diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,17207 @@
+{
+  "best_metric": 0.5207100591715976,
+  "best_model_checkpoint": "videomae-base-finetuned-kisa/checkpoint-1320",
+  "epoch": 21.040366972477063,
+  "eval_steps": 500,
+  "global_step": 2420,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0003669724770642202,
+      "grad_norm": 3.6520371437072754,
+      "learning_rate": 5e-05,
+      "loss": 0.6833,
+      "step": 1
+    },
+    {
+      "epoch": 0.0007339449541284404,
+      "grad_norm": 4.897116184234619,
+      "learning_rate": 5e-05,
+      "loss": 0.5987,
+      "step": 2
+    },
+    {
+      "epoch": 0.0011009174311926607,
+      "grad_norm": 4.032791614532471,
+      "learning_rate": 5e-05,
+      "loss": 0.6721,
+      "step": 3
+    },
+    {
+      "epoch": 0.0014678899082568807,
+      "grad_norm": 11.829678535461426,
+      "learning_rate": 5e-05,
+      "loss": 0.8827,
+      "step": 4
+    },
+    {
+      "epoch": 0.001834862385321101,
+      "grad_norm": 4.306095600128174,
+      "learning_rate": 5e-05,
+      "loss": 0.426,
+      "step": 5
+    },
+    {
+      "epoch": 0.0022018348623853213,
+      "grad_norm": 3.91937518119812,
+      "learning_rate": 5e-05,
+      "loss": 0.4953,
+      "step": 6
+    },
+    {
+      "epoch": 0.002568807339449541,
+      "grad_norm": 3.876131534576416,
+      "learning_rate": 5e-05,
+      "loss": 0.4024,
+      "step": 7
+    },
+    {
+      "epoch": 0.0029357798165137615,
+      "grad_norm": 7.274229526519775,
+      "learning_rate": 5e-05,
+      "loss": 0.7389,
+      "step": 8
+    },
+    {
+      "epoch": 0.0033027522935779817,
+      "grad_norm": 8.303215980529785,
+      "learning_rate": 5e-05,
+      "loss": 0.6945,
+      "step": 9
+    },
+    {
+      "epoch": 0.003669724770642202,
+      "grad_norm": 3.1192452907562256,
+      "learning_rate": 5e-05,
+      "loss": 0.2942,
+      "step": 10
+    },
+    {
+      "epoch": 0.004036697247706422,
+      "grad_norm": 5.009522914886475,
+      "learning_rate": 5e-05,
+      "loss": 0.3938,
+      "step": 11
+    },
+    {
+      "epoch": 0.004403669724770643,
+      "grad_norm": 11.767984390258789,
+      "learning_rate": 5e-05,
+      "loss": 0.3367,
+      "step": 12
+    },
+    {
+      "epoch": 0.0047706422018348625,
+      "grad_norm": 5.413540363311768,
+      "learning_rate": 5e-05,
+      "loss": 0.4643,
+      "step": 13
+    },
+    {
+      "epoch": 0.005137614678899082,
+      "grad_norm": 4.5803704261779785,
+      "learning_rate": 5e-05,
+      "loss": 0.3817,
+      "step": 14
+    },
+    {
+      "epoch": 0.005504587155963303,
+      "grad_norm": 3.5025928020477295,
+      "learning_rate": 5e-05,
+      "loss": 0.1948,
+      "step": 15
+    },
+    {
+      "epoch": 0.005871559633027523,
+      "grad_norm": 7.923089027404785,
+      "learning_rate": 5e-05,
+      "loss": 0.3397,
+      "step": 16
+    },
+    {
+      "epoch": 0.006238532110091743,
+      "grad_norm": 5.565113067626953,
+      "learning_rate": 5e-05,
+      "loss": 0.3566,
+      "step": 17
+    },
+    {
+      "epoch": 0.0066055045871559635,
+      "grad_norm": 9.009073257446289,
+      "learning_rate": 5e-05,
+      "loss": 0.6565,
+      "step": 18
+    },
+    {
+      "epoch": 0.006972477064220183,
+      "grad_norm": 4.929457187652588,
+      "learning_rate": 5e-05,
+      "loss": 0.4387,
+      "step": 19
+    },
+    {
+      "epoch": 0.007339449541284404,
+      "grad_norm": 1.428413987159729,
+      "learning_rate": 5e-05,
+      "loss": 0.0621,
+      "step": 20
+    },
+    {
+      "epoch": 0.007706422018348624,
+      "grad_norm": 2.003269672393799,
+      "learning_rate": 5e-05,
+      "loss": 0.0793,
+      "step": 21
+    },
+    {
+      "epoch": 0.008073394495412844,
+      "grad_norm": 4.0943427085876465,
+      "learning_rate": 5e-05,
+      "loss": 0.2708,
+      "step": 22
+    },
+    {
+      "epoch": 0.008440366972477064,
+      "grad_norm": 3.9067351818084717,
+      "learning_rate": 5e-05,
+      "loss": 0.4436,
+      "step": 23
+    },
+    {
+      "epoch": 0.008807339449541285,
+      "grad_norm": 6.453662395477295,
+      "learning_rate": 5e-05,
+      "loss": 0.3319,
+      "step": 24
+    },
+    {
+      "epoch": 0.009174311926605505,
+      "grad_norm": 9.729504585266113,
+      "learning_rate": 5e-05,
+      "loss": 0.8182,
+      "step": 25
+    },
+    {
+      "epoch": 0.009541284403669725,
+      "grad_norm": 10.723071098327637,
+      "learning_rate": 5e-05,
+      "loss": 0.8879,
+      "step": 26
+    },
+    {
+      "epoch": 0.009908256880733945,
+      "grad_norm": 13.677897453308105,
+      "learning_rate": 5e-05,
+      "loss": 0.2614,
+      "step": 27
+    },
+    {
+      "epoch": 0.010275229357798165,
+      "grad_norm": 4.850866794586182,
+      "learning_rate": 5e-05,
+      "loss": 0.1127,
+      "step": 28
+    },
+    {
+      "epoch": 0.010642201834862385,
+      "grad_norm": 2.1186203956604004,
+      "learning_rate": 5e-05,
+      "loss": 0.0717,
+      "step": 29
+    },
+    {
+      "epoch": 0.011009174311926606,
+      "grad_norm": 12.493134498596191,
+      "learning_rate": 5e-05,
+      "loss": 0.2431,
+      "step": 30
+    },
+    {
+      "epoch": 0.011376146788990826,
+      "grad_norm": 7.8452067375183105,
+      "learning_rate": 5e-05,
+      "loss": 0.3278,
+      "step": 31
+    },
+    {
+      "epoch": 0.011743119266055046,
+      "grad_norm": 34.506736755371094,
+      "learning_rate": 5e-05,
+      "loss": 0.5556,
+      "step": 32
+    },
+    {
+      "epoch": 0.012110091743119266,
+      "grad_norm": 8.73371410369873,
+      "learning_rate": 5e-05,
+      "loss": 0.1854,
+      "step": 33
+    },
+    {
+      "epoch": 0.012477064220183486,
+      "grad_norm": 2.640326499938965,
+      "learning_rate": 5e-05,
+      "loss": 0.0779,
+      "step": 34
+    },
+    {
+      "epoch": 0.012844036697247707,
+      "grad_norm": 11.442755699157715,
+      "learning_rate": 5e-05,
+      "loss": 0.0736,
+      "step": 35
+    },
+    {
+      "epoch": 0.013211009174311927,
+      "grad_norm": 5.950634479522705,
+      "learning_rate": 5e-05,
+      "loss": 0.0595,
+      "step": 36
+    },
+    {
+      "epoch": 0.013577981651376147,
+      "grad_norm": 4.315831184387207,
+      "learning_rate": 5e-05,
+      "loss": 0.3076,
+      "step": 37
+    },
+    {
+      "epoch": 0.013944954128440367,
+      "grad_norm": 19.644577026367188,
+      "learning_rate": 5e-05,
+      "loss": 0.5271,
+      "step": 38
+    },
+    {
+      "epoch": 0.014311926605504587,
+      "grad_norm": 2.4374496936798096,
+      "learning_rate": 5e-05,
+      "loss": 0.0613,
+      "step": 39
+    },
+    {
+      "epoch": 0.014678899082568808,
+      "grad_norm": 0.5387063026428223,
+      "learning_rate": 5e-05,
+      "loss": 0.0228,
+      "step": 40
+    },
+    {
+      "epoch": 0.015045871559633028,
+      "grad_norm": 5.887943267822266,
+      "learning_rate": 5e-05,
+      "loss": 0.5045,
+      "step": 41
+    },
+    {
+      "epoch": 0.015412844036697248,
+      "grad_norm": 6.322305202484131,
+      "learning_rate": 5e-05,
+      "loss": 0.2003,
+      "step": 42
+    },
+    {
+      "epoch": 0.01577981651376147,
+      "grad_norm": 14.017230987548828,
+      "learning_rate": 5e-05,
+      "loss": 0.2012,
+      "step": 43
+    },
+    {
+      "epoch": 0.016146788990825688,
+      "grad_norm": 6.268453598022461,
+      "learning_rate": 5e-05,
+      "loss": 0.3335,
+      "step": 44
+    },
+    {
+      "epoch": 0.01651376146788991,
+      "grad_norm": 11.06840991973877,
+      "learning_rate": 5e-05,
+      "loss": 0.2125,
+      "step": 45
+    },
+    {
+      "epoch": 0.016880733944954127,
+      "grad_norm": 2.9422080516815186,
+      "learning_rate": 5e-05,
+      "loss": 0.0381,
+      "step": 46
+    },
+    {
+      "epoch": 0.01724770642201835,
+      "grad_norm": 3.210777759552002,
+      "learning_rate": 5e-05,
+      "loss": 0.0305,
+      "step": 47
+    },
+    {
+      "epoch": 0.01761467889908257,
+      "grad_norm": 3.826173782348633,
+      "learning_rate": 5e-05,
+      "loss": 0.4479,
+      "step": 48
+    },
+    {
+      "epoch": 0.01798165137614679,
+      "grad_norm": 5.312827110290527,
+      "learning_rate": 5e-05,
+      "loss": 0.734,
+      "step": 49
+    },
+    {
+      "epoch": 0.01834862385321101,
+      "grad_norm": 1.9761258363723755,
+      "learning_rate": 5e-05,
+      "loss": 0.0622,
+      "step": 50
+    },
+    {
+      "epoch": 0.01871559633027523,
+      "grad_norm": 2.034306764602661,
+      "learning_rate": 5e-05,
+      "loss": 0.0544,
+      "step": 51
+    },
+    {
+      "epoch": 0.01908256880733945,
+      "grad_norm": 25.11560821533203,
+      "learning_rate": 5e-05,
+      "loss": 0.8473,
+      "step": 52
+    },
+    {
+      "epoch": 0.019449541284403668,
+      "grad_norm": 0.31734830141067505,
+      "learning_rate": 5e-05,
+      "loss": 0.0136,
+      "step": 53
+    },
+    {
+      "epoch": 0.01981651376146789,
+      "grad_norm": 1.6658449172973633,
+      "learning_rate": 5e-05,
+      "loss": 0.0467,
+      "step": 54
+    },
+    {
+      "epoch": 0.02018348623853211,
+      "grad_norm": 4.054420471191406,
+      "learning_rate": 5e-05,
+      "loss": 0.4589,
+      "step": 55
+    },
+    {
+      "epoch": 0.02055045871559633,
+      "grad_norm": 4.88888692855835,
+      "learning_rate": 5e-05,
+      "loss": 0.3952,
+      "step": 56
+    },
+    {
+      "epoch": 0.02091743119266055,
+      "grad_norm": 3.6318199634552,
+      "learning_rate": 5e-05,
+      "loss": 0.5339,
+      "step": 57
+    },
+    {
+      "epoch": 0.02128440366972477,
+      "grad_norm": 6.343443393707275,
+      "learning_rate": 5e-05,
+      "loss": 0.0726,
+      "step": 58
+    },
+    {
+      "epoch": 0.02165137614678899,
+      "grad_norm": 3.2057759761810303,
+      "learning_rate": 5e-05,
+      "loss": 0.4775,
+      "step": 59
+    },
+    {
+      "epoch": 0.022018348623853212,
+      "grad_norm": 0.8520527482032776,
+      "learning_rate": 5e-05,
+      "loss": 0.0307,
+      "step": 60
+    },
+    {
+      "epoch": 0.02238532110091743,
+      "grad_norm": 9.82697582244873,
+      "learning_rate": 5e-05,
+      "loss": 0.4445,
+      "step": 61
+    },
+    {
+      "epoch": 0.022752293577981652,
+      "grad_norm": 2.9992215633392334,
+      "learning_rate": 5e-05,
+      "loss": 0.336,
+      "step": 62
+    },
+    {
+      "epoch": 0.02311926605504587,
+      "grad_norm": 3.1180429458618164,
+      "learning_rate": 5e-05,
+      "loss": 0.1017,
+      "step": 63
+    },
+    {
+      "epoch": 0.02348623853211009,
+      "grad_norm": 2.5434699058532715,
+      "learning_rate": 5e-05,
+      "loss": 0.0647,
+      "step": 64
+    },
+    {
+      "epoch": 0.023853211009174313,
+      "grad_norm": 1.4112409353256226,
+      "learning_rate": 5e-05,
+      "loss": 0.0575,
+      "step": 65
+    },
+    {
+      "epoch": 0.02422018348623853,
+      "grad_norm": 4.961060523986816,
+      "learning_rate": 5e-05,
+      "loss": 0.3283,
+      "step": 66
+    },
+    {
+      "epoch": 0.024587155963302753,
+      "grad_norm": 1.3271359205245972,
+      "learning_rate": 5e-05,
+      "loss": 0.0486,
+      "step": 67
+    },
+    {
+      "epoch": 0.02495412844036697,
+      "grad_norm": 12.50844669342041,
+      "learning_rate": 5e-05,
+      "loss": 0.4417,
+      "step": 68
+    },
+    {
+      "epoch": 0.025321100917431193,
+      "grad_norm": 6.534421443939209,
+      "learning_rate": 5e-05,
+      "loss": 0.6418,
+      "step": 69
+    },
+    {
+      "epoch": 0.025688073394495414,
+      "grad_norm": 3.0277347564697266,
+      "learning_rate": 5e-05,
+      "loss": 0.135,
+      "step": 70
+    },
+    {
+      "epoch": 0.026055045871559632,
+      "grad_norm": 8.132633209228516,
+      "learning_rate": 5e-05,
+      "loss": 0.1136,
+      "step": 71
+    },
+    {
+      "epoch": 0.026422018348623854,
+      "grad_norm": 19.286867141723633,
+      "learning_rate": 5e-05,
+      "loss": 0.5652,
+      "step": 72
+    },
+    {
+      "epoch": 0.026788990825688072,
+      "grad_norm": 0.6927171945571899,
+      "learning_rate": 5e-05,
+      "loss": 0.0257,
+      "step": 73
+    },
+    {
+      "epoch": 0.027155963302752294,
+      "grad_norm": 0.12636712193489075,
+      "learning_rate": 5e-05,
+      "loss": 0.0064,
+      "step": 74
+    },
+    {
+      "epoch": 0.027522935779816515,
+      "grad_norm": 6.38351583480835,
+      "learning_rate": 5e-05,
