0.3918,
+      "step": 27825
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4021317008273713e-07,
+      "loss": 0.3738,
+      "step": 27826
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.393070528447506e-07,
+      "loss": 0.4056,
+      "step": 27827
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3840214183337157e-07,
+      "loss": 0.4692,
+      "step": 27828
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3749843705953577e-07,
+      "loss": 0.4143,
+      "step": 27829
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.365959385341899e-07,
+      "loss": 0.4053,
+      "step": 27830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3569464626826976e-07,
+      "loss": 0.3962,
+      "step": 27831
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.347945602726554e-07,
+      "loss": 0.3917,
+      "step": 27832
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.338956805582605e-07,
+      "loss": 0.3553,
+      "step": 27833
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.32998007135954e-07,
+      "loss": 0.3999,
+      "step": 27834
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.321015400166161e-07,
+      "loss": 0.4037,
+      "step": 27835
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.312062792110826e-07,
+      "loss": 0.4229,
+      "step": 27836
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.303122247302004e-07,
+      "loss": 0.4505,
+      "step": 27837
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2941937658477197e-07,
+      "loss": 0.4016,
+      "step": 27838
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.285277347856219e-07,
+      "loss": 0.4356,
+      "step": 27839
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.276372993435417e-07,
+      "loss": 0.4034,
+      "step": 27840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2674807026930044e-07,
+      "loss": 0.4008,
+      "step": 27841
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.258600475736562e-07,
+      "loss": 0.3723,
+      "step": 27842
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.24973231267367e-07,
+      "loss": 0.4461,
+      "step": 27843
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2408762136116875e-07,
+      "loss": 0.362,
+      "step": 27844
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2320321786577514e-07,
+      "loss": 0.352,
+      "step": 27845
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.223200207918775e-07,
+      "loss": 0.341,
+      "step": 27846
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.214380301501896e-07,
+      "loss": 0.4745,
+      "step": 27847
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.205572459513695e-07,
+      "loss": 0.4402,
+      "step": 27848
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.196776682060865e-07,
+      "loss": 0.4236,
+      "step": 27849
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1879929692498757e-07,
+      "loss": 0.4291,
+      "step": 27850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1792213211869757e-07,
+      "loss": 0.4613,
+      "step": 27851
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.170461737978303e-07,
+      "loss": 0.4526,
+      "step": 27852
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.161714219729883e-07,
+      "loss": 0.4686,
+      "step": 27853
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.152978766547743e-07,
+      "loss": 0.4141,
+      "step": 27854
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.144255378537464e-07,
+      "loss": 0.4502,
+      "step": 27855
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.135544055804629e-07,
+      "loss": 0.5132,
+      "step": 27856
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1268447984547087e-07,
+      "loss": 0.4097,
+      "step": 27857
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.118157606593064e-07,
+      "loss": 0.4589,
+      "step": 27858
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1094824803247213e-07,
+      "loss": 0.4327,
+      "step": 27859
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.100819419754708e-07,
+      "loss": 0.3616,
+      "step": 27860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.092168424987829e-07,
+      "loss": 0.4711,
+      "step": 27861
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0835294961290005e-07,
+      "loss": 0.4158,
+      "step": 27862
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0749026332824725e-07,
+      "loss": 0.3873,
+      "step": 27863
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0662878365529393e-07,
+      "loss": 0.3859,
+      "step": 27864
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.057685106044539e-07,
+      "loss": 0.3931,
+      "step": 27865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.049094441861411e-07,
+      "loss": 0.4814,
+      "step": 27866
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.040515844107583e-07,
+      "loss": 0.4629,
+      "step": 27867
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0319493128866396e-07,
+      "loss": 0.3827,
+      "step": 27868
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0233948483026077e-07,
+      "loss": 0.3493,
+      "step": 27869
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0148524504588493e-07,
+      "loss": 0.41,
+      "step": 27870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.006322119458727e-07,
+      "loss": 0.4599,
+      "step": 27871
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.997803855405601e-07,
+      "loss": 0.3754,
+      "step": 27872
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.989297658402501e-07,
+      "loss": 0.3694,
+      "step": 27873
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.980803528552345e-07,
+      "loss": 0.3962,
+      "step": 27874
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.972321465957939e-07,
+      "loss": 0.383,
+      "step": 27875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9638514707220897e-07,
+      "loss": 0.4358,
+      "step": 27876
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.95539354294716e-07,
+      "loss": 0.4742,
+      "step": 27877
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9469476827355123e-07,
+      "loss": 0.4315,
+      "step": 27878
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9385138901895094e-07,
+      "loss": 0.4401,
+      "step": 27879
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.930092165411069e-07,
+      "loss": 0.3965,
+      "step": 27880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9216825085022214e-07,
+      "loss": 0.4369,
+      "step": 27881
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9132849195647737e-07,
+      "loss": 0.3962,
+      "step": 27882
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9048993987002004e-07,
+      "loss": 0.3742,
+      "step": 27883
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.896525946010198e-07,
+      "loss": 0.4047,
+      "step": 27884
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8881645615960184e-07,
+      "loss": 0.3428,
+      "step": 27885
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8798152455589146e-07,
+      "loss": 0.4644,
+      "step": 27886
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8714779979999163e-07,
+      "loss": 0.4248,
+      "step": 27887
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.863152819019832e-07,
+      "loss": 0.4229,
+      "step": 27888
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.85483970871947e-07,
+      "loss": 0.4083,
+      "step": 27889
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8465386671995275e-07,
+      "loss": 0.3634,
+      "step": 27890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.83824969456048e-07,
+      "loss": 0.3887,
+      "step": 27891
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.82997279090258e-07,
+      "loss": 0.3741,
+      "step": 27892
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8217079563259694e-07,
+      "loss": 0.3575,
+      "step": 27893
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.813455190930681e-07,
+      "loss": 0.4335,
+      "step": 27894
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.805214494816744e-07,
+      "loss": 0.4129,
+      "step": 27895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.796985868083857e-07,
+      "loss": 0.3854,
+      "step": 27896
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7887693108314963e-07,
+      "loss": 0.3957,
+      "step": 27897
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7805648231592485e-07,
+      "loss": 0.4773,
+      "step": 27898
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.772372405166257e-07,
+      "loss": 0.3998,
+      "step": 27899
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7641920569517753e-07,
+      "loss": 0.4176,
+      "step": 27900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.756023778614836e-07,
+      "loss": 0.4144,
+      "step": 27901
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.747867570254359e-07,
+      "loss": 0.3498,
+      "step": 27902
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7397234319688215e-07,
+      "loss": 0.3957,
+      "step": 27903
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.731591363857033e-07,
+      "loss": 0.401,
+      "step": 27904
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7234713660173603e-07,
+      "loss": 0.5813,
+      "step": 27905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7153634385479466e-07,
+      "loss": 0.3991,
+      "step": 27906
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.707267581547157e-07,
+      "loss": 0.4247,
+      "step": 27907
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.699183795112803e-07,
+      "loss": 0.4691,
+      "step": 27908
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6911120793428056e-07,
+      "loss": 0.3716,
+      "step": 27909
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.683052434334754e-07,
+      "loss": 0.5392,
+      "step": 27910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.675004860186348e-07,
+      "loss": 0.4601,
+      "step": 27911
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.666969356994842e-07,
+      "loss": 0.4051,
+      "step": 27912
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.658945924857603e-07,
+      "loss": 0.3774,
+      "step": 27913
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6509345638717764e-07,
+      "loss": 0.4141,
+      "step": 27914
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.642935274134173e-07,
+      "loss": 0.3904,
+      "step": 27915
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.634948055741715e-07,
+      "loss": 0.4032,
+      "step": 27916
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6269729087909924e-07,
+      "loss": 0.4657,
+      "step": 27917
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.619009833378705e-07,
+      "loss": 0.4373,
+      "step": 27918
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6110588296009986e-07,
+      "loss": 0.3351,
+      "step": 27919
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.603119897554351e-07,
+      "loss": 0.4671,
+      "step": 27920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5951930373346864e-07,
+      "loss": 0.3347,
+      "step": 27921
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5872782490379277e-07,
+      "loss": 0.4295,
+      "step": 27922
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5793755327599977e-07,
+      "loss": 0.4749,
+      "step": 27923
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.571484888596376e-07,
+      "loss": 0.3779,
+      "step": 27924
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.563606316642764e-07,
+      "loss": 0.4037,
+      "step": 27925
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5557398169943067e-07,
+      "loss": 0.3978,
+      "step": 27926
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.547885389746485e-07,
+      "loss": 0.3729,
+      "step": 27927
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5400430349941107e-07,
+      "loss": 0.4451,
+      "step": 27928
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5322127528322194e-07,
+      "loss": 0.3476,
+      "step": 27929
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.524394543355513e-07,
+      "loss": 0.3459,
+      "step": 27930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.516588406658693e-07,
+      "loss": 0.3723,
+      "step": 27931
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5087943428362405e-07,
+      "loss": 0.396,
+      "step": 27932
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.501012351982523e-07,
+      "loss": 0.4007,
+      "step": 27933
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4932424341915785e-07,
+      "loss": 0.4036,
+      "step": 27934
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.485484589557552e-07,
+      "loss": 0.3824,
+      "step": 27935
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4777388181743687e-07,
+      "loss": 0.4931,
+      "step": 27936
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.470005120135732e-07,
+      "loss": 0.3842,
+      "step": 27937
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.462283495535234e-07,
+      "loss": 0.3901,
+      "step": 27938
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.454573944466243e-07,
+      "loss": 0.4421,
+      "step": 27939
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.446876467022352e-07,
+      "loss": 0.4296,
+      "step": 27940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4391910632963755e-07,
+      "loss": 0.3775,
+      "step": 27941
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.431517733381683e-07,
+      "loss": 0.4216,
+      "step": 27942
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.423856477370867e-07,
+      "loss": 0.43,
+      "step": 27943
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.416207295356743e-07,
+      "loss": 0.3955,
+      "step": 27944
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4085701874319024e-07,
+      "loss": 0.4934,
+      "step": 27945
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4009451536888274e-07,
+      "loss": 0.4351,
+      "step": 27946
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.393332194219777e-07,
+      "loss": 0.3718,
+      "step": 27947
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3857313091169007e-07,
+      "loss": 0.4266,
+      "step": 27948
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3781424984721246e-07,
+      "loss": 0.3736,
+      "step": 27949
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3705657623772636e-07,
+      "loss": 0.3894,
+      "step": 27950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3630011009242447e-07,
+      "loss": 0.4623,
+      "step": 27951
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.355448514204439e-07,
+      "loss": 0.3498,
+      "step": 27952
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.347908002309329e-07,
+      "loss": 0.3862,
+      "step": 27953
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.340379565330064e-07,
+      "loss": 0.3571,
+      "step": 27954
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3328632033580156e-07,
+      "loss": 0.4546,
+      "step": 27955
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3253589164837774e-07,
+      "loss": 0.3546,
+      "step": 27956
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3178667047986103e-07,
+      "loss": 0.4033,
+      "step": 27957
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.310386568392886e-07,
+      "loss": 0.3902,
+      "step": 27958
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.302918507357199e-07,
+      "loss": 0.402,
+      "step": 27959
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2954625217820324e-07,
+      "loss": 0.4439,
+      "step": 27960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2880186117575363e-07,
+      "loss": 0.4295,
+      "step": 27961
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.28058677737375e-07,
+      "loss": 0.4615,
+      "step": 27962
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2731670187208233e-07,
+      "loss": 0.4005,
+      "step": 27963
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.265759335888462e-07,
+      "loss": 0.4126,
+      "step": 27964
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2583637289662618e-07,
+      "loss": 0.445,
+      "step": 27965
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2509801980438173e-07,
+      "loss": 0.3226,
+      "step": 27966
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2436087432103904e-07,
+      "loss": 0.4312,
+      "step": 27967
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.236249364555354e-07,
+      "loss": 0.4151,
+      "step": 27968
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.228902062167637e-07,
+      "loss": 0.4157,
+      "step": 27969
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2215668361362795e-07,
+      "loss": 0.4091,
+      "step": 27970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2142436865499882e-07,
+      "loss": 0.392,
+      "step": 27971
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.206932613497359e-07,
+      "loss": 0.3907,
+      "step": 27972
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1996336170669873e-07,
+      "loss": 0.4087,
+      "step": 27973
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1923466973471363e-07,
+      "loss": 0.4011,
+      "step": 27974
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1850718544260684e-07,
+      "loss": 0.3822,
+      "step": 27975
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1778090883917134e-07,
+      "loss": 0.3529,
+      "step": 27976
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1705583993322231e-07,
+      "loss": 0.4131,
+      "step": 27977
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.163319787335083e-07,
+      "loss": 0.4869,
+      "step": 27978
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1560932524880007e-07,
+      "loss": 0.4367,
+      "step": 27979
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1488787948784616e-07,
+      "loss": 0.4213,
+      "step": 27980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1416764145938407e-07,
+      "loss": 0.3828,
+      "step": 27981
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.134486111721179e-07,
+      "loss": 0.4443,
+      "step": 27982
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1273078863476292e-07,
+      "loss": 0.4477,
+      "step": 27983
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1201417385598998e-07,
+      "loss": 0.3552,
+      "step": 27984
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1129876684449213e-07,
+      "loss": 0.4903,
+      "step": 27985
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1058456760891798e-07,
+      "loss": 0.4526,
+      "step": 27986
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.098715761579162e-07,
+      "loss": 0.3762,
+      "step": 27987
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.091597925001132e-07,
+      "loss": 0.4862,
+      "step": 27988
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.084492166441243e-07,
+      "loss": 0.4031,
+      "step": 27989
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.077398485985427e-07,
+      "loss": 0.4073,
+      "step": 27990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.070316883719725e-07,
+      "loss": 0.3731,
+      "step": 27991
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0632473597296253e-07,
+      "loss": 0.4715,
+      "step": 27992
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0561899141009476e-07,
+      "loss": 0.4416,
+      "step": 27993
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0491445469188463e-07,
+      "loss": 0.4478,
+      "step": 27994
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0421112582688085e-07,
+      "loss": 0.475,
+      "step": 27995
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.035090048235766e-07,
+      "loss": 0.3817,
+      "step": 27996
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0280809169048732e-07,
+      "loss": 0.4568,
+      "step": 27997
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0210838643608398e-07,
+      "loss": 0.449,
+      "step": 27998
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0140988906883762e-07,
+      "loss": 0.4619,
+      "step": 27999
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0071259959721922e-07,
+      "loss": 0.3959,
+      "step": 28000
+    },
+    {
+      "epoch": 0.98,
+      "eval_loss": 0.3964395821094513,
+      "eval_runtime": 848.064,
+      "eval_samples_per_second": 28.371,
+      "eval_steps_per_second": 14.185,
+      "step": 28000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0001651802964428e-07,
+      "loss": 0.4097,
+      "step": 28001
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9932164437455047e-07,
+      "loss": 0.3799,
+      "step": 28002
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9862797864034221e-07,
+      "loss": 0.4018,
+      "step": 28003
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9793552083541278e-07,
+      "loss": 0.4298,
+      "step": 28004
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9724427096816656e-07,
+      "loss": 0.3656,
+      "step": 28005
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9655422904693022e-07,
+      "loss": 0.4754,
+      "step": 28006
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9586539508009704e-07,
+      "loss": 0.4215,
+      "step": 28007
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.951777690759715e-07,
+      "loss": 0.424,
+      "step": 28008
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.944913510428914e-07,
+      "loss": 0.3461,
+      "step": 28009
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9380614098917226e-07,
+      "loss": 0.401,
+      "step": 28010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9312213892308528e-07,
+      "loss": 0.4471,
+      "step": 28011
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.924393448529349e-07,
+      "loss": 0.4103,
+      "step": 28012