+      "loss": 0.8959,
+      "step": 75
+    },
+    {
+      "epoch": 0.027889908256880733,
+      "grad_norm": 0.23792296648025513,
+      "learning_rate": 5e-05,
+      "loss": 0.011,
+      "step": 76
+    },
+    {
+      "epoch": 0.028256880733944955,
+      "grad_norm": 4.081106662750244,
+      "learning_rate": 5e-05,
+      "loss": 0.2215,
+      "step": 77
+    },
+    {
+      "epoch": 0.028623853211009173,
+      "grad_norm": 5.544277667999268,
+      "learning_rate": 5e-05,
+      "loss": 0.0833,
+      "step": 78
+    },
+    {
+      "epoch": 0.028990825688073395,
+      "grad_norm": 5.9235687255859375,
+      "learning_rate": 5e-05,
+      "loss": 0.5039,
+      "step": 79
+    },
+    {
+      "epoch": 0.029357798165137616,
+      "grad_norm": 33.5960693359375,
+      "learning_rate": 5e-05,
+      "loss": 0.3264,
+      "step": 80
+    },
+    {
+      "epoch": 0.029724770642201834,
+      "grad_norm": 0.29501473903656006,
+      "learning_rate": 5e-05,
+      "loss": 0.0124,
+      "step": 81
+    },
+    {
+      "epoch": 0.030091743119266056,
+      "grad_norm": 2.078266143798828,
+      "learning_rate": 5e-05,
+      "loss": 0.0609,
+      "step": 82
+    },
+    {
+      "epoch": 0.030458715596330274,
+      "grad_norm": 4.236815929412842,
+      "learning_rate": 5e-05,
+      "loss": 0.1367,
+      "step": 83
+    },
+    {
+      "epoch": 0.030825688073394496,
+      "grad_norm": 3.3054449558258057,
+      "learning_rate": 5e-05,
+      "loss": 0.4739,
+      "step": 84
+    },
+    {
+      "epoch": 0.031192660550458717,
+      "grad_norm": 0.7153921723365784,
+      "learning_rate": 5e-05,
+      "loss": 0.0235,
+      "step": 85
+    },
+    {
+      "epoch": 0.03155963302752294,
+      "grad_norm": 15.103848457336426,
+      "learning_rate": 5e-05,
+      "loss": 0.1547,
+      "step": 86
+    },
+    {
+      "epoch": 0.031926605504587154,
+      "grad_norm": 7.31617546081543,
+      "learning_rate": 5e-05,
+      "loss": 0.2991,
+      "step": 87
+    },
+    {
+      "epoch": 0.032293577981651375,
+      "grad_norm": 0.8391185998916626,
+      "learning_rate": 5e-05,
+      "loss": 0.0304,
+      "step": 88
+    },
+    {
+      "epoch": 0.0326605504587156,
+      "grad_norm": 9.793542861938477,
+      "learning_rate": 5e-05,
+      "loss": 0.0772,
+      "step": 89
+    },
+    {
+      "epoch": 0.03302752293577982,
+      "grad_norm": 10.501496315002441,
+      "learning_rate": 5e-05,
+      "loss": 0.9207,
+      "step": 90
+    },
+    {
+      "epoch": 0.03339449541284404,
+      "grad_norm": 2.911019802093506,
+      "learning_rate": 5e-05,
+      "loss": 0.0248,
+      "step": 91
+    },
+    {
+      "epoch": 0.033761467889908255,
+      "grad_norm": 2.5744543075561523,
+      "learning_rate": 5e-05,
+      "loss": 0.0182,
+      "step": 92
+    },
+    {
+      "epoch": 0.034128440366972476,
+      "grad_norm": 0.26191648840904236,
+      "learning_rate": 5e-05,
+      "loss": 0.0109,
+      "step": 93
+    },
+    {
+      "epoch": 0.0344954128440367,
+      "grad_norm": 3.0673983097076416,
+      "learning_rate": 5e-05,
+      "loss": 0.0653,
+      "step": 94
+    },
+    {
+      "epoch": 0.03486238532110092,
+      "grad_norm": 7.1955342292785645,
+      "learning_rate": 5e-05,
+      "loss": 1.1587,
+      "step": 95
+    },
+    {
+      "epoch": 0.03522935779816514,
+      "grad_norm": 7.954893589019775,
+      "learning_rate": 5e-05,
+      "loss": 1.0011,
+      "step": 96
+    },
+    {
+      "epoch": 0.035596330275229356,
+      "grad_norm": 0.5798079371452332,
+      "learning_rate": 5e-05,
+      "loss": 0.0182,
+      "step": 97
+    },
+    {
+      "epoch": 0.03596330275229358,
+      "grad_norm": 0.4644419252872467,
+      "learning_rate": 5e-05,
+      "loss": 0.0194,
+      "step": 98
+    },
+    {
+      "epoch": 0.0363302752293578,
+      "grad_norm": 0.26596662402153015,
+      "learning_rate": 5e-05,
+      "loss": 0.0112,
+      "step": 99
+    },
+    {
+      "epoch": 0.03669724770642202,
+      "grad_norm": 3.482236862182617,
+      "learning_rate": 5e-05,
+      "loss": 0.5393,
+      "step": 100
+    },
+    {
+      "epoch": 0.03706422018348624,
+      "grad_norm": 3.809786558151245,
+      "learning_rate": 5e-05,
+      "loss": 0.4622,
+      "step": 101
+    },
+    {
+      "epoch": 0.03743119266055046,
+      "grad_norm": 9.85364818572998,
+      "learning_rate": 5e-05,
+      "loss": 0.1608,
+      "step": 102
+    },
+    {
+      "epoch": 0.03779816513761468,
+      "grad_norm": 0.39617234468460083,
+      "learning_rate": 5e-05,
+      "loss": 0.0168,
+      "step": 103
+    },
+    {
+      "epoch": 0.0381651376146789,
+      "grad_norm": 12.427167892456055,
+      "learning_rate": 5e-05,
+      "loss": 0.1535,
+      "step": 104
+    },
+    {
+      "epoch": 0.03853211009174312,
+      "grad_norm": 8.598713874816895,
+      "learning_rate": 5e-05,
+      "loss": 0.4384,
+      "step": 105
+    },
+    {
+      "epoch": 0.038899082568807336,
+      "grad_norm": 4.784875392913818,
+      "learning_rate": 5e-05,
+      "loss": 0.4216,
+      "step": 106
+    },
+    {
+      "epoch": 0.03926605504587156,
+      "grad_norm": 2.0999863147735596,
+      "learning_rate": 5e-05,
+      "loss": 0.04,
+      "step": 107
+    },
+    {
+      "epoch": 0.03963302752293578,
+      "grad_norm": 6.801530361175537,
+      "learning_rate": 5e-05,
+      "loss": 0.5133,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0184106826782227,
+      "learning_rate": 5e-05,
+      "loss": 0.1484,
+      "step": 109
+    },
+    {
+      "epoch": 0.04036697247706422,
+      "grad_norm": 1.8532726764678955,
+      "learning_rate": 5e-05,
+      "loss": 0.0741,
+      "step": 110
+    },
+    {
+      "epoch": 0.04036697247706422,
+      "eval_accuracy": 0.5,
+      "eval_loss": 1.1352694034576416,
+      "eval_runtime": 28.2851,
+      "eval_samples_per_second": 11.95,
+      "eval_steps_per_second": 1.52,
+      "step": 110
+    },
+    {
+      "epoch": 1.0003669724770643,
+      "grad_norm": 3.2624194622039795,
+      "learning_rate": 5e-05,
+      "loss": 0.4381,
+      "step": 111
+    },
+    {
+      "epoch": 1.0007339449541284,
+      "grad_norm": 0.3712124824523926,
+      "learning_rate": 5e-05,
+      "loss": 0.0149,
+      "step": 112
+    },
+    {
+      "epoch": 1.0011009174311927,
+      "grad_norm": 12.700252532958984,
+      "learning_rate": 5e-05,
+      "loss": 0.1207,
+      "step": 113
+    },
+    {
+      "epoch": 1.001467889908257,
+      "grad_norm": 0.20263834297657013,
+      "learning_rate": 5e-05,
+      "loss": 0.0087,
+      "step": 114
+    },
+    {
+      "epoch": 1.001834862385321,
+      "grad_norm": 0.9390329718589783,
+      "learning_rate": 5e-05,
+      "loss": 0.0404,
+      "step": 115
+    },
+    {
+      "epoch": 1.0022018348623853,
+      "grad_norm": 0.35725080966949463,
+      "learning_rate": 5e-05,
+      "loss": 0.0152,
+      "step": 116
+    },
+    {
+      "epoch": 1.0025688073394496,
+      "grad_norm": 0.8551551103591919,
+      "learning_rate": 5e-05,
+      "loss": 0.0355,
+      "step": 117
+    },
+    {
+      "epoch": 1.0029357798165137,
+      "grad_norm": 0.29797810316085815,
+      "learning_rate": 5e-05,
+      "loss": 0.0119,
+      "step": 118
+    },
+    {
+      "epoch": 1.003302752293578,
+      "grad_norm": 5.283618927001953,
+      "learning_rate": 5e-05,
+      "loss": 0.1695,
+      "step": 119
+    },
+    {
+      "epoch": 1.003669724770642,
+      "grad_norm": 3.9632015228271484,
+      "learning_rate": 5e-05,
+      "loss": 0.4226,
+      "step": 120
+    },
+    {
+      "epoch": 1.0040366972477064,
+      "grad_norm": 10.07297134399414,
+      "learning_rate": 5e-05,
+      "loss": 0.7977,
+      "step": 121
+    },
+    {
+      "epoch": 1.0044036697247707,
+      "grad_norm": 1.6393853425979614,
+      "learning_rate": 5e-05,
+      "loss": 0.0125,
+      "step": 122
+    },
+    {
+      "epoch": 1.0047706422018348,
+      "grad_norm": 10.305487632751465,
+      "learning_rate": 5e-05,
+      "loss": 0.7063,
+      "step": 123
+    },
+    {
+      "epoch": 1.005137614678899,
+      "grad_norm": 0.8910521268844604,
+      "learning_rate": 5e-05,
+      "loss": 0.0261,
+      "step": 124
+    },
+    {
+      "epoch": 1.0055045871559634,
+      "grad_norm": 0.10523223131895065,
+      "learning_rate": 5e-05,
+      "loss": 0.0044,
+      "step": 125
+    },
+    {
+      "epoch": 1.0058715596330274,
+      "grad_norm": 0.13839223980903625,
+      "learning_rate": 5e-05,
+      "loss": 0.0056,
+      "step": 126
+    },
+    {
+      "epoch": 1.0062385321100917,
+      "grad_norm": 6.318146705627441,
+      "learning_rate": 5e-05,
+      "loss": 0.0842,
+      "step": 127
+    },
+    {
+      "epoch": 1.006605504587156,
+      "grad_norm": 3.8025062084198,
+      "learning_rate": 5e-05,
+      "loss": 0.8147,
+      "step": 128
+    },
+    {
+      "epoch": 1.00697247706422,
+      "grad_norm": 4.658647537231445,
+      "learning_rate": 5e-05,
+      "loss": 0.3894,
+      "step": 129
+    },
+    {
+      "epoch": 1.0073394495412844,
+      "grad_norm": 0.16271570324897766,
+      "learning_rate": 5e-05,
+      "loss": 0.0065,
+      "step": 130
+    },
+    {
+      "epoch": 1.0077064220183487,
+      "grad_norm": 4.8438286781311035,
+      "learning_rate": 5e-05,
+      "loss": 0.3602,
+      "step": 131
+    },
+    {
+      "epoch": 1.0080733944954128,
+      "grad_norm": 47.5307731628418,
+      "learning_rate": 5e-05,
+      "loss": 0.3416,
+      "step": 132
+    },
+    {
+      "epoch": 1.008440366972477,
+      "grad_norm": 0.18173491954803467,
+      "learning_rate": 5e-05,
+      "loss": 0.008,
+      "step": 133
+    },
+    {
+      "epoch": 1.0088073394495414,
+      "grad_norm": 5.772665500640869,
+      "learning_rate": 5e-05,
+      "loss": 0.3185,
+      "step": 134
+    },
+    {
+      "epoch": 1.0091743119266054,
+      "grad_norm": 0.1098528727889061,
+      "learning_rate": 5e-05,
+      "loss": 0.0044,
+      "step": 135
+    },
+    {
+      "epoch": 1.0095412844036697,
+      "grad_norm": 4.924429416656494,
+      "learning_rate": 5e-05,
+      "loss": 0.0879,
+      "step": 136
+    },
+    {
+      "epoch": 1.009908256880734,
+      "grad_norm": 0.06098955124616623,
+      "learning_rate": 5e-05,
+      "loss": 0.0028,
+      "step": 137
+    },
+    {
+      "epoch": 1.0102752293577981,
+      "grad_norm": 8.785345077514648,
+      "learning_rate": 5e-05,
+      "loss": 0.4549,
+      "step": 138
+    },
+    {
+      "epoch": 1.0106422018348624,
+      "grad_norm": 1.1175813674926758,
+      "learning_rate": 5e-05,
+      "loss": 0.0256,
+      "step": 139
+    },
+    {
+      "epoch": 1.0110091743119267,
+      "grad_norm": 3.097990036010742,
+      "learning_rate": 5e-05,
+      "loss": 0.0386,
+      "step": 140
+    },
+    {
+      "epoch": 1.0113761467889908,
+      "grad_norm": 0.31747254729270935,
+      "learning_rate": 5e-05,
+      "loss": 0.0105,
+      "step": 141
+    },
+    {
+      "epoch": 1.011743119266055,
+      "grad_norm": 0.19412679970264435,
+      "learning_rate": 5e-05,
+      "loss": 0.0083,
+      "step": 142
+    },
+    {
+      "epoch": 1.0121100917431192,
+      "grad_norm": 0.10393175482749939,
+      "learning_rate": 5e-05,
+      "loss": 0.0047,
+      "step": 143
+    },
+    {
+      "epoch": 1.0124770642201835,
+      "grad_norm": 0.6574366092681885,
+      "learning_rate": 5e-05,
+      "loss": 0.0193,
+      "step": 144
+    },
+    {
+      "epoch": 1.0128440366972478,
+      "grad_norm": 0.10559283941984177,
+      "learning_rate": 5e-05,
+      "loss": 0.0044,
+      "step": 145
+    },
+    {
+      "epoch": 1.0132110091743118,
+      "grad_norm": 4.850479602813721,
+      "learning_rate": 5e-05,
+      "loss": 0.0182,
+      "step": 146
+    },
+    {
+      "epoch": 1.0135779816513761,
+      "grad_norm": 0.11620081216096878,
+      "learning_rate": 5e-05,
+      "loss": 0.0043,
+      "step": 147
+    },
+    {
+      "epoch": 1.0139449541284404,
+      "grad_norm": 0.03569629415869713,
+      "learning_rate": 5e-05,
+      "loss": 0.0018,
+      "step": 148
+    },
+    {
+      "epoch": 1.0143119266055045,
+      "grad_norm": 3.962449789047241,
+      "learning_rate": 5e-05,
+      "loss": 0.6633,
+      "step": 149
+    },
+    {
+      "epoch": 1.0146788990825688,
+      "grad_norm": 0.03495300933718681,
+      "learning_rate": 5e-05,
+      "loss": 0.0017,