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.91757758786959e-07,
+      "loss": 0.3795,
+      "step": 28013
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9107738073342874e-07,
+      "loss": 0.3583,
+      "step": 28014
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9039821070057084e-07,
+      "loss": 0.4066,
+      "step": 28015
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.897202486965899e-07,
+      "loss": 0.4839,
+      "step": 28016
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.890434947297237e-07,
+      "loss": 0.41,
+      "step": 28017
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8836794880814358e-07,
+      "loss": 0.4334,
+      "step": 28018
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.876936109400207e-07,
+      "loss": 0.3907,
+      "step": 28019
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8702048113352631e-07,
+      "loss": 0.4977,
+      "step": 28020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.863485593968095e-07,
+      "loss": 0.4466,
+      "step": 28021
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8567784573799707e-07,
+      "loss": 0.3515,
+      "step": 28022
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8500834016520475e-07,
+      "loss": 0.3773,
+      "step": 28023
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8434004268654825e-07,
+      "loss": 0.4091,
+      "step": 28024
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8367295331011003e-07,
+      "loss": 0.4809,
+      "step": 28025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8300707204395028e-07,
+      "loss": 0.4348,
+      "step": 28026
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8234239889614036e-07,
+      "loss": 0.4576,
+      "step": 28027
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8167893387472935e-07,
+      "loss": 0.3438,
+      "step": 28028
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8101667698774415e-07,
+      "loss": 0.4248,
+      "step": 28029
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8035562824320063e-07,
+      "loss": 0.4252,
+      "step": 28030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7969578764909235e-07,
+      "loss": 0.3799,
+      "step": 28031
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.790371552134129e-07,
+      "loss": 0.3976,
+      "step": 28032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7837973094413375e-07,
+      "loss": 0.4164,
+      "step": 28033
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7772351484920402e-07,
+      "loss": 0.4071,
+      "step": 28034
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7706850693658407e-07,
+      "loss": 0.3659,
+      "step": 28035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7641470721417864e-07,
+      "loss": 0.3795,
+      "step": 28036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7576211568991474e-07,
+      "loss": 0.4452,
+      "step": 28037
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7511073237168607e-07,
+      "loss": 0.466,
+      "step": 28038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.744605572673863e-07,
+      "loss": 0.4391,
+      "step": 28039
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7381159038487583e-07,
+      "loss": 0.4762,
+      "step": 28040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.731638317320039e-07,
+      "loss": 0.3273,
+      "step": 28041
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7251728131661983e-07,
+      "loss": 0.4092,
+      "step": 28042
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7187193914655063e-07,
+      "loss": 0.5011,
+      "step": 28043
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7122780522960123e-07,
+      "loss": 0.4494,
+      "step": 28044
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.705848795735654e-07,
+      "loss": 0.3844,
+      "step": 28045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6994316218623685e-07,
+      "loss": 0.3742,
+      "step": 28046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6930265307536497e-07,
+      "loss": 0.4016,
+      "step": 28047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.686633522487213e-07,
+      "loss": 0.4516,
+      "step": 28048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6802525971403304e-07,
+      "loss": 0.3729,
+      "step": 28049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.673883754790162e-07,
+      "loss": 0.4141,
+      "step": 28050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6675269955139794e-07,
+      "loss": 0.423,
+      "step": 28051
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.661182319388499e-07,
+      "loss": 0.4865,
+      "step": 28052
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6548497264906592e-07,
+      "loss": 0.3916,
+      "step": 28053
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6485292168970656e-07,
+      "loss": 0.3229,
+      "step": 28054
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6422207906843234e-07,
+      "loss": 0.3666,
+      "step": 28055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.635924447928594e-07,
+      "loss": 0.3741,
+      "step": 28056
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6296401887061497e-07,
+      "loss": 0.4941,
+      "step": 28057
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6233680130931518e-07,
+      "loss": 0.4338,
+      "step": 28058
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6171079211653172e-07,
+      "loss": 0.3361,
+      "step": 28059
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6108599129986969e-07,
+      "loss": 0.4224,
+      "step": 28060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6046239886686743e-07,
+      "loss": 0.3989,
+      "step": 28061
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.598400148250745e-07,
+      "loss": 0.4911,
+      "step": 28062
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.592188391820404e-07,
+      "loss": 0.4042,
+      "step": 28063
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5859887194527023e-07,
+      "loss": 0.3916,
+      "step": 28064
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57980113122258e-07,
+      "loss": 0.4345,
+      "step": 28065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5736256272051998e-07,
+      "loss": 0.4013,
+      "step": 28066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5674622074750567e-07,
+      "loss": 0.4556,
+      "step": 28067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5613108721068693e-07,
+      "loss": 0.4107,
+      "step": 28068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5551716211751334e-07,
+      "loss": 0.4532,
+      "step": 28069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5490444547540118e-07,
+      "loss": 0.4418,
+      "step": 28070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5429293729177786e-07,
+      "loss": 0.3854,
+      "step": 28071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5368263757403745e-07,
+      "loss": 0.4028,
+      "step": 28072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5307354632957404e-07,
+      "loss": 0.4159,
+      "step": 28073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5246566356574842e-07,
+      "loss": 0.4069,
+      "step": 28074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5185898928993247e-07,
+      "loss": 0.5097,
+      "step": 28075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5125352350945366e-07,
+      "loss": 0.4206,
+      "step": 28076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5064926623166166e-07,
+      "loss": 0.374,
+      "step": 28077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5004621746383952e-07,
+      "loss": 0.3991,
+      "step": 28078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4944437721331473e-07,
+      "loss": 0.4027,
+      "step": 28079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4884374548734814e-07,
+      "loss": 0.4483,
+      "step": 28080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4824432229323393e-07,
+      "loss": 0.4148,
+      "step": 28081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4764610763821073e-07,
+      "loss": 0.3957,
+      "step": 28082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4704910152951723e-07,
+      "loss": 0.3905,
+      "step": 28083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4645330397439206e-07,
+      "loss": 0.4614,
+      "step": 28084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4585871498002946e-07,
+      "loss": 0.3574,
+      "step": 28085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4526533455364587e-07,
+      "loss": 0.3672,
+      "step": 28086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4467316270240228e-07,
+      "loss": 0.4519,
+      "step": 28087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4408219943348178e-07,
+      "loss": 0.3777,
+      "step": 28088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4349244475403423e-07,
+      "loss": 0.4033,
+      "step": 28089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4290389867118726e-07,
+      "loss": 0.393,
+      "step": 28090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4231656119207958e-07,
+      "loss": 0.4387,
+      "step": 28091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4173043232380557e-07,
+      "loss": 0.3707,
+      "step": 28092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.411455120734706e-07,
+      "loss": 0.4334,
+      "step": 28093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4056180044814682e-07,
+      "loss": 0.3626,
+      "step": 28094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3997929745489524e-07,
+      "loss": 0.5072,
+      "step": 28095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.39398003100788e-07,
+      "loss": 0.3995,
+      "step": 28096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3881791739283057e-07,
+      "loss": 0.4293,
+      "step": 28097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.382390403380618e-07,
+      "loss": 0.3348,
+      "step": 28098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3766137194348716e-07,
+      "loss": 0.3869,
+      "step": 28099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.370849122161011e-07,
+      "loss": 0.4057,
+      "step": 28100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.365096611628758e-07,
+      "loss": 0.4662,
+      "step": 28101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.359356187907612e-07,
+      "loss": 0.3755,
+      "step": 28102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3536278510672963e-07,
+      "loss": 0.3609,
+      "step": 28103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3479116011769767e-07,
+      "loss": 0.3853,
+      "step": 28104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3422074383059314e-07,
+      "loss": 0.4025,
+      "step": 28105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3365153625231052e-07,
+      "loss": 0.4048,
+      "step": 28106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3308353738973322e-07,
+      "loss": 0.3884,
+      "step": 28107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3251674724975571e-07,
+      "loss": 0.4289,
+      "step": 28108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3195116583923916e-07,
+      "loss": 0.4493,
+      "step": 28109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3138679316500035e-07,
+      "loss": 0.4214,
+      "step": 28110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3082362923390046e-07,
+      "loss": 0.3712,
+      "step": 28111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3026167405274515e-07,
+      "loss": 0.4262,
+      "step": 28112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2970092762832897e-07,
+      "loss": 0.3486,
+      "step": 28113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.291413899674465e-07,
+      "loss": 0.4299,
+      "step": 28114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2858306107685902e-07,
+      "loss": 0.4356,
+      "step": 28115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2802594096333886e-07,
+      "loss": 0.425,
+      "step": 28116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2747002963362508e-07,
+      "loss": 0.3814,
+      "step": 28117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2691532709444565e-07,
+      "loss": 0.4089,
+      "step": 28118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.263618333525063e-07,
+      "loss": 0.4214,
+      "step": 28119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2580954841452386e-07,
+      "loss": 0.4251,
+      "step": 28120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.252584722871597e-07,
+      "loss": 0.4531,
+      "step": 28121
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2470860497709735e-07,
+      "loss": 0.4966,
+      "step": 28122
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2415994649099817e-07,
+      "loss": 0.3474,
+      "step": 28123
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2361249683549015e-07,
+      "loss": 0.4473,
+      "step": 28124
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2306625601720133e-07,
+      "loss": 0.3909,
+      "step": 28125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2252122404274868e-07,
+      "loss": 0.4195,
+      "step": 28126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.219774009187269e-07,
+      "loss": 0.4758,
+      "step": 28127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.214347866517085e-07,
+      "loss": 0.4237,
+      "step": 28128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2089338124827708e-07,
+      "loss": 0.4077,
+      "step": 28129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2035318471497193e-07,
+      "loss": 0.4492,
+      "step": 28130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.198141970583433e-07,
+      "loss": 0.4045,
+      "step": 28131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.192764182849082e-07,
+      "loss": 0.4023,
+      "step": 28132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1873984840117259e-07,
+      "loss": 0.4553,
+      "step": 28133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1820448741363121e-07,
+      "loss": 0.5205,
+      "step": 28134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1767033532876781e-07,
+      "loss": 0.4659,
+      "step": 28135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1713739215305497e-07,
+      "loss": 0.4307,
+      "step": 28136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1660565789293199e-07,
+      "loss": 0.4141,
+      "step": 28137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1607513255482704e-07,
+      "loss": 0.3532,
+      "step": 28138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1554581614517945e-07,
+      "loss": 0.3873,
+      "step": 28139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1501770867038408e-07,
+      "loss": 0.4149,
+      "step": 28140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1449081013683582e-07,
+      "loss": 0.3553,
+      "step": 28141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1396512055090736e-07,
+      "loss": 0.4728,
+      "step": 28142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1344063991897135e-07,
+      "loss": 0.4467,
+      "step": 28143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.129173682473783e-07,
+      "loss": 0.4394,
+      "step": 28144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1239530554244537e-07,
+      "loss": 0.4773,
+      "step": 28145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1187445181050082e-07,
+      "loss": 0.3236,
+      "step": 28146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1135480705783963e-07,
+      "loss": 0.3786,
+      "step": 28147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1083637129076785e-07,
+      "loss": 0.4161,
+      "step": 28148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1031914451554715e-07,
+      "loss": 0.3935,
+      "step": 28149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0980312673845027e-07,
+      "loss": 0.4294,
+      "step": 28150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.092883179657056e-07,
+      "loss": 0.4543,
+      "step": 28151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0877471820356366e-07,
+      "loss": 0.4698,
+      "step": 28152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0826232745821951e-07,
+      "loss": 0.5098,
+      "step": 28153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.077511457358904e-07,
+      "loss": 0.3842,
+      "step": 28154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0724117304274916e-07,
+      "loss": 0.4439,
+      "step": 28155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0673240938499085e-07,
+      "loss": 0.4171,
+      "step": 28156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0622485476875499e-07,
+      "loss": 0.3952,
+      "step": 28157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0571850920018111e-07,
+      "loss": 0.3637,
+      "step": 28158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0521337268541986e-07,
+      "loss": 0.3351,
+      "step": 28159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0470944523055526e-07,
+      "loss": 0.3674,
+      "step": 28160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0420672684171572e-07,
+      "loss": 0.391,
+      "step": 28161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0370521752496309e-07,
+      "loss": 0.3457,
+      "step": 28162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0320491728638137e-07,
+      "loss": 0.4072,
+      "step": 28163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0270582613202129e-07,
+      "loss": 0.3438,
+      "step": 28164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0220794406792244e-07,
+      "loss": 0.3731,
+      "step": 28165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0171127110011336e-07,
+      "loss": 0.4011,
+      "step": 28166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0121580723460033e-07,
+      "loss": 0.3881,
+      "step": 28167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0072155247738968e-07,
+      "loss": 0.3276,
+      "step": 28168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.002285068344544e-07,
+      "loss": 0.3959,
+      "step": 28169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.973667031176748e-08,
+      "loss": 0.4042,
+      "step": 28170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.924604291526862e-08,
+      "loss": 0.3839,
+      "step": 28171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.875662465091973e-08,
+      "loss": 0.4574,
+      "step": 28172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.826841552462718e-08,
+      "loss": 0.4008,
+      "step": 28173
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.778141554229736e-08,
+      "loss": 0.394,
+      "step": 28174
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.729562470983666e-08,
+      "loss": 0.3768,
+      "step": 28175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.681104303311817e-08,
+      "loss": 0.3802,
+      "step": 28176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.632767051801494e-08,
+      "loss": 0.452,
+      "step": 28177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.584550717036678e-08,
+      "loss": 0.3931,
+      "step": 28178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.536455299601344e-08,
+      "loss": 0.4217,
+      "step": 28179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.488480800077249e-08,
+      "loss": 0.5675,
+      "step": 28180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.44062721904615e-08,
+      "loss": 0.4111,
+      "step": 28181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.392894557085364e-08,
+      "loss": 0.3301,
+      "step": 28182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.345282814774426e-08,
+      "loss": 0.3737,
+      "step": 28183
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.297791992689542e-08,
+      "loss": 0.5173,
+      "step": 28184
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.250422091403588e-08,
+      "loss": 0.3891,
+      "step": 28185
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.203173111492768e-08,
+      "loss": 0.4124,
+      "step": 28186
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.156045053525519e-08,
+      "loss": 0.3543,
+      "step": 28187
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.109037918075825e-08,
+      "loss": 0.3511,
+      "step": 28188
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.062151705709898e-08,
+      "loss": 0.4026,
+      "step": 28189
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.015386416996174e-08,
+      "loss": 0.4244,
+      "step": 28190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.968742052500867e-08,
+      "loss": 0.3221,
+      "step": 28191
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.92221861278908e-08,
+      "loss": 0.3527,
+      "step": 28192
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.875816098422585e-08,
+      "loss": 0.3614,
+      "step": 28193
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.829534509963155e-08,
+      "loss": 0.4543,
+      "step": 28194
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.783373847972565e-08,
+      "loss": 0.3263,
+      "step": 28195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.737334113007034e-08,
+      "loss": 0.3406,
+      "step": 28196
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.691415305626116e-08,
+      "loss": 0.4705,
+      "step": 28197
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.645617426384923e-08,
+      "loss": 0.3684,
+      "step": 28198
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.599940475836343e-08,
+      "loss": 0.4627,
+      "step": 28199
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.554384454535492e-08,
+      "loss": 0.3666,
+      "step": 28200
+    },
+    {
+      "epoch": 0.99,
+      "eval_loss": 0.3963852822780609,
+      "eval_runtime": 853.5974,
+      "eval_samples_per_second": 28.187,
+      "eval_steps_per_second": 14.093,