+      "step": 150
+    },
+    {
+      "epoch": 1.015045871559633,
+      "grad_norm": 3.539942979812622,
+      "learning_rate": 5e-05,
+      "loss": 0.739,
+      "step": 151
+    },
+    {
+      "epoch": 1.0154128440366972,
+      "grad_norm": 7.318427085876465,
+      "learning_rate": 5e-05,
+      "loss": 0.5628,
+      "step": 152
+    },
+    {
+      "epoch": 1.0157798165137615,
+      "grad_norm": 0.5549555420875549,
+      "learning_rate": 5e-05,
+      "loss": 0.0184,
+      "step": 153
+    },
+    {
+      "epoch": 1.0161467889908258,
+      "grad_norm": 4.45099401473999,
+      "learning_rate": 5e-05,
+      "loss": 0.071,
+      "step": 154
+    },
+    {
+      "epoch": 1.0165137614678899,
+      "grad_norm": 7.420391082763672,
+      "learning_rate": 5e-05,
+      "loss": 0.3171,
+      "step": 155
+    },
+    {
+      "epoch": 1.0168807339449542,
+      "grad_norm": 32.71088790893555,
+      "learning_rate": 5e-05,
+      "loss": 0.9586,
+      "step": 156
+    },
+    {
+      "epoch": 1.0172477064220184,
+      "grad_norm": 4.373344421386719,
+      "learning_rate": 5e-05,
+      "loss": 0.3713,
+      "step": 157
+    },
+    {
+      "epoch": 1.0176146788990825,
+      "grad_norm": 27.9835147857666,
+      "learning_rate": 5e-05,
+      "loss": 1.6796,
+      "step": 158
+    },
+    {
+      "epoch": 1.0179816513761468,
+      "grad_norm": 3.565764904022217,
+      "learning_rate": 5e-05,
+      "loss": 0.3572,
+      "step": 159
+    },
+    {
+      "epoch": 1.018348623853211,
+      "grad_norm": 12.718873977661133,
+      "learning_rate": 5e-05,
+      "loss": 0.7293,
+      "step": 160
+    },
+    {
+      "epoch": 1.0187155963302752,
+      "grad_norm": 14.63851547241211,
+      "learning_rate": 5e-05,
+      "loss": 0.4575,
+      "step": 161
+    },
+    {
+      "epoch": 1.0190825688073395,
+      "grad_norm": 2.6600899696350098,
+      "learning_rate": 5e-05,
+      "loss": 0.1451,
+      "step": 162
+    },
+    {
+      "epoch": 1.0194495412844036,
+      "grad_norm": 3.707906723022461,
+      "learning_rate": 5e-05,
+      "loss": 0.2055,
+      "step": 163
+    },
+    {
+      "epoch": 1.0198165137614679,
+      "grad_norm": 2.595513343811035,
+      "learning_rate": 5e-05,
+      "loss": 0.1865,
+      "step": 164
+    },
+    {
+      "epoch": 1.0201834862385322,
+      "grad_norm": 2.2449002265930176,
+      "learning_rate": 5e-05,
+      "loss": 0.0649,
+      "step": 165
+    },
+    {
+      "epoch": 1.0205504587155962,
+      "grad_norm": 1.119431495666504,
+      "learning_rate": 5e-05,
+      "loss": 0.0431,
+      "step": 166
+    },
+    {
+      "epoch": 1.0209174311926605,
+      "grad_norm": 0.9740021228790283,
+      "learning_rate": 5e-05,
+      "loss": 0.0416,
+      "step": 167
+    },
+    {
+      "epoch": 1.0212844036697248,
+      "grad_norm": 0.6569391489028931,
+      "learning_rate": 5e-05,
+      "loss": 0.0237,
+      "step": 168
+    },
+    {
+      "epoch": 1.021651376146789,
+      "grad_norm": 3.094494581222534,
+      "learning_rate": 5e-05,
+      "loss": 0.5642,
+      "step": 169
+    },
+    {
+      "epoch": 1.0220183486238532,
+      "grad_norm": 2.9208061695098877,
+      "learning_rate": 5e-05,
+      "loss": 0.5584,
+      "step": 170
+    },
+    {
+      "epoch": 1.0223853211009175,
+      "grad_norm": 0.10827453434467316,
+      "learning_rate": 5e-05,
+      "loss": 0.0053,
+      "step": 171
+    },
+    {
+      "epoch": 1.0227522935779816,
+      "grad_norm": 8.481878280639648,
+      "learning_rate": 5e-05,
+      "loss": 0.2166,
+      "step": 172
+    },
+    {
+      "epoch": 1.0231192660550459,
+      "grad_norm": 4.289131164550781,
+      "learning_rate": 5e-05,
+      "loss": 0.3037,
+      "step": 173
+    },
+    {
+      "epoch": 1.0234862385321102,
+      "grad_norm": 7.553821086883545,
+      "learning_rate": 5e-05,
+      "loss": 0.119,
+      "step": 174
+    },
+    {
+      "epoch": 1.0238532110091743,
+      "grad_norm": 3.613996744155884,
+      "learning_rate": 5e-05,
+      "loss": 0.0451,
+      "step": 175
+    },
+    {
+      "epoch": 1.0242201834862386,
+      "grad_norm": 5.399818420410156,
+      "learning_rate": 5e-05,
+      "loss": 0.1578,
+      "step": 176
+    },
+    {
+      "epoch": 1.0245871559633029,
+      "grad_norm": 3.0832746028900146,
+      "learning_rate": 5e-05,
+      "loss": 0.5418,
+      "step": 177
+    },
+    {
+      "epoch": 1.024954128440367,
+      "grad_norm": 10.228167533874512,
+      "learning_rate": 5e-05,
+      "loss": 1.1447,
+      "step": 178
+    },
+    {
+      "epoch": 1.0253211009174312,
+      "grad_norm": 5.832758903503418,
+      "learning_rate": 5e-05,
+      "loss": 0.2518,
+      "step": 179
+    },
+    {
+      "epoch": 1.0256880733944955,
+      "grad_norm": 2.290520429611206,
+      "learning_rate": 5e-05,
+      "loss": 0.085,
+      "step": 180
+    },
+    {
+      "epoch": 1.0260550458715596,
+      "grad_norm": 5.634564399719238,
+      "learning_rate": 5e-05,
+      "loss": 0.1121,
+      "step": 181
+    },
+    {
+      "epoch": 1.026422018348624,
+      "grad_norm": 6.528543472290039,
+      "learning_rate": 5e-05,
+      "loss": 0.1542,
+      "step": 182
+    },
+    {
+      "epoch": 1.026788990825688,
+      "grad_norm": 0.2736799120903015,
+      "learning_rate": 5e-05,
+      "loss": 0.0123,
+      "step": 183
+    },
+    {
+      "epoch": 1.0271559633027523,
+      "grad_norm": 2.2989137172698975,
+      "learning_rate": 5e-05,
+      "loss": 0.2866,
+      "step": 184
+    },
+    {
+      "epoch": 1.0275229357798166,
+      "grad_norm": 11.668277740478516,
+      "learning_rate": 5e-05,
+      "loss": 0.19,
+      "step": 185
+    },
+    {
+      "epoch": 1.0278899082568806,
+      "grad_norm": 1.8051328659057617,
+      "learning_rate": 5e-05,
+      "loss": 0.0783,
+      "step": 186
+    },
+    {
+      "epoch": 1.028256880733945,
+      "grad_norm": 5.763457298278809,
+      "learning_rate": 5e-05,
+      "loss": 0.2919,
+      "step": 187
+    },
+    {
+      "epoch": 1.0286238532110092,
+      "grad_norm": 5.3835015296936035,
+      "learning_rate": 5e-05,
+      "loss": 0.1674,
+      "step": 188
+    },
+    {
+      "epoch": 1.0289908256880733,
+      "grad_norm": 12.367846488952637,
+      "learning_rate": 5e-05,
+      "loss": 0.6034,
+      "step": 189
+    },
+    {
+      "epoch": 1.0293577981651376,
+      "grad_norm": 3.675715446472168,
+      "learning_rate": 5e-05,
+      "loss": 0.1131,
+      "step": 190
+    },
+    {
+      "epoch": 1.029724770642202,
+      "grad_norm": 1.8906902074813843,
+      "learning_rate": 5e-05,
+      "loss": 0.0827,
+      "step": 191
+    },
+    {
+      "epoch": 1.030091743119266,
+      "grad_norm": 9.142536163330078,
+      "learning_rate": 5e-05,
+      "loss": 0.481,
+      "step": 192
+    },
+    {
+      "epoch": 1.0304587155963303,
+      "grad_norm": 1.3598240613937378,
+      "learning_rate": 5e-05,
+      "loss": 0.0445,
+      "step": 193
+    },
+    {
+      "epoch": 1.0308256880733946,
+      "grad_norm": 9.976537704467773,
+      "learning_rate": 5e-05,
+      "loss": 0.3771,
+      "step": 194
+    },
+    {
+      "epoch": 1.0311926605504587,
+      "grad_norm": 2.8891141414642334,
+      "learning_rate": 5e-05,
+      "loss": 0.1017,
+      "step": 195
+    },
+    {
+      "epoch": 1.031559633027523,
+      "grad_norm": 0.397853821516037,
+      "learning_rate": 5e-05,
+      "loss": 0.0174,
+      "step": 196
+    },
+    {
+      "epoch": 1.0319266055045873,
+      "grad_norm": 0.43675440549850464,
+      "learning_rate": 5e-05,
+      "loss": 0.0176,
+      "step": 197
+    },
+    {
+      "epoch": 1.0322935779816513,
+      "grad_norm": 11.94718074798584,
+      "learning_rate": 5e-05,
+      "loss": 0.3562,
+      "step": 198
+    },
+    {
+      "epoch": 1.0326605504587156,
+      "grad_norm": 0.2326168417930603,
+      "learning_rate": 5e-05,
+      "loss": 0.0101,
+      "step": 199
+    },
+    {
+      "epoch": 1.0330275229357797,
+      "grad_norm": 0.24757659435272217,
+      "learning_rate": 5e-05,
+      "loss": 0.01,
+      "step": 200
+    },
+    {
+      "epoch": 1.033394495412844,
+      "grad_norm": 3.0414950847625732,
+      "learning_rate": 5e-05,
+      "loss": 0.0805,
+      "step": 201
+    },
+    {
+      "epoch": 1.0337614678899083,
+      "grad_norm": 14.541916847229004,
+      "learning_rate": 5e-05,
+      "loss": 0.1291,
+      "step": 202
+    },
+    {
+      "epoch": 1.0341284403669724,
+      "grad_norm": 3.3583457469940186,
+      "learning_rate": 5e-05,
+      "loss": 0.0347,
+      "step": 203
+    },
+    {
+      "epoch": 1.0344954128440367,
+      "grad_norm": 10.220564842224121,
+      "learning_rate": 5e-05,
+      "loss": 0.4121,
+      "step": 204
+    },
+    {
+      "epoch": 1.034862385321101,
+      "grad_norm": 0.0713169276714325,
+      "learning_rate": 5e-05,
+      "loss": 0.0032,
+      "step": 205
+    },
+    {
+      "epoch": 1.035229357798165,
+      "grad_norm": 4.899472236633301,
+      "learning_rate": 5e-05,
+      "loss": 0.3429,
+      "step": 206
+    },
+    {
+      "epoch": 1.0355963302752293,
+      "grad_norm": 16.711999893188477,
+      "learning_rate": 5e-05,
+      "loss": 0.1311,
+      "step": 207
+    },
+    {
+      "epoch": 1.0359633027522936,
+      "grad_norm": 3.8642189502716064,
+      "learning_rate": 5e-05,
+      "loss": 0.029,
+      "step": 208
+    },
+    {
+      "epoch": 1.0363302752293577,
+      "grad_norm": 6.308407306671143,
+      "learning_rate": 5e-05,
+      "loss": 1.2885,
+      "step": 209
+    },
+    {
+      "epoch": 1.036697247706422,
+      "grad_norm": 12.80115032196045,
+      "learning_rate": 5e-05,
+      "loss": 0.1282,
+      "step": 210
+    },
+    {
+      "epoch": 1.0370642201834863,
+      "grad_norm": 28.702354431152344,
+      "learning_rate": 5e-05,
+      "loss": 0.6809,
+      "step": 211
+    },
+    {
+      "epoch": 1.0374311926605504,
+      "grad_norm": 6.560705184936523,
+      "learning_rate": 5e-05,
+      "loss": 1.2442,
+      "step": 212
+    },
+    {
+      "epoch": 1.0377981651376147,
+      "grad_norm": 0.3281303346157074,
+      "learning_rate": 5e-05,
+      "loss": 0.012,
+      "step": 213
+    },
+    {
+      "epoch": 1.038165137614679,
+      "grad_norm": 0.08594893664121628,
+      "learning_rate": 5e-05,
+      "loss": 0.0038,
+      "step": 214
+    },
+    {
+      "epoch": 1.038532110091743,
+      "grad_norm": 0.43805766105651855,
+      "learning_rate": 5e-05,
+      "loss": 0.0157,
+      "step": 215
+    },
+    {
+      "epoch": 1.0388990825688074,
+      "grad_norm": 0.2981579899787903,
+      "learning_rate": 5e-05,
+      "loss": 0.0108,
+      "step": 216
+    },
+    {
+      "epoch": 1.0392660550458717,
+      "grad_norm": 0.18283423781394958,
+      "learning_rate": 5e-05,
+      "loss": 0.0049,
+      "step": 217
+    },
+    {
+      "epoch": 1.0396330275229357,
+      "grad_norm": 0.2474643886089325,
+      "learning_rate": 5e-05,
+      "loss": 0.0086,
+      "step": 218
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.838778555393219,
+      "learning_rate": 5e-05,
+      "loss": 0.0104,
+      "step": 219
+    },
+    {
+      "epoch": 1.0403669724770643,
+      "grad_norm": 1.1746125221252441,
+      "learning_rate": 5e-05,
+      "loss": 0.0259,
+      "step": 220
+    },
+    {
+      "epoch": 1.0403669724770643,
+      "eval_accuracy": 0.11834319526627218,
+      "eval_loss": 3.714170217514038,
+      "eval_runtime": 27.8403,
+      "eval_samples_per_second": 12.141,
+      "eval_steps_per_second": 1.545,
+      "step": 220
+    },
+    {
+      "epoch": 2.0003669724770643,
+      "grad_norm": 3.3098745346069336,
+      "learning_rate": 5e-05,
+      "loss": 0.5984,
+      "step": 221
+    },
+    {
+      "epoch": 2.0007339449541286,
+      "grad_norm": 3.1905641555786133,
+      "learning_rate": 5e-05,
+      "loss": 0.6037,
+      "step": 222
+    },
+    {
+      "epoch": 2.0011009174311924,
+      "grad_norm": 0.11920172721147537,
+      "learning_rate": 5e-05,
+      "loss": 0.005,
+      "step": 223
+    },
+    {
+      "epoch": 2.0014678899082567,
+      "grad_norm": 0.2815546989440918,
+      "learning_rate": 5e-05,
+      "loss": 0.0108,
+      "step": 224
+    },
+    {
+      "epoch": 2.001834862385321,
+      "grad_norm": 3.7425575256347656,
+      "learning_rate": 5e-05,
+      "loss": 0.8452,
+      "step": 225
+    },
+    {