+      "step": 28200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.508949363031927e-08,
+      "loss": 0.3751,
+      "step": 28201
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.463635201876319e-08,
+      "loss": 0.3674,
+      "step": 28202
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.418441971616009e-08,
+      "loss": 0.4495,
+      "step": 28203
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.373369672800557e-08,
+      "loss": 0.3992,
+      "step": 28204
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.328418305972863e-08,
+      "loss": 0.4202,
+      "step": 28205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.283587871678045e-08,
+      "loss": 0.3994,
+      "step": 28206
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.238878370459003e-08,
+      "loss": 0.3817,
+      "step": 28207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.194289802857525e-08,
+      "loss": 0.3644,
+      "step": 28208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.149822169410959e-08,
+      "loss": 0.4196,
+      "step": 28209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.105475470659985e-08,
+      "loss": 0.4413,
+      "step": 28210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.061249707138619e-08,
+      "loss": 0.4265,
+      "step": 28211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.017144879385318e-08,
+      "loss": 0.366,
+      "step": 28212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.973160987931883e-08,
+      "loss": 0.3932,
+      "step": 28213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.929298033311217e-08,
+      "loss": 0.4293,
+      "step": 28214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.885556016054008e-08,
+      "loss": 0.3134,
+      "step": 28215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.841934936688722e-08,
+      "loss": 0.4031,
+      "step": 28216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.798434795746045e-08,
+      "loss": 0.4231,
+      "step": 28217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.755055593750004e-08,
+      "loss": 0.4302,
+      "step": 28218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.711797331226845e-08,
+      "loss": 0.4061,
+      "step": 28219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.66866000869837e-08,
+      "loss": 0.3764,
+      "step": 28220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.625643626689715e-08,
+      "loss": 0.3994,
+      "step": 28221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.582748185719358e-08,
+      "loss": 0.4586,
+      "step": 28222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.539973686306879e-08,
+      "loss": 0.3776,
+      "step": 28223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.497320128969643e-08,
+      "loss": 0.3742,
+      "step": 28224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.454787514225014e-08,
+      "loss": 0.4354,
+      "step": 28225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.412375842587028e-08,
+      "loss": 0.4491,
+      "step": 28226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.370085114568604e-08,
+      "loss": 0.3789,
+      "step": 28227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.327915330682666e-08,
+      "loss": 0.4292,
+      "step": 28228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.285866491438808e-08,
+      "loss": 0.3767,
+      "step": 28229
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.243938597345511e-08,
+      "loss": 0.3816,
+      "step": 28230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.202131648911259e-08,
+      "loss": 0.4573,
+      "step": 28231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.160445646642311e-08,
+      "loss": 0.3181,
+      "step": 28232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.118880591042709e-08,
+      "loss": 0.4566,
+      "step": 28233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.077436482614275e-08,
+      "loss": 0.38,
+      "step": 28234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.036113321861049e-08,
+      "loss": 0.4336,
+      "step": 28235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.994911109281522e-08,
+      "loss": 0.416,
+      "step": 28236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.953829845374182e-08,
+      "loss": 0.4138,
+      "step": 28237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.912869530636412e-08,
+      "loss": 0.3981,
+      "step": 28238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.87203016556559e-08,
+      "loss": 0.435,
+      "step": 28239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.831311750654656e-08,
+      "loss": 0.4097,
+      "step": 28240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.790714286396549e-08,
+      "loss": 0.515,
+      "step": 28241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.750237773281987e-08,
+      "loss": 0.3825,
+      "step": 28242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.709882211801688e-08,
+      "loss": 0.426,
+      "step": 28243
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.669647602444152e-08,
+      "loss": 0.4073,
+      "step": 28244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.629533945696764e-08,
+      "loss": 0.3453,
+      "step": 28245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.589541242043585e-08,
+      "loss": 0.4561,
+      "step": 28246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.549669491969778e-08,
+      "loss": 0.3242,
+      "step": 28247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.509918695957184e-08,
+      "loss": 0.3994,
+      "step": 28248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.470288854487639e-08,
+      "loss": 0.4276,
+      "step": 28249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.430779968040757e-08,
+      "loss": 0.4447,
+      "step": 28250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.391392037095046e-08,
+      "loss": 0.3694,
+      "step": 28251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.35212506212568e-08,
+      "loss": 0.3859,
+      "step": 28252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.312979043608946e-08,
+      "loss": 0.4539,
+      "step": 28253
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.27395398201891e-08,
+      "loss": 0.4585,
+      "step": 28254
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.235049877828524e-08,
+      "loss": 0.3843,
+      "step": 28255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.196266731506306e-08,
+      "loss": 0.4464,
+      "step": 28256
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.157604543524098e-08,
+      "loss": 0.4617,
+      "step": 28257
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.119063314348194e-08,
+      "loss": 0.3693,
+      "step": 28258
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.080643044445999e-08,
+      "loss": 0.4042,
+      "step": 28259
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.042343734282696e-08,
+      "loss": 0.4933,
+      "step": 28260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.004165384320137e-08,
+      "loss": 0.4029,
+      "step": 28261
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.966107995022397e-08,
+      "loss": 0.4033,
+      "step": 28262
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.928171566850216e-08,
+      "loss": 0.4706,
+      "step": 28263
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.890356100261008e-08,
+      "loss": 0.3619,
+      "step": 28264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8526615957132935e-08,
+      "loss": 0.4432,
+      "step": 28265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.815088053663376e-08,
+      "loss": 0.3843,
+      "step": 28266
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.777635474566445e-08,
+      "loss": 0.3753,
+      "step": 28267
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.740303858874363e-08,
+      "loss": 0.3796,
+      "step": 28268
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.70309320704121e-08,
+      "loss": 0.3218,
+      "step": 28269
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.666003519515517e-08,
+      "loss": 0.38,
+      "step": 28270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.629034796746924e-08,
+      "loss": 0.4195,
+      "step": 28271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.59218703918285e-08,
+      "loss": 0.4135,
+      "step": 28272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.555460247268496e-08,
+      "loss": 0.3968,
+      "step": 28273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.51885442145017e-08,
+      "loss": 0.4541,
+      "step": 28274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.4823695621686324e-08,
+      "loss": 0.3961,
+      "step": 28275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.4460056698668606e-08,
+      "loss": 0.4224,
+      "step": 28276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.409762744985614e-08,
+      "loss": 0.3695,
+      "step": 28277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3736407879623195e-08,
+      "loss": 0.4468,
+      "step": 28278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.337639799234406e-08,
+      "loss": 0.3639,
+      "step": 28279
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.301759779238191e-08,
+      "loss": 0.4209,
+      "step": 28280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2660007284066616e-08,
+      "loss": 0.4154,
+      "step": 28281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.230362647173914e-08,
+      "loss": 0.3888,
+      "step": 28282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.194845535970716e-08,
+      "loss": 0.3784,
+      "step": 28283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.159449395226723e-08,
+      "loss": 0.3836,
+      "step": 28284
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.124174225371592e-08,
+      "loss": 0.4729,
+      "step": 28285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.089020026831648e-08,
+      "loss": 0.3216,
+      "step": 28286
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0539868000309967e-08,
+      "loss": 0.4208,
+      "step": 28287
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.019074545394853e-08,
+      "loss": 0.3968,
+      "step": 28288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9842832633462124e-08,
+      "loss": 0.3903,
+      "step": 28289
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.949612954305849e-08,
+      "loss": 0.4056,
+      "step": 28290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.915063618692317e-08,
+      "loss": 0.4138,
+      "step": 28291
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8806352569252814e-08,
+      "loss": 0.4236,
+      "step": 28292
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.846327869419964e-08,
+      "loss": 0.4287,
+      "step": 28293
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8121414565927e-08,
+      "loss": 0.393,
+      "step": 28294
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.778076018857602e-08,
+      "loss": 0.4337,