+      "epoch": 2.0022018348623853,
+      "grad_norm": 5.340523719787598,
+      "learning_rate": 5e-05,
+      "loss": 0.0343,
+      "step": 226
+    },
+    {
+      "epoch": 2.0025688073394496,
+      "grad_norm": 4.530514240264893,
+      "learning_rate": 5e-05,
+      "loss": 0.0478,
+      "step": 227
+    },
+    {
+      "epoch": 2.002935779816514,
+      "grad_norm": 0.5523406863212585,
+      "learning_rate": 5e-05,
+      "loss": 0.0122,
+      "step": 228
+    },
+    {
+      "epoch": 2.003302752293578,
+      "grad_norm": 0.2626419961452484,
+      "learning_rate": 5e-05,
+      "loss": 0.0113,
+      "step": 229
+    },
+    {
+      "epoch": 2.003669724770642,
+      "grad_norm": 4.385184288024902,
+      "learning_rate": 5e-05,
+      "loss": 0.6609,
+      "step": 230
+    },
+    {
+      "epoch": 2.0040366972477064,
+      "grad_norm": 0.6880325078964233,
+      "learning_rate": 5e-05,
+      "loss": 0.0258,
+      "step": 231
+    },
+    {
+      "epoch": 2.0044036697247707,
+      "grad_norm": 0.16113348305225372,
+      "learning_rate": 5e-05,
+      "loss": 0.0076,
+      "step": 232
+    },
+    {
+      "epoch": 2.004770642201835,
+      "grad_norm": 0.18330299854278564,
+      "learning_rate": 5e-05,
+      "loss": 0.0082,
+      "step": 233
+    },
+    {
+      "epoch": 2.0051376146788993,
+      "grad_norm": 0.11417803168296814,
+      "learning_rate": 5e-05,
+      "loss": 0.0057,
+      "step": 234
+    },
+    {
+      "epoch": 2.005504587155963,
+      "grad_norm": 5.610725402832031,
+      "learning_rate": 5e-05,
+      "loss": 0.5976,
+      "step": 235
+    },
+    {
+      "epoch": 2.0058715596330274,
+      "grad_norm": 2.915940046310425,
+      "learning_rate": 5e-05,
+      "loss": 0.5405,
+      "step": 236
+    },
+    {
+      "epoch": 2.0062385321100917,
+      "grad_norm": 0.19295038282871246,
+      "learning_rate": 5e-05,
+      "loss": 0.0092,
+      "step": 237
+    },
+    {
+      "epoch": 2.006605504587156,
+      "grad_norm": 0.18379826843738556,
+      "learning_rate": 5e-05,
+      "loss": 0.0084,
+      "step": 238
+    },
+    {
+      "epoch": 2.0069724770642203,
+      "grad_norm": 0.21146495640277863,
+      "learning_rate": 5e-05,
+      "loss": 0.0097,
+      "step": 239
+    },
+    {
+      "epoch": 2.007339449541284,
+      "grad_norm": 15.179059028625488,
+      "learning_rate": 5e-05,
+      "loss": 1.4366,
+      "step": 240
+    },
+    {
+      "epoch": 2.0077064220183485,
+      "grad_norm": 2.813278913497925,
+      "learning_rate": 5e-05,
+      "loss": 0.485,
+      "step": 241
+    },
+    {
+      "epoch": 2.0080733944954128,
+      "grad_norm": 2.852019786834717,
+      "learning_rate": 5e-05,
+      "loss": 0.566,
+      "step": 242
+    },
+    {
+      "epoch": 2.008440366972477,
+      "grad_norm": 9.99439811706543,
+      "learning_rate": 5e-05,
+      "loss": 1.2538,
+      "step": 243
+    },
+    {
+      "epoch": 2.0088073394495414,
+      "grad_norm": 2.9242894649505615,
+      "learning_rate": 5e-05,
+      "loss": 0.4865,
+      "step": 244
+    },
+    {
+      "epoch": 2.0091743119266057,
+      "grad_norm": 3.482128143310547,
+      "learning_rate": 5e-05,
+      "loss": 0.364,
+      "step": 245
+    },
+    {
+      "epoch": 2.0095412844036695,
+      "grad_norm": 3.0923054218292236,
+      "learning_rate": 5e-05,
+      "loss": 0.4441,
+      "step": 246
+    },
+    {
+      "epoch": 2.009908256880734,
+      "grad_norm": 0.8021335601806641,
+      "learning_rate": 5e-05,
+      "loss": 0.0384,
+      "step": 247
+    },
+    {
+      "epoch": 2.010275229357798,
+      "grad_norm": 6.722145080566406,
+      "learning_rate": 5e-05,
+      "loss": 0.3031,
+      "step": 248
+    },
+    {
+      "epoch": 2.0106422018348624,
+      "grad_norm": 2.6598613262176514,
+      "learning_rate": 5e-05,
+      "loss": 0.1012,
+      "step": 249
+    },
+    {
+      "epoch": 2.0110091743119267,
+      "grad_norm": 0.8569692969322205,
+      "learning_rate": 5e-05,
+      "loss": 0.0404,
+      "step": 250
+    },
+    {
+      "epoch": 2.011376146788991,
+      "grad_norm": 6.0418243408203125,
+      "learning_rate": 5e-05,
+      "loss": 0.3094,
+      "step": 251
+    },
+    {
+      "epoch": 2.011743119266055,
+      "grad_norm": 2.082373857498169,
+      "learning_rate": 5e-05,
+      "loss": 0.0965,
+      "step": 252
+    },
+    {
+      "epoch": 2.012110091743119,
+      "grad_norm": 1.664687991142273,
+      "learning_rate": 5e-05,
+      "loss": 0.0797,
+      "step": 253
+    },
+    {
+      "epoch": 2.0124770642201835,
+      "grad_norm": 0.8474786877632141,
+      "learning_rate": 5e-05,
+      "loss": 0.0396,
+      "step": 254
+    },
+    {
+      "epoch": 2.0128440366972478,
+      "grad_norm": 1.3707972764968872,
+      "learning_rate": 5e-05,
+      "loss": 0.0645,
+      "step": 255
+    },
+    {
+      "epoch": 2.013211009174312,
+      "grad_norm": 0.8410927057266235,
+      "learning_rate": 5e-05,
+      "loss": 0.0389,
+      "step": 256
+    },
+    {
+      "epoch": 2.0135779816513764,
+      "grad_norm": 0.5474077463150024,
+      "learning_rate": 5e-05,
+      "loss": 0.0293,
+      "step": 257
+    },
+    {
+      "epoch": 2.01394495412844,
+      "grad_norm": 7.09504508972168,
+      "learning_rate": 5e-05,
+      "loss": 1.0457,
+      "step": 258
+    },
+    {
+      "epoch": 2.0143119266055045,
+      "grad_norm": 0.5468025803565979,
+      "learning_rate": 5e-05,
+      "loss": 0.0234,
+      "step": 259
+    },
+    {
+      "epoch": 2.014678899082569,
+      "grad_norm": 10.472004890441895,
+      "learning_rate": 5e-05,
+      "loss": 0.0663,
+      "step": 260
+    },
+    {
+      "epoch": 2.015045871559633,
+      "grad_norm": 0.1386280655860901,
+      "learning_rate": 5e-05,
+      "loss": 0.0068,
+      "step": 261
+    },
+    {
+      "epoch": 2.0154128440366974,
+      "grad_norm": 2.8551340103149414,
+      "learning_rate": 5e-05,
+      "loss": 0.4273,
+      "step": 262
+    },
+    {
+      "epoch": 2.0157798165137613,
+      "grad_norm": 0.34334972500801086,
+      "learning_rate": 5e-05,
+      "loss": 0.0152,
+      "step": 263
+    },
+    {
+      "epoch": 2.0161467889908256,
+      "grad_norm": 3.112992286682129,
+      "learning_rate": 5e-05,
+      "loss": 0.4194,
+      "step": 264
+    },
+    {
+      "epoch": 2.01651376146789,
+      "grad_norm": 0.31794703006744385,
+      "learning_rate": 5e-05,
+      "loss": 0.0113,
+      "step": 265
+    },
+    {
+      "epoch": 2.016880733944954,
+      "grad_norm": 5.8122878074646,
+      "learning_rate": 5e-05,
+      "loss": 0.8694,
+      "step": 266
+    },
+    {
+      "epoch": 2.0172477064220184,
+      "grad_norm": 12.84292984008789,
+      "learning_rate": 5e-05,
+      "loss": 0.0437,
+      "step": 267
+    },
+    {
+      "epoch": 2.0176146788990827,
+      "grad_norm": 15.84395980834961,
+      "learning_rate": 5e-05,
+      "loss": 0.6422,
+      "step": 268
+    },
+    {
+      "epoch": 2.0179816513761466,
+      "grad_norm": 2.8891398906707764,
+      "learning_rate": 5e-05,
+      "loss": 0.5171,
+      "step": 269
+    },
+    {
+      "epoch": 2.018348623853211,
+      "grad_norm": 3.0113930702209473,
+      "learning_rate": 5e-05,
+      "loss": 0.4853,
+      "step": 270
+    },
+    {
+      "epoch": 2.018715596330275,
+      "grad_norm": 0.2399570345878601,
+      "learning_rate": 5e-05,
+      "loss": 0.0111,
+      "step": 271
+    },
+    {
+      "epoch": 2.0190825688073395,
+      "grad_norm": 0.3151020407676697,
+      "learning_rate": 5e-05,
+      "loss": 0.0137,
+      "step": 272
+    },
+    {
+      "epoch": 2.019449541284404,
+      "grad_norm": 3.697421073913574,
+      "learning_rate": 5e-05,
+      "loss": 0.3946,
+      "step": 273
+    },
+    {
+      "epoch": 2.019816513761468,
+      "grad_norm": 0.46915239095687866,
+      "learning_rate": 5e-05,
+      "loss": 0.0214,
+      "step": 274
+    },
+    {
+      "epoch": 2.020183486238532,
+      "grad_norm": 0.3693371117115021,
+      "learning_rate": 5e-05,
+      "loss": 0.0167,
+      "step": 275
+    },
+    {
+      "epoch": 2.0205504587155962,
+      "grad_norm": 4.579241752624512,
+      "learning_rate": 5e-05,
+      "loss": 0.281,
+      "step": 276
+    },
+    {
+      "epoch": 2.0209174311926605,
+      "grad_norm": 0.2460261881351471,
+      "learning_rate": 5e-05,
+      "loss": 0.0113,
+      "step": 277
+    },
+    {
+      "epoch": 2.021284403669725,
+      "grad_norm": 0.8817834854125977,
+      "learning_rate": 5e-05,
+      "loss": 0.0397,
+      "step": 278
+    },
+    {
+      "epoch": 2.021651376146789,
+      "grad_norm": 0.4477933645248413,
+      "learning_rate": 5e-05,
+      "loss": 0.0211,
+      "step": 279
+    },
+    {
+      "epoch": 2.0220183486238534,
+      "grad_norm": 0.49993953108787537,
+      "learning_rate": 5e-05,
+      "loss": 0.0228,
+      "step": 280
+    },
+    {
+      "epoch": 2.0223853211009173,
+      "grad_norm": 1.093253254890442,
+      "learning_rate": 5e-05,
+      "loss": 0.0232,
+      "step": 281
+    },
+    {
+      "epoch": 2.0227522935779816,
+      "grad_norm": 2.9838614463806152,
+      "learning_rate": 5e-05,
+      "loss": 0.3975,
+      "step": 282
+    },
+    {
+      "epoch": 2.023119266055046,
+      "grad_norm": 5.79507303237915,
+      "learning_rate": 5e-05,
+      "loss": 0.5574,
+      "step": 283
+    },
+    {
+      "epoch": 2.02348623853211,
+      "grad_norm": 2.6290011405944824,
+      "learning_rate": 5e-05,
+      "loss": 0.1956,
+      "step": 284
+    },
+    {
+      "epoch": 2.0238532110091745,
+      "grad_norm": 0.25618574023246765,
+      "learning_rate": 5e-05,
+      "loss": 0.0125,
+      "step": 285
+    },
+    {
+      "epoch": 2.0242201834862383,
+      "grad_norm": 0.26679301261901855,
+      "learning_rate": 5e-05,
+      "loss": 0.0118,
+      "step": 286
+    },
+    {
+      "epoch": 2.0245871559633026,
+      "grad_norm": 0.615793764591217,
+      "learning_rate": 5e-05,
+      "loss": 0.0211,
+      "step": 287
+    },
+    {
+      "epoch": 2.024954128440367,
+      "grad_norm": 3.446815013885498,
+      "learning_rate": 5e-05,
+      "loss": 0.462,
+      "step": 288
+    },
+    {
+      "epoch": 2.0253211009174312,
+      "grad_norm": 2.6503331661224365,
+      "learning_rate": 5e-05,
+      "loss": 0.5185,
+      "step": 289
+    },
+    {
+      "epoch": 2.0256880733944955,
+      "grad_norm": 2.8919334411621094,
+      "learning_rate": 5e-05,
+      "loss": 0.4913,
+      "step": 290
+    },
+    {
+      "epoch": 2.02605504587156,
+      "grad_norm": 6.523483753204346,
+      "learning_rate": 5e-05,
+      "loss": 0.3171,
+      "step": 291
+    },
+    {
+      "epoch": 2.0264220183486237,
+      "grad_norm": 0.46601414680480957,
+      "learning_rate": 5e-05,
+      "loss": 0.0205,
+      "step": 292
+    },
+    {
+      "epoch": 2.026788990825688,
+      "grad_norm": 0.3771248459815979,
+      "learning_rate": 5e-05,
+      "loss": 0.0163,
+      "step": 293
+    },
+    {
+      "epoch": 2.0271559633027523,
+      "grad_norm": 0.16291184723377228,
+      "learning_rate": 5e-05,
+      "loss": 0.0084,
+      "step": 294
+    },
+    {
+      "epoch": 2.0275229357798166,
+      "grad_norm": 3.0164453983306885,
+      "learning_rate": 5e-05,
+      "loss": 0.1478,
+      "step": 295
+    },
+    {
+      "epoch": 2.027889908256881,
+      "grad_norm": 0.2621011435985565,
+      "learning_rate": 5e-05,
+      "loss": 0.0126,
+      "step": 296
+    },
+    {
+      "epoch": 2.028256880733945,
+      "grad_norm": 6.095289707183838,
+      "learning_rate": 5e-05,
+      "loss": 0.408,
+      "step": 297
+    },
+    {
+      "epoch": 2.028623853211009,
+      "grad_norm": 2.1245741844177246,
+      "learning_rate": 5e-05,
+      "loss": 0.0909,
+      "step": 298
+    },
+    {
+      "epoch": 2.0289908256880733,
+      "grad_norm": 0.8265531063079834,
+      "learning_rate": 5e-05,
+      "loss": 0.0358,
+      "step": 299
+    },
+    {
+      "epoch": 2.0293577981651376,
+      "grad_norm": 0.4165724813938141,
+      "learning_rate": 5e-05,
+      "loss": 0.0182,
+      "step": 300
+    },
+    {
+      "epoch": 2.029724770642202,
+      "grad_norm": 0.7106027603149414,
+      "learning_rate": 5e-05,
+      "loss": 0.0308,
+      "step": 301
+    },
+    {
+      "epoch": 2.030091743119266,
+      "grad_norm": 0.5110293626785278,
+      "learning_rate": 5e-05,
+      "loss": 0.0221,
+      "step": 302
+    },