+      "step": 28295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.744131556626563e-08,
+      "loss": 0.4229,
+      "step": 28296
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.710308070309255e-08,
+      "loss": 0.3755,
+      "step": 28297
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.67660556031646e-08,
+      "loss": 0.3722,
+      "step": 28298
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6430240270556316e-08,
+      "loss": 0.4214,
+      "step": 28299
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.609563470933109e-08,
+      "loss": 0.3915,
+      "step": 28300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.576223892354126e-08,
+      "loss": 0.3504,
+      "step": 28301
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.543005291722802e-08,
+      "loss": 0.364,
+      "step": 28302
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.509907669438818e-08,
+      "loss": 0.4164,
+      "step": 28303
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4769310259062944e-08,
+      "loss": 0.4321,
+      "step": 28304
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4440753615215823e-08,
+      "loss": 0.4672,
+      "step": 28305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.411340676683251e-08,
+      "loss": 0.3726,
+      "step": 28306
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3787269717865396e-08,
+      "loss": 0.4286,
+      "step": 28307
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3462342472289086e-08,
+      "loss": 0.3862,
+      "step": 28308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3138625034000455e-08,
+      "loss": 0.375,
+      "step": 28309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.28161174069519e-08,
+      "loss": 0.3476,
+      "step": 28310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.249481959501811e-08,
+      "loss": 0.3745,
+      "step": 28311
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.217473160210705e-08,
+      "loss": 0.4293,
+      "step": 28312
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1855853432082316e-08,
+      "loss": 0.3765,
+      "step": 28313
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.153818508881857e-08,
+      "loss": 0.4986,
+      "step": 28314
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.122172657613499e-08,
+      "loss": 0.3511,
+      "step": 28315
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0906477897884046e-08,
+      "loss": 0.37,
+      "step": 28316
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.05924390578738e-08,
+      "loss": 0.3875,
+      "step": 28317
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.027961005990122e-08,
+      "loss": 0.3756,
+      "step": 28318
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.996799090775216e-08,
+      "loss": 0.3682,
+      "step": 28319
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.965758160521249e-08,
+      "loss": 0.3972,
+      "step": 28320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.934838215602365e-08,
+      "loss": 0.3814,
+      "step": 28321
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.904039256392711e-08,
+      "loss": 0.3795,
+      "step": 28322
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.87336128326643e-08,
+      "loss": 0.437,
+      "step": 28323
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.842804296594338e-08,
+      "loss": 0.502,
+      "step": 28324
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8123682967450283e-08,
+      "loss": 0.3586,
+      "step": 28325
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7820532840882053e-08,
+      "loss": 0.3273,
+      "step": 28326
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7518592589913524e-08,
+      "loss": 0.444,
+      "step": 28327
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.721786221817514e-08,
+      "loss": 0.3738,
+      "step": 28328
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.691834172933062e-08,
+      "loss": 0.3032,
+      "step": 28329
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.66200311269993e-08,
+      "loss": 0.409,
+      "step": 28330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.63229304147783e-08,
+      "loss": 0.345,
+      "step": 28331
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6027039596286947e-08,
+      "loss": 0.4915,
+      "step": 28332
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.573235867507796e-08,
+      "loss": 0.3859,
+      "step": 28333
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5438887654737355e-08,
+      "loss": 0.3774,
+      "step": 28334
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.514662653881784e-08,
+      "loss": 0.4609,
+      "step": 28335
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.485557533084993e-08,
+      "loss": 0.4291,
+      "step": 28336
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.456573403436414e-08,
+      "loss": 0.3929,
+      "step": 28337
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.427710265285766e-08,
+      "loss": 0.4324,
+      "step": 28338
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.39896811898277e-08,
+      "loss": 0.3773,
+      "step": 28339
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.370346964876036e-08,
+      "loss": 0.4477,
+      "step": 28340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.341846803310844e-08,
+      "loss": 0.3478,
+      "step": 28341
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.313467634632472e-08,
+      "loss": 0.384,
+      "step": 28342
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2852094591850903e-08,
+      "loss": 0.4951,
+      "step": 28343
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.257072277309536e-08,
+      "loss": 0.3715,
+      "step": 28344
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2290560893477597e-08,
+      "loss": 0.3701,
+      "step": 28345
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2011608956372675e-08,
+      "loss": 0.4146,
+      "step": 28346
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1733866965177885e-08,
+      "loss": 0.337,
+      "step": 28347
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.145733492323499e-08,
+      "loss": 0.4093,
+      "step": 28348
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.118201283389688e-08,
+      "loss": 0.419,
+      "step": 28349
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0907900700494205e-08,
+      "loss": 0.3696,
+      "step": 28350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.063499852635765e-08,
+      "loss": 0.4213,
+      "step": 28351
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.036330631477347e-08,
+      "loss": 0.5459,
+      "step": 28352
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.009282406905012e-08,
+      "loss": 0.4621,
+      "step": 28353
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.982355179242946e-08,
+      "loss": 0.3806,
+      "step": 28354
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9555489488197752e-08,
+      "loss": 0.3716,
+      "step": 28355
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9288637159596845e-08,
+      "loss": 0.368,
+      "step": 28356
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9022994809846384e-08,
+      "loss": 0.3598,
+      "step": 28357
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8758562442166014e-08,
+      "loss": 0.4418,
+      "step": 28358
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8495340059753185e-08,
+      "loss": 0.3642,
+      "step": 28359
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8233327665794228e-08,
+      "loss": 0.4369,
+      "step": 28360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.797252526347549e-08,
+      "loss": 0.4854,
+      "step": 28361
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.77129328559278e-08,
+      "loss": 0.3878,
+      "step": 28362
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7454550446315287e-08,
+      "loss": 0.4999,
+      "step": 28363
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.719737803774658e-08,
+      "loss": 0.4155,
+      "step": 28364
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6941415633352508e-08,
+      "loss": 0.3379,
+      "step": 28365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.668666323621949e-08,
+      "loss": 0.4005,
+      "step": 28366
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6433120849433946e-08,
+      "loss": 0.4201,
+      "step": 28367
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6180788476071195e-08,
+      "loss": 0.4279,
+      "step": 28368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5929666119162144e-08,
+      "loss": 0.4106,
+      "step": 28369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.567975378178211e-08,
+      "loss": 0.3721,
+      "step": 28370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.54310514669176e-08,
+      "loss": 0.4738,
+      "step": 28371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5183559177610617e-08,
+      "loss": 0.3846,
+      "step": 28372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.493727691683656e-08,
+      "loss": 0.4492,
+      "step": 28373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4692204687593033e-08,
+      "loss": 0.3831,
+      "step": 28374
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.444834249283323e-08,
+      "loss": 0.3838,
+      "step": 28375
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4205690335499244e-08,
+      "loss": 0.4909,
+      "step": 28376
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3964248218555364e-08,
+      "loss": 0.4283,
+      "step": 28377
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3724016144910376e-08,
+      "loss": 0.4045,
+      "step": 28378
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3484994117461966e-08,
+      "loss": 0.3175,
+      "step": 28379
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3247182139107815e-08,
+      "loss": 0.4345,
+      "step": 28380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.30105802127456e-08,
+      "loss": 0.356,
+      "step": 28381
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.27751883412064e-08,
+      "loss": 0.3689,
+      "step": 28382
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2541006527365683e-08,
+      "loss": 0.3665,
+      "step": 28383
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.230803477405452e-08,
+      "loss": 0.3874,
+      "step": 28384
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2076273084070677e-08,
+      "loss": 0.4043,
+      "step": 28385
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1845721460245218e-08,
+      "loss": 0.4233,
+      "step": 28386
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1616379905364802e-08,
+      "loss": 0.3653,
+      "step": 28387
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.138824842219389e-08,
+      "loss": 0.4363,
+      "step": 28388
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1161327013496936e-08,
+      "loss": 0.4178,
+      "step": 28389
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0935615682016185e-08,