+    {
+      "epoch": 2.03045871559633,
+      "grad_norm": 2.828913688659668,
+      "learning_rate": 5e-05,
+      "loss": 0.1115,
+      "step": 303
+    },
+    {
+      "epoch": 2.0308256880733944,
+      "grad_norm": 0.2149876207113266,
+      "learning_rate": 5e-05,
+      "loss": 0.0097,
+      "step": 304
+    },
+    {
+      "epoch": 2.0311926605504587,
+      "grad_norm": 3.2597672939300537,
+      "learning_rate": 5e-05,
+      "loss": 0.4539,
+      "step": 305
+    },
+    {
+      "epoch": 2.031559633027523,
+      "grad_norm": 0.5128521919250488,
+      "learning_rate": 5e-05,
+      "loss": 0.0196,
+      "step": 306
+    },
+    {
+      "epoch": 2.0319266055045873,
+      "grad_norm": 6.016701698303223,
+      "learning_rate": 5e-05,
+      "loss": 0.5247,
+      "step": 307
+    },
+    {
+      "epoch": 2.0322935779816516,
+      "grad_norm": 0.08408281207084656,
+      "learning_rate": 5e-05,
+      "loss": 0.0038,
+      "step": 308
+    },
+    {
+      "epoch": 2.0326605504587154,
+      "grad_norm": 11.709476470947266,
+      "learning_rate": 5e-05,
+      "loss": 1.2183,
+      "step": 309
+    },
+    {
+      "epoch": 2.0330275229357797,
+      "grad_norm": 1.2272305488586426,
+      "learning_rate": 5e-05,
+      "loss": 0.0208,
+      "step": 310
+    },
+    {
+      "epoch": 2.033394495412844,
+      "grad_norm": 0.18302415311336517,
+      "learning_rate": 5e-05,
+      "loss": 0.0082,
+      "step": 311
+    },
+    {
+      "epoch": 2.0337614678899083,
+      "grad_norm": 0.1822645217180252,
+      "learning_rate": 5e-05,
+      "loss": 0.0081,
+      "step": 312
+    },
+    {
+      "epoch": 2.0341284403669726,
+      "grad_norm": 5.434505462646484,
+      "learning_rate": 5e-05,
+      "loss": 0.0704,
+      "step": 313
+    },
+    {
+      "epoch": 2.034495412844037,
+      "grad_norm": 0.1395518183708191,
+      "learning_rate": 5e-05,
+      "loss": 0.0062,
+      "step": 314
+    },
+    {
+      "epoch": 2.0348623853211008,
+      "grad_norm": 0.18503399193286896,
+      "learning_rate": 5e-05,
+      "loss": 0.0079,
+      "step": 315
+    },
+    {
+      "epoch": 2.035229357798165,
+      "grad_norm": 19.39935874938965,
+      "learning_rate": 5e-05,
+      "loss": 0.8194,
+      "step": 316
+    },
+    {
+      "epoch": 2.0355963302752293,
+      "grad_norm": 11.234166145324707,
+      "learning_rate": 5e-05,
+      "loss": 1.0814,
+      "step": 317
+    },
+    {
+      "epoch": 2.0359633027522936,
+      "grad_norm": 0.2770557701587677,
+      "learning_rate": 5e-05,
+      "loss": 0.0095,
+      "step": 318
+    },
+    {
+      "epoch": 2.036330275229358,
+      "grad_norm": 4.017146587371826,
+      "learning_rate": 5e-05,
+      "loss": 0.2492,
+      "step": 319
+    },
+    {
+      "epoch": 2.036697247706422,
+      "grad_norm": 0.0824546366930008,
+      "learning_rate": 5e-05,
+      "loss": 0.0041,
+      "step": 320
+    },
+    {
+      "epoch": 2.037064220183486,
+      "grad_norm": 0.18984024226665497,
+      "learning_rate": 5e-05,
+      "loss": 0.0069,
+      "step": 321
+    },
+    {
+      "epoch": 2.0374311926605504,
+      "grad_norm": 9.199568748474121,
+      "learning_rate": 5e-05,
+      "loss": 0.0527,
+      "step": 322
+    },
+    {
+      "epoch": 2.0377981651376147,
+      "grad_norm": 0.3961876332759857,
+      "learning_rate": 5e-05,
+      "loss": 0.0094,
+      "step": 323
+    },
+    {
+      "epoch": 2.038165137614679,
+      "grad_norm": 0.03565814346075058,
+      "learning_rate": 5e-05,
+      "loss": 0.0019,
+      "step": 324
+    },
+    {
+      "epoch": 2.0385321100917433,
+      "grad_norm": 2.880885362625122,
+      "learning_rate": 5e-05,
+      "loss": 0.7127,
+      "step": 325
+    },
+    {
+      "epoch": 2.038899082568807,
+      "grad_norm": 17.756855010986328,
+      "learning_rate": 5e-05,
+      "loss": 0.3802,
+      "step": 326
+    },
+    {
+      "epoch": 2.0392660550458714,
+      "grad_norm": 0.07551128417253494,
+      "learning_rate": 5e-05,
+      "loss": 0.0035,
+      "step": 327
+    },
+    {
+      "epoch": 2.0396330275229357,
+      "grad_norm": 0.07160859555006027,
+      "learning_rate": 5e-05,
+      "loss": 0.0034,
+      "step": 328
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 27.81930160522461,
+      "learning_rate": 5e-05,
+      "loss": 0.3666,
+      "step": 329
+    },
+    {
+      "epoch": 2.0403669724770643,
+      "grad_norm": 7.301838397979736,
+      "learning_rate": 5e-05,
+      "loss": 0.5784,
+      "step": 330
+    },
+    {
+      "epoch": 2.0403669724770643,
+      "eval_accuracy": 0.5,
+      "eval_loss": 2.2692294120788574,
+      "eval_runtime": 27.5028,
+      "eval_samples_per_second": 12.29,
+      "eval_steps_per_second": 1.563,
+      "step": 330
+    },
+    {
+      "epoch": 3.0003669724770643,
+      "grad_norm": 2.8246610164642334,
+      "learning_rate": 5e-05,
+      "loss": 0.6286,
+      "step": 331
+    },
+    {
+      "epoch": 3.0007339449541286,
+      "grad_norm": 0.1554068624973297,
+      "learning_rate": 5e-05,
+      "loss": 0.0072,
+      "step": 332
+    },
+    {
+      "epoch": 3.0011009174311924,
+      "grad_norm": 2.9072225093841553,
+      "learning_rate": 5e-05,
+      "loss": 0.633,
+      "step": 333
+    },
+    {
+      "epoch": 3.0014678899082567,
+      "grad_norm": 3.0170562267303467,
+      "learning_rate": 5e-05,
+      "loss": 0.5699,
+      "step": 334
+    },
+    {
+      "epoch": 3.001834862385321,
+      "grad_norm": 0.21071328222751617,
+      "learning_rate": 5e-05,
+      "loss": 0.0097,
+      "step": 335
+    },
+    {
+      "epoch": 3.0022018348623853,
+      "grad_norm": 0.2662504017353058,
+      "learning_rate": 5e-05,
+      "loss": 0.0126,
+      "step": 336
+    },
+    {
+      "epoch": 3.0025688073394496,
+      "grad_norm": 0.11802137643098831,
+      "learning_rate": 5e-05,
+      "loss": 0.0062,
+      "step": 337
+    },
+    {
+      "epoch": 3.002935779816514,
+      "grad_norm": 0.8615747690200806,
+      "learning_rate": 5e-05,
+      "loss": 0.025,
+      "step": 338
+    },
+    {
+      "epoch": 3.003302752293578,
+      "grad_norm": 2.4851205348968506,
+      "learning_rate": 5e-05,
+      "loss": 0.4648,
+      "step": 339
+    },
+    {
+      "epoch": 3.003669724770642,
+      "grad_norm": 13.307538032531738,
+      "learning_rate": 5e-05,
+      "loss": 0.67,
+      "step": 340
+    },
+    {
+      "epoch": 3.0040366972477064,
+      "grad_norm": 4.551970481872559,
+      "learning_rate": 5e-05,
+      "loss": 0.0565,
+      "step": 341
+    },
+    {
+      "epoch": 3.0044036697247707,
+      "grad_norm": 0.583398163318634,
+      "learning_rate": 5e-05,
+      "loss": 0.0267,
+      "step": 342
+    },
+    {
+      "epoch": 3.004770642201835,
+      "grad_norm": 0.41236770153045654,
+      "learning_rate": 5e-05,
+      "loss": 0.0205,
+      "step": 343
+    },
+    {
+      "epoch": 3.0051376146788993,
+      "grad_norm": 0.3104024827480316,
+      "learning_rate": 5e-05,
+      "loss": 0.0151,
+      "step": 344
+    },
+    {
+      "epoch": 3.005504587155963,
+      "grad_norm": 7.46708869934082,
+      "learning_rate": 5e-05,
+      "loss": 0.0573,
+      "step": 345
+    },
+    {
+      "epoch": 3.0058715596330274,
+      "grad_norm": 0.4326479136943817,
+      "learning_rate": 5e-05,
+      "loss": 0.024,
+      "step": 346
+    },
+    {
+      "epoch": 3.0062385321100917,
+      "grad_norm": 0.33921730518341064,
+      "learning_rate": 5e-05,
+      "loss": 0.0159,
+      "step": 347
+    },
+    {
+      "epoch": 3.006605504587156,
+      "grad_norm": 0.24354976415634155,
+      "learning_rate": 5e-05,
+      "loss": 0.0117,
+      "step": 348
+    },
+    {
+      "epoch": 3.0069724770642203,
+      "grad_norm": 2.931462287902832,
+      "learning_rate": 5e-05,
+      "loss": 0.3715,
+      "step": 349
+    },
+    {
+      "epoch": 3.007339449541284,
+      "grad_norm": 0.5845007300376892,
+      "learning_rate": 5e-05,
+      "loss": 0.0209,
+      "step": 350
+    },
+    {
+      "epoch": 3.0077064220183485,
+      "grad_norm": 18.826570510864258,
+      "learning_rate": 5e-05,
+      "loss": 0.3154,
+      "step": 351
+    },
+    {
+      "epoch": 3.0080733944954128,
+      "grad_norm": 17.036834716796875,
+      "learning_rate": 5e-05,
+      "loss": 0.0905,
+      "step": 352
+    },
+    {
+      "epoch": 3.008440366972477,
+      "grad_norm": 1.953502893447876,
+      "learning_rate": 5e-05,
+      "loss": 0.1254,
+      "step": 353
+    },
+    {
+      "epoch": 3.0088073394495414,
+      "grad_norm": 1.0132907629013062,
+      "learning_rate": 5e-05,
+      "loss": 0.0313,
+      "step": 354
+    },
+    {
+      "epoch": 3.0091743119266057,
+      "grad_norm": 0.7406408786773682,
+      "learning_rate": 5e-05,
+      "loss": 0.0243,
+      "step": 355
+    },
+    {
+      "epoch": 3.0095412844036695,
+      "grad_norm": 0.384557843208313,
+      "learning_rate": 5e-05,
+      "loss": 0.017,
+      "step": 356
+    },
+    {
+      "epoch": 3.009908256880734,
+      "grad_norm": 0.3357013761997223,
+      "learning_rate": 5e-05,
+      "loss": 0.0152,
+      "step": 357
+    },
+    {
+      "epoch": 3.010275229357798,
+      "grad_norm": 12.639945030212402,
+      "learning_rate": 5e-05,
+      "loss": 0.5209,
+      "step": 358
+    },
+    {
+      "epoch": 3.0106422018348624,
+      "grad_norm": 0.1706286519765854,
+      "learning_rate": 5e-05,
+      "loss": 0.0081,
+      "step": 359
+    },
+    {
+      "epoch": 3.0110091743119267,
+      "grad_norm": 0.1683771312236786,
+      "learning_rate": 5e-05,
+      "loss": 0.0075,
+      "step": 360
+    },
+    {
+      "epoch": 3.011376146788991,
+      "grad_norm": 0.1635533720254898,
+      "learning_rate": 5e-05,
+      "loss": 0.007,
+      "step": 361
+    },
+    {
+      "epoch": 3.011743119266055,
+      "grad_norm": 3.003115177154541,
+      "learning_rate": 5e-05,
+      "loss": 0.5533,
+      "step": 362
+    },
+    {
+      "epoch": 3.012110091743119,
+      "grad_norm": 2.90537691116333,
+      "learning_rate": 5e-05,
+      "loss": 0.5029,
+      "step": 363
+    },
+    {
+      "epoch": 3.0124770642201835,
+      "grad_norm": 0.17986638844013214,
+      "learning_rate": 5e-05,
+      "loss": 0.0081,
+      "step": 364
+    },
+    {
+      "epoch": 3.0128440366972478,
+      "grad_norm": 0.1346598118543625,
+      "learning_rate": 5e-05,
+      "loss": 0.0056,
+      "step": 365
+    },
+    {
+      "epoch": 3.013211009174312,
+      "grad_norm": 0.17465481162071228,
+      "learning_rate": 5e-05,
+      "loss": 0.0083,
+      "step": 366
+    },
+    {
+      "epoch": 3.0135779816513764,
+      "grad_norm": 0.14251567423343658,
+      "learning_rate": 5e-05,
+      "loss": 0.0065,
+      "step": 367
+    },
+    {
+      "epoch": 3.01394495412844,
+      "grad_norm": 3.489738941192627,
+      "learning_rate": 5e-05,
+      "loss": 0.516,
+      "step": 368
+    },
+    {
+      "epoch": 3.0143119266055045,
+      "grad_norm": 33.34955596923828,
+      "learning_rate": 5e-05,
+      "loss": 0.2351,
+      "step": 369
+    },
+    {
+      "epoch": 3.014678899082569,
+      "grad_norm": 0.19406218826770782,
+      "learning_rate": 5e-05,
+      "loss": 0.0089,
+      "step": 370
+    },
+    {
+      "epoch": 3.015045871559633,
+      "grad_norm": 2.9808249473571777,
+      "learning_rate": 5e-05,
+      "loss": 0.4352,
+      "step": 371
+    },
+    {
+      "epoch": 3.0154128440366974,
+      "grad_norm": 0.21648350358009338,
+      "learning_rate": 5e-05,
+      "loss": 0.0096,
+      "step": 372
+    },
+    {
+      "epoch": 3.0157798165137613,
+      "grad_norm": 0.13927467167377472,
+      "learning_rate": 5e-05,
+      "loss": 0.0063,
+      "step": 373
+    },
+    {
+      "epoch": 3.0161467889908256,
+      "grad_norm": 11.033052444458008,
+      "learning_rate": 5e-05,
+      "loss": 0.39,
+      "step": 374
+    },
+    {
+      "epoch": 3.01651376146789,
+      "grad_norm": 2.9069337844848633,
+      "learning_rate": 5e-05,
+      "loss": 0.4442,
+      "step": 375
+    },
+    {
+      "epoch": 3.016880733944954,
+      "grad_norm": 0.3146663308143616,
+      "learning_rate": 5e-05,
+      "loss": 0.0136,
+      "step": 376
+    },
+    {
+      "epoch": 3.0172477064220184,
+      "grad_norm": 0.4670899510383606,
+      "learning_rate": 5e-05,
+      "loss": 0.0199,
+      "step": 377
+    },
+    {
+      "epoch": 3.0176146788990827,
+      "grad_norm": 1.011947751045227,
+      "learning_rate": 5e-05,
+      "loss": 0.0448,
+      "step": 378
+    },
+    {
+      "epoch": 3.0179816513761466,
+      "grad_norm": 0.45602405071258545,
+      "learning_rate": 5e-05,
+      "loss": 0.0194,
+      "step": 379
+    },
+    {
+      "epoch": 3.018348623853211,
+      "grad_norm": 0.35276544094085693,
+      "learning_rate": 5e-05,
+      "loss": 0.0155,
+      "step": 380
+    },
+    {
+      "epoch": 3.018715596330275,
+      "grad_norm": 6.134914875030518,
+      "learning_rate": 5e-05,
+      "loss": 0.0485,
+      "step": 381
+    },
+    {
+      "epoch": 3.0190825688073395,
+      "grad_norm": 2.9193053245544434,
+      "learning_rate": 5e-05,
+      "loss": 0.4323,
+      "step": 382
+    },
+    {
+      "epoch": 3.019449541284404,
+      "grad_norm": 14.471090316772461,
+      "learning_rate": 5e-05,
+      "loss": 0.9457,
+      "step": 383
+    },
+    {
+      "epoch": 3.019816513761468,
+      "grad_norm": 0.19935713708400726,
+      "learning_rate": 5e-05,
+      "loss": 0.0088,
+      "step": 384
+    },
+    {
+      "epoch": 3.020183486238532,
+      "grad_norm": 5.313055515289307,
+      "learning_rate": 5e-05,
+      "loss": 0.3795,
+      "step": 385
+    },
+    {
+      "epoch": 3.0205504587155962,
+      "grad_norm": 0.11737678200006485,
+      "learning_rate": 5e-05,
+      "loss": 0.0048,
+      "step": 386
+    },
+    {
+      "epoch": 3.0209174311926605,
+      "grad_norm": 0.7507890462875366,
+      "learning_rate": 5e-05,
+      "loss": 0.024,
+      "step": 387
+    },
+    {
+      "epoch": 3.021284403669725,
+      "grad_norm": 0.38982895016670227,
+      "learning_rate": 5e-05,
+      "loss": 0.0159,
+      "step": 388
+    },
+    {
+      "epoch": 3.021651376146789,
+      "grad_norm": 13.67441177368164,
+      "learning_rate": 5e-05,
+      "loss": 0.305,
+      "step": 389
+    },
+    {
+      "epoch": 3.0220183486238534,
+      "grad_norm": 6.475154399871826,
+      "learning_rate": 5e-05,
+      "loss": 0.2253,
+      "step": 390
+    },
+    {
+      "epoch": 3.0223853211009173,
+      "grad_norm": 10.348758697509766,
+      "learning_rate": 5e-05,
+      "loss": 0.3454,
+      "step": 391
+    },
+    {
+      "epoch": 3.0227522935779816,
+      "grad_norm": 0.384656697511673,
+      "learning_rate": 5e-05,
+      "loss": 0.0133,
+      "step": 392
+    },
+    {
+      "epoch": 3.023119266055046,
+      "grad_norm": 0.22271838784217834,
+      "learning_rate": 5e-05,
+      "loss": 0.0091,
+      "step": 393
+    },
+    {
+      "epoch": 3.02348623853211,
+      "grad_norm": 0.17015402019023895,
+      "learning_rate": 5e-05,
+      "loss": 0.0069,
+      "step": 394
+    },
+    {
+      "epoch": 3.0238532110091745,
+      "grad_norm": 0.21346238255500793,
+      "learning_rate": 5e-05,
+      "loss": 0.0085,
+      "step": 395
+    },
+    {
+      "epoch": 3.0242201834862383,
+      "grad_norm": 0.24555295705795288,
+      "learning_rate": 5e-05,
+      "loss": 0.0108,
+      "step": 396
+    },
+    {
+      "epoch": 3.0245871559633026,
+      "grad_norm": 0.5599361658096313,
+      "learning_rate": 5e-05,
+      "loss": 0.0194,
+      "step": 397
+    },
+    {
+      "epoch": 3.024954128440367,
+      "grad_norm": 1.0569020509719849,
+      "learning_rate": 5e-05,
+      "loss": 0.0289,
+      "step": 398
+    },
+    {
+      "epoch": 3.0253211009174312,
+      "grad_norm": 0.1300659030675888,
+      "learning_rate": 5e-05,
+      "loss": 0.0057,
+      "step": 399
+    },
+    {
+      "epoch": 3.0256880733944955,
+      "grad_norm": 0.13957509398460388,
+      "learning_rate": 5e-05,
+      "loss": 0.0063,
+      "step": 400
+    },
+    {
+      "epoch": 3.02605504587156,
+      "grad_norm": 7.443248271942139,
+      "learning_rate": 5e-05,
+      "loss": 0.1163,
+      "step": 401
+    },
+    {
+      "epoch": 3.0264220183486237,
+      "grad_norm": 5.293422698974609,
+      "learning_rate": 5e-05,
+      "loss": 0.4377,
+      "step": 402
+    },
+    {
+      "epoch": 3.026788990825688,
+      "grad_norm": 0.09510613232851028,
+      "learning_rate": 5e-05,
+      "loss": 0.0043,
+      "step": 403
+    },
+    {
+      "epoch": 3.0271559633027523,
+      "grad_norm": 0.07236023992300034,
+      "learning_rate": 5e-05,
+      "loss": 0.0033,
+      "step": 404
+    },
+    {
+      "epoch": 3.0275229357798166,
+      "grad_norm": 0.6020928025245667,
+      "learning_rate": 5e-05,
+      "loss": 0.0159,
+      "step": 405
+    },
+    {
+      "epoch": 3.027889908256881,
+      "grad_norm": 0.07929949462413788,
+      "learning_rate": 5e-05,
+      "loss": 0.0036,
+      "step": 406
+    },
+    {
+      "epoch": 3.028256880733945,
+      "grad_norm": 0.0823131799697876,
+      "learning_rate": 5e-05,
+      "loss": 0.003,
+      "step": 407
+    },
+    {
+      "epoch": 3.028623853211009,
+      "grad_norm": 3.037116289138794,
+      "learning_rate": 5e-05,
+      "loss": 0.6657,
+      "step": 408
+    },
+    {
+      "epoch": 3.0289908256880733,
+      "grad_norm": 0.041281409561634064,
+      "learning_rate": 5e-05,
+      "loss": 0.0019,
+      "step": 409
+    },
+    {
+      "epoch": 3.0293577981651376,
+      "grad_norm": 0.01862535998225212,
+      "learning_rate": 5e-05,
+      "loss": 0.001,
+      "step": 410
+    },
+    {
+      "epoch": 3.029724770642202,
+      "grad_norm": 2.199099063873291,
+      "learning_rate": 5e-05,
+      "loss": 0.0107,
+      "step": 411
+    },
+    {
+      "epoch": 3.030091743119266,
+      "grad_norm": 3.811396598815918,
+      "learning_rate": 5e-05,
+      "loss": 0.5222,
+      "step": 412
+    },
+    {
+      "epoch": 3.03045871559633,
+      "grad_norm": 0.1260610818862915,
+      "learning_rate": 5e-05,
+      "loss": 0.0054,
+      "step": 413
+    },
+    {
+      "epoch": 3.0308256880733944,
+      "grad_norm": 7.156766414642334,
+      "learning_rate": 5e-05,
+      "loss": 0.9478,
+      "step": 414
+    },
+    {
+      "epoch": 3.0311926605504587,
+      "grad_norm": 9.729988098144531,
+      "learning_rate": 5e-05,
+      "loss": 0.5033,
+      "step": 415
+    },
+    {
+      "epoch": 3.031559633027523,
+      "grad_norm": 0.5098385214805603,
+      "learning_rate": 5e-05,
+      "loss": 0.0079,
+      "step": 416
+    },
+    {
+      "epoch": 3.0319266055045873,
+      "grad_norm": 27.979625701904297,
+      "learning_rate": 5e-05,
+      "loss": 0.1233,
+      "step": 417
+    },
+    {
+      "epoch": 3.0322935779816516,
+      "grad_norm": 3.594717264175415,
+      "learning_rate": 5e-05,
+      "loss": 0.6435,
+      "step": 418
+    },
+    {
+      "epoch": 3.0326605504587154,
+      "grad_norm": 0.21876324713230133,
+      "learning_rate": 5e-05,
+      "loss": 0.0082,
+      "step": 419
+    },
+    {
+      "epoch": 3.0330275229357797,
+      "grad_norm": 0.16088874638080597,
+      "learning_rate": 5e-05,
+      "loss": 0.0066,
+      "step": 420
+    },
+    {
+      "epoch": 3.033394495412844,
+      "grad_norm": 0.0883617103099823,
+      "learning_rate": 5e-05,
+      "loss": 0.0039,
+      "step": 421
+    },
+    {
+      "epoch": 3.0337614678899083,
+      "grad_norm": 5.720613956451416,
+      "learning_rate": 5e-05,
+      "loss": 0.0471,
+      "step": 422
+    },
+    {
+      "epoch": 3.0341284403669726,
+      "grad_norm": 4.0728254318237305,
+      "learning_rate": 5e-05,
+      "loss": 0.4594,
+      "step": 423
+    },
+    {
+      "epoch": 3.034495412844037,
+      "grad_norm": 0.08090603351593018,
+      "learning_rate": 5e-05,
+      "loss": 0.0037,
+      "step": 424
+    },
+    {
+      "epoch": 3.0348623853211008,
+      "grad_norm": 0.40366142988204956,
+      "learning_rate": 5e-05,
+      "loss": 0.017,
+      "step": 425
+    },
+    {
+      "epoch": 3.035229357798165,
+      "grad_norm": 0.14235635101795197,
+      "learning_rate": 5e-05,
+      "loss": 0.0065,
+      "step": 426
+    },
+    {
+      "epoch": 3.0355963302752293,
+      "grad_norm": 4.432626247406006,
+      "learning_rate": 5e-05,
+      "loss": 0.2543,
+      "step": 427
+    },
+    {
+      "epoch": 3.0359633027522936,
+      "grad_norm": 6.051278114318848,
+      "learning_rate": 5e-05,
+      "loss": 1.0286,
+      "step": 428
+    },
+    {
+      "epoch": 3.036330275229358,
+      "grad_norm": 0.3364741802215576,
+      "learning_rate": 5e-05,
+      "loss": 0.0112,
+      "step": 429
+    },
+    {
+      "epoch": 3.036697247706422,
+      "grad_norm": 3.4305264949798584,
+      "learning_rate": 5e-05,
+      "loss": 0.4368,
+      "step": 430
+    },
+    {
+      "epoch": 3.037064220183486,
+      "grad_norm": 0.3547607362270355,
+      "learning_rate": 5e-05,
+      "loss": 0.015,
+      "step": 431
+    },
+    {
+      "epoch": 3.0374311926605504,
+      "grad_norm": 0.28232720494270325,
+      "learning_rate": 5e-05,
+      "loss": 0.0126,
+      "step": 432
+    },
+    {
+      "epoch": 3.0377981651376147,
+      "grad_norm": 3.836404800415039,
+      "learning_rate": 5e-05,
+      "loss": 0.2958,
+      "step": 433
+    },
+    {
+      "epoch": 3.038165137614679,
+      "grad_norm": 0.6646902561187744,
+      "learning_rate": 5e-05,
+      "loss": 0.0283,
+      "step": 434
+    },
+    {
+      "epoch": 3.0385321100917433,
+      "grad_norm": 2.672466993331909,
+      "learning_rate": 5e-05,
+      "loss": 0.5025,
+      "step": 435
+    },
+    {
+      "epoch": 3.038899082568807,
+      "grad_norm": 0.44712260365486145,
+      "learning_rate": 5e-05,
+      "loss": 0.0214,
+      "step": 436
+    },
+    {
+      "epoch": 3.0392660550458714,
+      "grad_norm": 0.6815525889396667,
+      "learning_rate": 5e-05,
+      "loss": 0.0254,
+      "step": 437
+    },
+    {
+      "epoch": 3.0396330275229357,
+      "grad_norm": 6.344850540161133,
+      "learning_rate": 5e-05,
+      "loss": 0.1868,
+      "step": 438
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 6.064592361450195,
+      "learning_rate": 5e-05,
+      "loss": 0.2402,
+      "step": 439
+    },
+    {
+      "epoch": 3.0403669724770643,
+      "grad_norm": 16.158626556396484,
+      "learning_rate": 5e-05,
+      "loss": 0.1384,
+      "step": 440
+    },
+    {
+      "epoch": 3.0403669724770643,
+      "eval_accuracy": 0.5177514792899408,
+      "eval_loss": 1.3725639581680298,
+      "eval_runtime": 27.8621,
+      "eval_samples_per_second": 12.131,
+      "eval_steps_per_second": 1.543,
+      "step": 440
+    },
+    {
+      "epoch": 4.000366972477064,
+      "grad_norm": 0.25887179374694824,
+      "learning_rate": 5e-05,
+      "loss": 0.0116,
+      "step": 441
+    },
+    {
+      "epoch": 4.000733944954129,
+      "grad_norm": 0.14733389019966125,
+      "learning_rate": 5e-05,
+      "loss": 0.0067,
+      "step": 442
+    },
+    {
+      "epoch": 4.0011009174311924,
+      "grad_norm": 2.579551935195923,
+      "learning_rate": 5e-05,
+      "loss": 0.2906,
+      "step": 443
+    },
+    {
+      "epoch": 4.001467889908257,
+      "grad_norm": 0.3204147219657898,
+      "learning_rate": 5e-05,
+      "loss": 0.014,
+      "step": 444
+    },
+    {
+      "epoch": 4.001834862385321,
+      "grad_norm": 0.6737015247344971,
+      "learning_rate": 5e-05,
+      "loss": 0.0296,
+      "step": 445
+    },
+    {
+      "epoch": 4.002201834862385,
+      "grad_norm": 0.10879269242286682,
+      "learning_rate": 5e-05,
+      "loss": 0.0051,
+      "step": 446
+    },
+    {
+      "epoch": 4.00256880733945,
+      "grad_norm": 0.5761868357658386,
+      "learning_rate": 5e-05,
+      "loss": 0.0258,
+      "step": 447
+    },
+    {
+      "epoch": 4.0029357798165135,
+      "grad_norm": 0.17560754716396332,
+      "learning_rate": 5e-05,
+      "loss": 0.0078,
+      "step": 448
+    },
+    {
+      "epoch": 4.003302752293578,
+      "grad_norm": 0.38492515683174133,
+      "learning_rate": 5e-05,
+      "loss": 0.0101,
+      "step": 449
+    },
+    {
+      "epoch": 4.003669724770642,
+      "grad_norm": 0.16580255329608917,
+      "learning_rate": 5e-05,
+      "loss": 0.0075,
+      "step": 450
+    },
+    {
+      "epoch": 4.004036697247707,
+      "grad_norm": 7.445613384246826,
+      "learning_rate": 5e-05,
+      "loss": 0.1051,
+      "step": 451
+    },
+    {
+      "epoch": 4.004403669724771,
+      "grad_norm": 0.47299015522003174,
+      "learning_rate": 5e-05,
+      "loss": 0.0209,
+      "step": 452
+    },
+    {
+      "epoch": 4.0047706422018345,
+      "grad_norm": 5.689526081085205,