+      "loss": 0.4369,
+      "step": 28390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0711114430504998e-08,
+      "loss": 0.4232,
+      "step": 28391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0487823261650106e-08,
+      "loss": 0.365,
+      "step": 28392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0265742178182666e-08,
+      "loss": 0.395,
+      "step": 28393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004487118276721e-08,
+      "loss": 0.3835,
+      "step": 28394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9825210278090477e-08,
+      "loss": 0.489,
+      "step": 28395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9606759466805903e-08,
+      "loss": 0.4689,
+      "step": 28396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9389518751566917e-08,
+      "loss": 0.3643,
+      "step": 28397
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.917348813498254e-08,
+      "loss": 0.3753,
+      "step": 28398
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.895866761968401e-08,
+      "loss": 0.501,
+      "step": 28399
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8745057208269247e-08,
+      "loss": 0.3973,
+      "step": 28400
+    },
+    {
+      "epoch": 0.99,
+      "eval_loss": 0.3963080644607544,
+      "eval_runtime": 851.0338,
+      "eval_samples_per_second": 28.271,
+      "eval_steps_per_second": 14.136,
+      "step": 28400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8532656903325064e-08,
+      "loss": 0.3962,
+      "step": 28401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.832146670741608e-08,
+      "loss": 0.4174,
+      "step": 28402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.811148662310691e-08,
+      "loss": 0.4523,
+      "step": 28403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7902716652928863e-08,
+      "loss": 0.3672,
+      "step": 28404
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7695156799402145e-08,
+      "loss": 0.4229,
+      "step": 28405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7488807065069167e-08,
+      "loss": 0.3762,
+      "step": 28406
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7283667452394624e-08,
+      "loss": 0.4722,
+      "step": 28407
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.707973796387652e-08,
+      "loss": 0.456,
+      "step": 28408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6877018601990645e-08,
+      "loss": 0.3896,
+      "step": 28409
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6675509369179498e-08,
+      "loss": 0.4397,
+      "step": 28410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6475210267874465e-08,
+      "loss": 0.3749,
+      "step": 28411
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6276121300518034e-08,
+      "loss": 0.4978,
+      "step": 28412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6078242469508286e-08,
+      "loss": 0.3893,
+      "step": 28413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5881573777243308e-08,
+      "loss": 0.4245,
+      "step": 28414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5686115226098974e-08,
+      "loss": 0.3968,
+      "step": 28415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.549186681845116e-08,
+      "loss": 0.4995,
+      "step": 28416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5298828556642443e-08,
+      "loss": 0.3712,
+      "step": 28417
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5107000443015385e-08,
+      "loss": 0.3942,
+      "step": 28418
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4916382479890356e-08,
+      "loss": 0.4036,
+      "step": 28419
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.472697466956552e-08,
+      "loss": 0.4552,
+      "step": 28420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4538777014339034e-08,
+      "loss": 0.3783,
+      "step": 28421
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.435178951649796e-08,
+      "loss": 0.4252,
+      "step": 28422
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4166012178296051e-08,
+      "loss": 0.3485,
+      "step": 28423
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3981445001975957e-08,
+      "loss": 0.3971,
+      "step": 28424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3798087989780328e-08,
+      "loss": 0.4368,
+      "step": 28425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3615941143929612e-08,
+      "loss": 0.3439,
+      "step": 28426
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.343500446663315e-08,
+      "loss": 0.3992,
+      "step": 28427
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.325527796006698e-08,
+      "loss": 0.3813,
+      "step": 28428
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3076761626407141e-08,
+      "loss": 0.4092,
+      "step": 28429
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2899455467829669e-08,
+      "loss": 0.4047,
+      "step": 28430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2723359486455089e-08,
+      "loss": 0.3474,
+      "step": 28431
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2548473684448336e-08,
+      "loss": 0.3588,
+      "step": 28432
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.237479806389663e-08,
+      "loss": 0.3605,
+      "step": 28433
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2202332626920498e-08,
+      "loss": 0.4038,
+      "step": 28434
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2031077375596056e-08,
+      "loss": 0.4576,
+      "step": 28435
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1861032311999421e-08,
+      "loss": 0.4275,
+      "step": 28436
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1692197438195607e-08,
+      "loss": 0.356,
+      "step": 28437
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1524572756216323e-08,
+      "loss": 0.3988,
+      "step": 28438
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.135815826810438e-08,
+      "loss": 0.3701,
+      "step": 28439
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1192953975858178e-08,
+      "loss": 0.3737,
+      "step": 28440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1028959881487223e-08,
+      "loss": 0.3892,
+      "step": 28441
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.086617598697881e-08,
+      "loss": 0.3456,
+      "step": 28442
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0704602294286936e-08,
+      "loss": 0.356,
+      "step": 28443
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0544238805387796e-08,
+      "loss": 0.4202,
+      "step": 28444
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0385085522213178e-08,
+      "loss": 0.5276,
+      "step": 28445
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0227142446683768e-08,
+      "loss": 0.4417,
+      "step": 28446
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0070409580720253e-08,
+      "loss": 0.3673,
+      "step": 28447
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.914886926221113e-09,
+      "loss": 0.4756,
+      "step": 28448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.760574485051522e-09,
+      "loss": 0.3789,
+      "step": 28449
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.60747225909886e-09,
+      "loss": 0.4546,
+      "step": 28450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.455580250206097e-09,
+      "loss": 0.432,
+      "step": 28451
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.304898460216204e-09,
+      "loss": 0.3833,
+      "step": 28452
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.15542689096105e-09,
+      "loss": 0.4434,
+      "step": 28453
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.00716554422809e-09,
+      "loss": 0.4035,
+      "step": 28454
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.860114421826993e-09,
+      "loss": 0.4213,
+      "step": 28455
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.714273525534112e-09,
+      "loss": 0.3837,
+      "step": 28456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.5696428571036e-09,
+      "loss": 0.3849,
+      "step": 28457
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.426222418311814e-09,
+      "loss": 0.3845,
+      "step": 28458
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.284012210868497e-09,
+      "loss": 0.3362,
+      "step": 28459
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.1430122365167e-09,
+      "loss": 0.4963,
+      "step": 28460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.003222496943963e-09,
+      "loss": 0.3309,
+      "step": 28461
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.864642993860027e-09,
+      "loss": 0.4274,
+      "step": 28462
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.727273728919126e-09,
+      "loss": 0.382,
+      "step": 28463
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.5911147038088e-09,
+      "loss": 0.3988,
+      "step": 28464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.456165920161074e-09,
+      "loss": 0.4408,
+      "step": 28465
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.322427379607977e-09,
+      "loss": 0.3852,
+      "step": 28466
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.189899083781537e-09,
+      "loss": 0.3546,
+      "step": 28467
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.058581034280476e-09,
+      "loss": 0.3996,
+      "step": 28468
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.9284732326924116e-09,
+      "loss": 0.4265,
+      "step": 28469
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.799575680582759e-09,
+      "loss": 0.4523,
+      "step": 28470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.671888379528035e-09,
+      "loss": 0.3742,
+      "step": 28471
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.545411331060347e-09,
+      "loss": 0.3725,
+      "step": 28472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.420144536722905e-09,
+      "loss": 0.4188,
+      "step": 28473
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.296087998025613e-09,
+      "loss": 0.4964,
+      "step": 28474
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.17324171646727e-09,
+      "loss": 0.3428,
+      "step": 28475
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.051605693535578e-09,
+      "loss": 0.4072,
+      "step": 28476
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.931179930707131e-09,
+      "loss": 0.4361,
+      "step": 28477
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.811964429425221e-09,
+      "loss": 0.3906,
+      "step": 28478
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.6939591911553405e-09,
+      "loss": 0.4242,
+      "step": 28479
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.577164217318576e-09,
+      "loss": 0.4167,
+      "step": 28480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.4615795093138075e-09,
+      "loss": 0.4732,
+      "step": 28481
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.347205068562122e-09,
+      "loss": 0.4765,
+      "step": 28482
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.234040896429093e-09,