+      "learning_rate": 5e-05,
+      "loss": 0.9595,
+      "step": 453
+    },
+    {
+      "epoch": 4.005137614678899,
+      "grad_norm": 0.11032013595104218,
+      "learning_rate": 5e-05,
+      "loss": 0.0053,
+      "step": 454
+    },
+    {
+      "epoch": 4.005504587155963,
+      "grad_norm": 0.10374589264392853,
+      "learning_rate": 5e-05,
+      "loss": 0.0044,
+      "step": 455
+    },
+    {
+      "epoch": 4.005871559633028,
+      "grad_norm": 2.941288948059082,
+      "learning_rate": 5e-05,
+      "loss": 0.5899,
+      "step": 456
+    },
+    {
+      "epoch": 4.006238532110092,
+      "grad_norm": 0.14339561760425568,
+      "learning_rate": 5e-05,
+      "loss": 0.0066,
+      "step": 457
+    },
+    {
+      "epoch": 4.006605504587156,
+      "grad_norm": 0.08379548043012619,
+      "learning_rate": 5e-05,
+      "loss": 0.0037,
+      "step": 458
+    },
+    {
+      "epoch": 4.00697247706422,
+      "grad_norm": 0.16727282106876373,
+      "learning_rate": 5e-05,
+      "loss": 0.0074,
+      "step": 459
+    },
+    {
+      "epoch": 4.007339449541284,
+      "grad_norm": 0.15462461113929749,
+      "learning_rate": 5e-05,
+      "loss": 0.0067,
+      "step": 460
+    },
+    {
+      "epoch": 4.007706422018349,
+      "grad_norm": 0.4988681375980377,
+      "learning_rate": 5e-05,
+      "loss": 0.0146,
+      "step": 461
+    },
+    {
+      "epoch": 4.008073394495413,
+      "grad_norm": 10.861186027526855,
+      "learning_rate": 5e-05,
+      "loss": 0.7078,
+      "step": 462
+    },
+    {
+      "epoch": 4.008440366972477,
+      "grad_norm": 6.235146522521973,
+      "learning_rate": 5e-05,
+      "loss": 0.5016,
+      "step": 463
+    },
+    {
+      "epoch": 4.008807339449541,
+      "grad_norm": 33.146121978759766,
+      "learning_rate": 5e-05,
+      "loss": 0.5555,
+      "step": 464
+    },
+    {
+      "epoch": 4.009174311926605,
+      "grad_norm": 0.12248405069112778,
+      "learning_rate": 5e-05,
+      "loss": 0.0055,
+      "step": 465
+    },
+    {
+      "epoch": 4.00954128440367,
+      "grad_norm": 0.06532777845859528,
+      "learning_rate": 5e-05,
+      "loss": 0.0025,
+      "step": 466
+    },
+    {
+      "epoch": 4.009908256880734,
+      "grad_norm": 0.05290471389889717,
+      "learning_rate": 5e-05,
+      "loss": 0.0026,
+      "step": 467
+    },
+    {
+      "epoch": 4.010275229357799,
+      "grad_norm": 3.072108507156372,
+      "learning_rate": 5e-05,
+      "loss": 0.6163,
+      "step": 468
+    },
+    {
+      "epoch": 4.010642201834862,
+      "grad_norm": 26.443828582763672,
+      "learning_rate": 5e-05,
+      "loss": 0.3965,
+      "step": 469
+    },
+    {
+      "epoch": 4.011009174311926,
+      "grad_norm": 0.27494654059410095,
+      "learning_rate": 5e-05,
+      "loss": 0.0116,
+      "step": 470
+    },
+    {
+      "epoch": 4.011376146788991,
+      "grad_norm": 0.058719560503959656,
+      "learning_rate": 5e-05,
+      "loss": 0.0029,
+      "step": 471
+    },
+    {
+      "epoch": 4.011743119266055,
+      "grad_norm": 0.6328119039535522,
+      "learning_rate": 5e-05,
+      "loss": 0.0264,
+      "step": 472
+    },
+    {
+      "epoch": 4.01211009174312,
+      "grad_norm": 0.2899278402328491,
+      "learning_rate": 5e-05,
+      "loss": 0.0117,
+      "step": 473
+    },
+    {
+      "epoch": 4.0124770642201835,
+      "grad_norm": 0.16253183782100677,
+      "learning_rate": 5e-05,
+      "loss": 0.007,
+      "step": 474
+    },
+    {
+      "epoch": 4.012844036697247,
+      "grad_norm": 0.17079049348831177,
+      "learning_rate": 5e-05,
+      "loss": 0.0075,
+      "step": 475
+    },
+    {
+      "epoch": 4.013211009174312,
+      "grad_norm": 0.26379895210266113,
+      "learning_rate": 5e-05,
+      "loss": 0.0111,
+      "step": 476
+    },
+    {
+      "epoch": 4.013577981651376,
+      "grad_norm": 11.44698715209961,
+      "learning_rate": 5e-05,
+      "loss": 0.1339,
+      "step": 477
+    },
+    {
+      "epoch": 4.013944954128441,
+      "grad_norm": 0.20807303488254547,
+      "learning_rate": 5e-05,
+      "loss": 0.0088,
+      "step": 478
+    },
+    {
+      "epoch": 4.0143119266055045,
+      "grad_norm": 1.3006268739700317,
+      "learning_rate": 5e-05,
+      "loss": 0.0671,
+      "step": 479
+    },
+    {
+      "epoch": 4.014678899082568,
+      "grad_norm": 0.12875904142856598,
+      "learning_rate": 5e-05,
+      "loss": 0.0061,
+      "step": 480
+    },
+    {
+      "epoch": 4.015045871559633,
+      "grad_norm": 9.852110862731934,
+      "learning_rate": 5e-05,
+      "loss": 0.8958,
+      "step": 481
+    },
+    {
+      "epoch": 4.015412844036697,
+      "grad_norm": 0.14236675202846527,
+      "learning_rate": 5e-05,
+      "loss": 0.0062,
+      "step": 482
+    },
+    {
+      "epoch": 4.015779816513762,
+      "grad_norm": 0.10368365794420242,
+      "learning_rate": 5e-05,
+      "loss": 0.0047,
+      "step": 483
+    },
+    {
+      "epoch": 4.0161467889908256,
+      "grad_norm": 12.67244815826416,
+      "learning_rate": 5e-05,
+      "loss": 0.2997,
+      "step": 484
+    },
+    {
+      "epoch": 4.01651376146789,
+      "grad_norm": 0.24769026041030884,
+      "learning_rate": 5e-05,
+      "loss": 0.0085,
+      "step": 485
+    },
+    {
+      "epoch": 4.016880733944954,
+      "grad_norm": 0.3091563880443573,
+      "learning_rate": 5e-05,
+      "loss": 0.0098,
+      "step": 486
+    },
+    {
+      "epoch": 4.017247706422018,
+      "grad_norm": 10.314630508422852,
+      "learning_rate": 5e-05,
+      "loss": 0.1585,
+      "step": 487
+    },
+    {
+      "epoch": 4.017614678899083,
+      "grad_norm": 6.167266368865967,
+      "learning_rate": 5e-05,
+      "loss": 0.4035,
+      "step": 488
+    },
+    {
+      "epoch": 4.017981651376147,
+      "grad_norm": 8.486177444458008,
+      "learning_rate": 5e-05,
+      "loss": 0.3855,
+      "step": 489
+    },
+    {
+      "epoch": 4.018348623853211,
+      "grad_norm": 0.0736735537648201,
+      "learning_rate": 5e-05,
+      "loss": 0.0031,
+      "step": 490
+    },
+    {
+      "epoch": 4.018715596330275,
+      "grad_norm": 0.1895316243171692,
+      "learning_rate": 5e-05,
+      "loss": 0.0081,
+      "step": 491
+    },
+    {
+      "epoch": 4.019082568807339,
+      "grad_norm": 0.9338497519493103,
+      "learning_rate": 5e-05,
+      "loss": 0.0415,
+      "step": 492
+    },
+    {
+      "epoch": 4.019449541284404,
+      "grad_norm": 4.960600852966309,
+      "learning_rate": 5e-05,
+      "loss": 0.1473,
+      "step": 493
+    },
+    {
+      "epoch": 4.019816513761468,
+      "grad_norm": 0.07334944605827332,
+      "learning_rate": 5e-05,
+      "loss": 0.0033,
+      "step": 494
+    },
+    {
+      "epoch": 4.020183486238532,
+      "grad_norm": 0.15500545501708984,
+      "learning_rate": 5e-05,
+      "loss": 0.0065,
+      "step": 495
+    },
+    {
+      "epoch": 4.020550458715596,
+      "grad_norm": 0.06480821967124939,
+      "learning_rate": 5e-05,
+      "loss": 0.0029,
+      "step": 496
+    },
+    {
+      "epoch": 4.020917431192661,
+      "grad_norm": 0.32147422432899475,
+      "learning_rate": 5e-05,
+      "loss": 0.0077,
+      "step": 497
+    },
+    {
+      "epoch": 4.021284403669725,
+      "grad_norm": 2.988463878631592,
+      "learning_rate": 5e-05,
+      "loss": 0.6713,
+      "step": 498
+    },
+    {
+      "epoch": 4.021651376146789,
+      "grad_norm": 3.9235758781433105,
+      "learning_rate": 5e-05,
+      "loss": 0.2648,
+      "step": 499
+    },
+    {
+      "epoch": 4.022018348623853,
+      "grad_norm": 0.0843447595834732,
+      "learning_rate": 5e-05,
+      "loss": 0.0037,
+      "step": 500
+    },
+    {
+      "epoch": 4.022385321100917,
+      "grad_norm": 3.954972743988037,
+      "learning_rate": 5e-05,
+      "loss": 0.4679,
+      "step": 501
+    },
+    {
+      "epoch": 4.022752293577982,
+      "grad_norm": 4.359033584594727,
+      "learning_rate": 5e-05,
+      "loss": 0.4076,
+      "step": 502
+    },
+    {
+      "epoch": 4.023119266055046,
+      "grad_norm": 0.49513423442840576,
+      "learning_rate": 5e-05,
+      "loss": 0.0105,
+      "step": 503
+    },
+    {
+      "epoch": 4.02348623853211,
+      "grad_norm": 4.326723098754883,
+      "learning_rate": 5e-05,
+      "loss": 0.1771,
+      "step": 504
+    },
+    {
+      "epoch": 4.0238532110091745,
+      "grad_norm": 0.32503172755241394,
+      "learning_rate": 5e-05,
+      "loss": 0.0095,
+      "step": 505
+    },
+    {
+      "epoch": 4.024220183486238,
+      "grad_norm": 5.02016544342041,
+      "learning_rate": 5e-05,
+      "loss": 0.1406,
+      "step": 506
+    },
+    {
+      "epoch": 4.024587155963303,
+      "grad_norm": 9.759092330932617,
+      "learning_rate": 5e-05,
+      "loss": 0.2609,
+      "step": 507
+    },
+    {
+      "epoch": 4.024954128440367,
+      "grad_norm": 3.016780138015747,
+      "learning_rate": 5e-05,
+      "loss": 0.0916,
+      "step": 508
+    },
+    {
+      "epoch": 4.025321100917431,
+      "grad_norm": 3.435225248336792,
+      "learning_rate": 5e-05,
+      "loss": 0.1611,
+      "step": 509
+    },
+    {
+      "epoch": 4.0256880733944955,
+      "grad_norm": 13.51616382598877,
+      "learning_rate": 5e-05,
+      "loss": 0.4944,
+      "step": 510
+    },
+    {
+      "epoch": 4.026055045871559,
+      "grad_norm": 0.07008403539657593,
+      "learning_rate": 5e-05,
+      "loss": 0.0032,
+      "step": 511
+    },
+    {
+      "epoch": 4.026422018348624,
+      "grad_norm": 2.2766823768615723,
+      "learning_rate": 5e-05,
+      "loss": 0.0191,
+      "step": 512
+    },
+    {
+      "epoch": 4.026788990825688,
+      "grad_norm": 3.173746109008789,
+      "learning_rate": 5e-05,
+      "loss": 0.073,
+      "step": 513
+    },
+    {
+      "epoch": 4.027155963302753,
+      "grad_norm": 17.80497169494629,
+      "learning_rate": 5e-05,
+      "loss": 0.1241,
+      "step": 514
+    },
+    {
+      "epoch": 4.027522935779817,
+      "grad_norm": 0.15218138694763184,
+      "learning_rate": 5e-05,
+      "loss": 0.0066,
+      "step": 515
+    },
+    {
+      "epoch": 4.02788990825688,
+      "grad_norm": 19.21993064880371,
+      "learning_rate": 5e-05,
+      "loss": 0.2223,
+      "step": 516
+    },
+    {
+      "epoch": 4.028256880733945,
+      "grad_norm": 8.731097221374512,
+      "learning_rate": 5e-05,
+      "loss": 0.0971,
+      "step": 517
+    },
+    {
+      "epoch": 4.028623853211009,
+      "grad_norm": 0.6111873984336853,
+      "learning_rate": 5e-05,
+      "loss": 0.0135,
+      "step": 518
+    },
+    {
+      "epoch": 4.028990825688074,
+      "grad_norm": 4.371664047241211,
+      "learning_rate": 5e-05,
+      "loss": 0.0483,
+      "step": 519
+    },
+    {
+      "epoch": 4.029357798165138,
+      "grad_norm": 6.547307968139648,
+      "learning_rate": 5e-05,
+      "loss": 0.5885,
+      "step": 520
+    },
+    {
+      "epoch": 4.0297247706422015,
+      "grad_norm": 0.08560451120138168,
+      "learning_rate": 5e-05,
+      "loss": 0.0038,
+      "step": 521
+    },
+    {
+      "epoch": 4.030091743119266,
+      "grad_norm": 9.077963829040527,
+      "learning_rate": 5e-05,
+      "loss": 0.3581,
+      "step": 522
+    },
+    {
+      "epoch": 4.03045871559633,
+      "grad_norm": 10.413993835449219,
+      "learning_rate": 5e-05,
+      "loss": 0.5088,
+      "step": 523
+    },
+    {
+      "epoch": 4.030825688073395,
+      "grad_norm": 0.20086339116096497,
+      "learning_rate": 5e-05,
+      "loss": 0.0075,
+      "step": 524
+    },
+    {
+      "epoch": 4.031192660550459,
+      "grad_norm": 1.2687442302703857,
+      "learning_rate": 5e-05,
+      "loss": 0.0297,
+      "step": 525
+    },
+    {
+      "epoch": 4.0315596330275225,
+      "grad_norm": 0.49614784121513367,
+      "learning_rate": 5e-05,
+      "loss": 0.0101,
+      "step": 526
+    },
+    {
+      "epoch": 4.031926605504587,
+      "grad_norm": 1.610874891281128,
+      "learning_rate": 5e-05,
+      "loss": 0.0227,
+      "step": 527
+    },
+    {
+      "epoch": 4.032293577981651,
+      "grad_norm": 3.678208827972412,
+      "learning_rate": 5e-05,
+      "loss": 0.2042,
+      "step": 528
+    },
+    {
+      "epoch": 4.032660550458716,
+      "grad_norm": 12.703598022460938,
+      "learning_rate": 5e-05,
+      "loss": 0.2437,
+      "step": 529
+    },
+    {
+      "epoch": 4.03302752293578,
+      "grad_norm": 3.365461826324463,