+      "loss": 0.4261,
+      "step": 28483
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.122086994291397e-09,
+      "loss": 0.422,
+      "step": 28484
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.011343363503506e-09,
+      "loss": 0.4943,
+      "step": 28485
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.901810005408791e-09,
+      "loss": 0.3761,
+      "step": 28486
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.793486921328416e-09,
+      "loss": 0.4159,
+      "step": 28487
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.686374112583547e-09,
+      "loss": 0.4142,
+      "step": 28488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5804715804509404e-09,
+      "loss": 0.346,
+      "step": 28489
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.475779326229557e-09,
+      "loss": 0.4073,
+      "step": 28490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.372297351185051e-09,
+      "loss": 0.4042,
+      "step": 28491
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.270025656560872e-09,
+      "loss": 0.3536,
+      "step": 28492
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.168964243611573e-09,
+      "loss": 0.35,
+      "step": 28493
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.069113113536194e-09,
+      "loss": 0.3765,
+      "step": 28494
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9704722675559805e-09,
+      "loss": 0.3886,
+      "step": 28495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8730417068699734e-09,
+      "loss": 0.4716,
+      "step": 28496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.776821432655009e-09,
+      "loss": 0.4125,
+      "step": 28497
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6818114460657195e-09,
+      "loss": 0.5065,
+      "step": 28498
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5880117482678388e-09,
+      "loss": 0.3922,
+      "step": 28499
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4954223403826926e-09,
+      "loss": 0.3817,
+      "step": 28500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4040432235316055e-09,
+      "loss": 0.4148,
+      "step": 28501
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3138743988359034e-09,
+      "loss": 0.5068,
+      "step": 28502
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2249158673725022e-09,
+      "loss": 0.3637,
+      "step": 28503
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1371676302183184e-09,
+      "loss": 0.3357,
+      "step": 28504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.050629688439166e-09,
+      "loss": 0.4162,
+      "step": 28505
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.965302043078655e-09,
+      "loss": 0.4393,
+      "step": 28506
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8811846951803945e-09,
+      "loss": 0.4675,
+      "step": 28507
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.798277645754688e-09,
+      "loss": 0.3313,
+      "step": 28508
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7165808958007353e-09,
+      "loss": 0.4385,
+      "step": 28509
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6360944463177384e-09,
+      "loss": 0.4432,
+      "step": 28510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5568182982715905e-09,
+      "loss": 0.3404,
+      "step": 28511
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.478752452628186e-09,
+      "loss": 0.3601,
+      "step": 28512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4018969103201117e-09,
+      "loss": 0.3643,
+      "step": 28513
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3262516722910575e-09,
+      "loss": 0.4873,
+      "step": 28514
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2518167394514065e-09,
+      "loss": 0.383,
+      "step": 28515
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1785921127004396e-09,
+      "loss": 0.3477,
+      "step": 28516
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.106577792926334e-09,
+      "loss": 0.4118,
+      "step": 28517
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0357737810061674e-09,
+      "loss": 0.4262,
+      "step": 28518
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.966180077783708e-09,
+      "loss": 0.4123,
+      "step": 28519
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8977966841138284e-09,
+      "loss": 0.4239,
+      "step": 28520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8306236008180934e-09,
+      "loss": 0.395,
+      "step": 28521
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7646608287069655e-09,
+      "loss": 0.4477,
+      "step": 28522
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.699908368590908e-09,
+      "loss": 0.4197,
+      "step": 28523
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6363662212359742e-09,
+      "loss": 0.3608,
+      "step": 28524
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.574034387430423e-09,
+      "loss": 0.3523,
+      "step": 28525
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5129128679070015e-09,
+      "loss": 0.4141,
+      "step": 28526
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4530016634317633e-09,
+      "loss": 0.5187,
+      "step": 28527
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3943007747041492e-09,
+      "loss": 0.4281,
+      "step": 28528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.336810202445804e-09,
+      "loss": 0.4407,
+      "step": 28529
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2805299473561683e-09,
+      "loss": 0.4462,
+      "step": 28530
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.225460010112478e-09,
+      "loss": 0.4579,
+      "step": 28531
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.171600391380867e-09,
+      "loss": 0.3735,
+      "step": 28532
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.118951091816367e-09,
+      "loss": 0.4174,
+      "step": 28533
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.067512112051805e-09,
+      "loss": 0.3486,
+      "step": 28534
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.017283452708906e-09,
+      "loss": 0.4149,
+      "step": 28535
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.682651144093946e-10,
+      "loss": 0.43,
+      "step": 28536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.204570977194848e-10,
+      "loss": 0.474,
+      "step": 28537
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.738594032497993e-10,
+      "loss": 0.3397,
+      "step": 28538
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.28472031533245e-10,
+      "loss": 0.4553,
+      "step": 28539
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.84294983147138e-10,
+      "loss": 0.4766,
+      "step": 28540
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.413282586132831e-10,
+      "loss": 0.4215,
+      "step": 28541
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.995718584423827e-10,
+      "loss": 0.3451,
+      "step": 28542
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.590257831562419e-10,
+      "loss": 0.3845,
+      "step": 28543
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.196900332322564e-10,
+      "loss": 0.379,
+      "step": 28544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.815646091589244e-10,
+      "loss": 0.3927,
+      "step": 28545
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.446495113914374e-10,
+      "loss": 0.4438,
+      "step": 28546
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.089447403738845e-10,
+      "loss": 0.37,
+      "step": 28547
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.744502965392528e-10,
+      "loss": 0.4062,
+      "step": 28548
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4116618032052916e-10,
+      "loss": 0.417,
+      "step": 28549
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.090923920951895e-10,
+      "loss": 0.3917,
+      "step": 28550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.78228932262914e-10,
+      "loss": 0.3623,
+      "step": 28551
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.485758012011786e-10,
+      "loss": 0.4304,
+      "step": 28552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2013299926525463e-10,
+      "loss": 0.4665,
+      "step": 28553
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9290052679931124e-10,
+      "loss": 0.4551,
+      "step": 28554
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6687838413641533e-10,
+      "loss": 0.4735,
+      "step": 28555
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.420665715874293e-10,
+      "loss": 0.36,
+      "step": 28556
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1846508945211342e-10,
+      "loss": 0.3995,
+      "step": 28557
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9607393800802343e-10,
+      "loss": 0.3255,
+      "step": 28558
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7489311754381732e-10,
+      "loss": 0.436,
+      "step": 28559
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5492262831484638e-10,
+      "loss": 0.379,
+      "step": 28560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3616247055425745e-10,
+      "loss": 0.4143,
+      "step": 28561
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1861264448409514e-10,
+      "loss": 0.3755,
+      "step": 28562
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0227315032640406e-10,
+      "loss": 0.4294,
+      "step": 28563
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.71439882699221e-11,
+      "loss": 0.4325,
+      "step": 28564
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.322515851448941e-11,
+      "loss": 0.4234,
+      "step": 28565
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.051666120443499e-11,
+      "loss": 0.4007,
+      "step": 28566
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.901849651739454e-11,
+      "loss": 0.3894,
+      "step": 28567
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.873066457549257e-11,
+      "loss": 0.402,
+      "step": 28568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9653165511955846e-11,
+      "loss": 0.4283,
+      "step": 28569
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.178599943780668e-11,
+      "loss": 0.378,
+      "step": 28570
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5129166441862907e-11,
+      "loss": 0.4071,
+      "step": 28571
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.68266661294237e-12,
+      "loss": 0.4696,
+      "step": 28572
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.446500006556221e-12,
+      "loss": 0.4218,
+      "step": 28573
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4206666782156106e-12,
+      "loss": 0.401,
+      "step": 28574
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.051666723294603e-13,
+      "loss": 0.4331,
+      "step": 28575
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.402,
+      "step": 28576
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 28576,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 2.734103779564585e+19,
+  "trial_name": null,
+  "trial_params": null
+}