+      "learning_rate": 5e-05,
+      "loss": 0.5984,
+      "step": 530
+    },
+    {
+      "epoch": 4.0333944954128444,
+      "grad_norm": 0.5475956201553345,
+      "learning_rate": 5e-05,
+      "loss": 0.0118,
+      "step": 531
+    },
+    {
+      "epoch": 4.033761467889908,
+      "grad_norm": 0.5540260076522827,
+      "learning_rate": 5e-05,
+      "loss": 0.0095,
+      "step": 532
+    },
+    {
+      "epoch": 4.034128440366972,
+      "grad_norm": 3.213998317718506,
+      "learning_rate": 5e-05,
+      "loss": 0.6399,
+      "step": 533
+    },
+    {
+      "epoch": 4.034495412844037,
+      "grad_norm": 4.15159797668457,
+      "learning_rate": 5e-05,
+      "loss": 0.4785,
+      "step": 534
+    },
+    {
+      "epoch": 4.034862385321101,
+      "grad_norm": 0.1713489145040512,
+      "learning_rate": 5e-05,
+      "loss": 0.007,
+      "step": 535
+    },
+    {
+      "epoch": 4.0352293577981655,
+      "grad_norm": 0.7636466026306152,
+      "learning_rate": 5e-05,
+      "loss": 0.0204,
+      "step": 536
+    },
+    {
+      "epoch": 4.035596330275229,
+      "grad_norm": 0.16516229510307312,
+      "learning_rate": 5e-05,
+      "loss": 0.0079,
+      "step": 537
+    },
+    {
+      "epoch": 4.035963302752293,
+      "grad_norm": 4.2119364738464355,
+      "learning_rate": 5e-05,
+      "loss": 0.068,
+      "step": 538
+    },
+    {
+      "epoch": 4.036330275229358,
+      "grad_norm": 11.787577629089355,
+      "learning_rate": 5e-05,
+      "loss": 0.2251,
+      "step": 539
+    },
+    {
+      "epoch": 4.036697247706422,
+      "grad_norm": 0.15907607972621918,
+      "learning_rate": 5e-05,
+      "loss": 0.0074,
+      "step": 540
+    },
+    {
+      "epoch": 4.0370642201834865,
+      "grad_norm": 0.08161436766386032,
+      "learning_rate": 5e-05,
+      "loss": 0.0034,
+      "step": 541
+    },
+    {
+      "epoch": 4.03743119266055,
+      "grad_norm": 1.5476641654968262,
+      "learning_rate": 5e-05,
+      "loss": 0.0411,
+      "step": 542
+    },
+    {
+      "epoch": 4.037798165137614,
+      "grad_norm": 0.31121766567230225,
+      "learning_rate": 5e-05,
+      "loss": 0.0126,
+      "step": 543
+    },
+    {
+      "epoch": 4.038165137614679,
+      "grad_norm": 2.8357789516448975,
+      "learning_rate": 5e-05,
+      "loss": 0.5292,
+      "step": 544
+    },
+    {
+      "epoch": 4.038532110091743,
+      "grad_norm": 1.322967290878296,
+      "learning_rate": 5e-05,
+      "loss": 0.0271,
+      "step": 545
+    },
+    {
+      "epoch": 4.038899082568808,
+      "grad_norm": 0.09862636029720306,
+      "learning_rate": 5e-05,
+      "loss": 0.0042,
+      "step": 546
+    },
+    {
+      "epoch": 4.039266055045871,
+      "grad_norm": 1.0618364810943604,
+      "learning_rate": 5e-05,
+      "loss": 0.0159,
+      "step": 547
+    },
+    {
+      "epoch": 4.039633027522936,
+      "grad_norm": 2.8359313011169434,
+      "learning_rate": 5e-05,
+      "loss": 0.5355,
+      "step": 548
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 1.2886866331100464,
+      "learning_rate": 5e-05,
+      "loss": 0.0103,
+      "step": 549
+    },
+    {
+      "epoch": 4.040366972477064,
+      "grad_norm": 9.333901405334473,
+      "learning_rate": 5e-05,
+      "loss": 0.513,
+      "step": 550
+    },
+    {
+      "epoch": 4.040366972477064,
+      "eval_accuracy": 0.3727810650887574,
+      "eval_loss": 2.5340259075164795,
+      "eval_runtime": 26.0301,
+      "eval_samples_per_second": 12.985,
+      "eval_steps_per_second": 1.652,
+      "step": 550
+    },
+    {
+      "epoch": 5.000366972477064,
+      "grad_norm": 0.09391362965106964,
+      "learning_rate": 5e-05,
+      "loss": 0.0041,
+      "step": 551
+    },
+    {
+      "epoch": 5.000733944954129,
+      "grad_norm": 6.501461982727051,
+      "learning_rate": 5e-05,
+      "loss": 0.726,
+      "step": 552
+    },
+    {
+      "epoch": 5.0011009174311924,
+      "grad_norm": 0.11173416674137115,
+      "learning_rate": 5e-05,
+      "loss": 0.0049,
+      "step": 553
+    },
+    {
+      "epoch": 5.001467889908257,
+      "grad_norm": 11.773106575012207,
+      "learning_rate": 5e-05,
+      "loss": 0.2458,
+      "step": 554
+    },
+    {
+      "epoch": 5.001834862385321,
+      "grad_norm": 3.067979097366333,
+      "learning_rate": 5e-05,
+      "loss": 0.4383,
+      "step": 555
+    },
+    {
+      "epoch": 5.002201834862385,
+      "grad_norm": 0.28603827953338623,
+      "learning_rate": 5e-05,
+      "loss": 0.0127,
+      "step": 556
+    },
+    {
+      "epoch": 5.00256880733945,
+      "grad_norm": 2.988393545150757,
+      "learning_rate": 5e-05,
+      "loss": 0.0435,
+      "step": 557
+    },
+    {
+      "epoch": 5.0029357798165135,
+      "grad_norm": 3.135219097137451,
+      "learning_rate": 5e-05,
+      "loss": 0.4889,
+      "step": 558
+    },
+    {
+      "epoch": 5.003302752293578,
+      "grad_norm": 0.22969892621040344,
+      "learning_rate": 5e-05,
+      "loss": 0.0102,
+      "step": 559
+    },
+    {
+      "epoch": 5.003669724770642,
+      "grad_norm": 7.6010589599609375,
+      "learning_rate": 5e-05,
+      "loss": 0.0448,
+      "step": 560
+    },
+    {
+      "epoch": 5.004036697247707,
+      "grad_norm": 0.4112565815448761,
+      "learning_rate": 5e-05,
+      "loss": 0.018,
+      "step": 561
+    },
+    {
+      "epoch": 5.004403669724771,
+      "grad_norm": 10.279380798339844,
+      "learning_rate": 5e-05,
+      "loss": 0.0823,
+      "step": 562
+    },
+    {
+      "epoch": 5.0047706422018345,
+      "grad_norm": 1.4434973001480103,
+      "learning_rate": 5e-05,
+      "loss": 0.0266,
+      "step": 563
+    },
+    {
+      "epoch": 5.005137614678899,
+      "grad_norm": 0.5896565914154053,
+      "learning_rate": 5e-05,
+      "loss": 0.0252,
+      "step": 564
+    },
+    {
+      "epoch": 5.005504587155963,
+      "grad_norm": 1.1192433834075928,
+      "learning_rate": 5e-05,
+      "loss": 0.0551,
+      "step": 565
+    },
+    {
+      "epoch": 5.005871559633028,
+      "grad_norm": 5.902185440063477,
+      "learning_rate": 5e-05,
+      "loss": 0.2355,
+      "step": 566
+    },
+    {
+      "epoch": 5.006238532110092,
+      "grad_norm": 12.040647506713867,
+      "learning_rate": 5e-05,
+      "loss": 0.6534,
+      "step": 567
+    },
+    {
+      "epoch": 5.006605504587156,
+      "grad_norm": 0.6709261536598206,
+      "learning_rate": 5e-05,
+      "loss": 0.0285,
+      "step": 568
+    },
+    {
+      "epoch": 5.00697247706422,
+      "grad_norm": 0.9721537232398987,
+      "learning_rate": 5e-05,
+      "loss": 0.0402,
+      "step": 569
+    },
+    {
+      "epoch": 5.007339449541284,
+      "grad_norm": 3.0798592567443848,
+      "learning_rate": 5e-05,
+      "loss": 0.0597,
+      "step": 570
+    },
+    {
+      "epoch": 5.007706422018349,
+      "grad_norm": 18.241010665893555,
+      "learning_rate": 5e-05,
+      "loss": 0.1666,
+      "step": 571
+    },
+    {
+      "epoch": 5.008073394495413,
+      "grad_norm": 17.412858963012695,
+      "learning_rate": 5e-05,
+      "loss": 0.6353,
+      "step": 572
+    },
+    {
+      "epoch": 5.008440366972477,
+      "grad_norm": 4.945258140563965,
+      "learning_rate": 5e-05,
+      "loss": 0.1863,
+      "step": 573
+    },
+    {
+      "epoch": 5.008807339449541,
+      "grad_norm": 0.41937756538391113,
+      "learning_rate": 5e-05,
+      "loss": 0.0153,
+      "step": 574
+    },
+    {
+      "epoch": 5.009174311926605,
+      "grad_norm": 10.010263442993164,
+      "learning_rate": 5e-05,
+      "loss": 0.4682,
+      "step": 575
+    },
+    {
+      "epoch": 5.00954128440367,
+      "grad_norm": 0.16260719299316406,
+      "learning_rate": 5e-05,
+      "loss": 0.0071,
+      "step": 576
+    },
+    {
+      "epoch": 5.009908256880734,
+      "grad_norm": 6.520624160766602,
+      "learning_rate": 5e-05,
+      "loss": 0.1814,
+      "step": 577
+    },
+    {
+      "epoch": 5.010275229357799,
+      "grad_norm": 3.0347862243652344,
+      "learning_rate": 5e-05,
+      "loss": 0.0125,
+      "step": 578
+    },
+    {
+      "epoch": 5.010642201834862,
+      "grad_norm": 3.8470888137817383,
+      "learning_rate": 5e-05,
+      "loss": 0.2867,
+      "step": 579
+    },
+    {
+      "epoch": 5.011009174311926,
+      "grad_norm": 2.9752800464630127,
+      "learning_rate": 5e-05,
+      "loss": 0.0245,
+      "step": 580
+    },
+    {
+      "epoch": 5.011376146788991,
+      "grad_norm": 0.1409146785736084,
+      "learning_rate": 5e-05,
+      "loss": 0.006,
+      "step": 581
+    },
+    {
+      "epoch": 5.011743119266055,
+      "grad_norm": 2.1914350986480713,
+      "learning_rate": 5e-05,
+      "loss": 0.1164,
+      "step": 582
+    },
+    {
+      "epoch": 5.01211009174312,
+      "grad_norm": 0.06727621704339981,
+      "learning_rate": 5e-05,
+      "loss": 0.003,
+      "step": 583
+    },
+    {
+      "epoch": 5.0124770642201835,
+      "grad_norm": 4.671161651611328,
+      "learning_rate": 5e-05,
+      "loss": 0.5731,
+      "step": 584
+    },
+    {
+      "epoch": 5.012844036697247,
+      "grad_norm": 0.4343807101249695,
+      "learning_rate": 5e-05,
+      "loss": 0.01,
+      "step": 585
+    },
+    {
+      "epoch": 5.013211009174312,
+      "grad_norm": 2.892256736755371,
+      "learning_rate": 5e-05,
+      "loss": 0.0594,
+      "step": 586
+    },
+    {
+      "epoch": 5.013577981651376,
+      "grad_norm": 7.202817440032959,
+      "learning_rate": 5e-05,
+      "loss": 0.2304,
+      "step": 587
+    },
+    {
+      "epoch": 5.013944954128441,
+      "grad_norm": 0.25226378440856934,
+      "learning_rate": 5e-05,
+      "loss": 0.0094,
+      "step": 588
+    },
+    {
+      "epoch": 5.0143119266055045,
+      "grad_norm": 0.08829619735479355,
+      "learning_rate": 5e-05,
+      "loss": 0.0036,
+      "step": 589
+    },
+    {
+      "epoch": 5.014678899082568,
+      "grad_norm": 5.163284778594971,
+      "learning_rate": 5e-05,
+      "loss": 0.4962,
+      "step": 590
+    },
+    {
+      "epoch": 5.015045871559633,
+      "grad_norm": 0.06816033273935318,
+      "learning_rate": 5e-05,
+      "loss": 0.003,
+      "step": 591
+    },
+    {
+      "epoch": 5.015412844036697,
+      "grad_norm": 0.07388438284397125,
+      "learning_rate": 5e-05,
+      "loss": 0.0032,
+      "step": 592
+    },
+    {
+      "epoch": 5.015779816513762,
+      "grad_norm": 4.521554946899414,
+      "learning_rate": 5e-05,
+      "loss": 0.104,
+      "step": 593
+    },
+    {
+      "epoch": 5.0161467889908256,
+      "grad_norm": 0.08599952608346939,
+      "learning_rate": 5e-05,
+      "loss": 0.0038,
+      "step": 594
+    },
+    {
+      "epoch": 5.01651376146789,
+      "grad_norm": 0.05113305151462555,
+      "learning_rate": 5e-05,
+      "loss": 0.0023,
+      "step": 595
+    },
+    {
+      "epoch": 5.016880733944954,
+      "grad_norm": 0.09497366845607758,
+      "learning_rate": 5e-05,
+      "loss": 0.0038,
+      "step": 596
+    },
+    {
+      "epoch": 5.017247706422018,
+      "grad_norm": 0.0758247971534729,
+      "learning_rate": 5e-05,
+      "loss": 0.0034,
+      "step": 597
+    },
+    {
+      "epoch": 5.017614678899083,
+      "grad_norm": 0.06351546943187714,
+      "learning_rate": 5e-05,
+      "loss": 0.0029,
+      "step": 598
+    },
+    {
+      "epoch": 5.017981651376147,
+      "grad_norm": 1.9657553434371948,
+      "learning_rate": 5e-05,
+      "loss": 0.0544,
+      "step": 599
+    },
+    {
+      "epoch": 5.018348623853211,
+      "grad_norm": 0.7355825901031494,
+      "learning_rate": 5e-05,
+      "loss": 0.0139,
+      "step": 600
+    },
+    {
+      "epoch": 5.018715596330275,
+      "grad_norm": 0.05275749787688255,
+      "learning_rate": 5e-05,
+      "loss": 0.0024,
+      "step": 601
+    },
+    {
+      "epoch": 5.019082568807339,
+      "grad_norm": 0.06981419026851654,
+      "learning_rate": 5e-05,
+      "loss": 0.0031,
+      "step": 602
+    },
+    {
+      "epoch": 5.019449541284404,
+      "grad_norm": 0.07637110352516174,
+      "learning_rate": 5e-05,
+      "loss": 0.0034,
+      "step": 603
+    },
+    {
+      "epoch": 5.019816513761468,
+      "grad_norm": 0.20633555948734283,
+      "learning_rate": 5e-05,
+      "loss": 0.005,
+      "step": 604
+    },
+    {
+      "epoch": 5.020183486238532,
+      "grad_norm": 0.06374866515398026,
+      "learning_rate": 5e-05,
+      "loss": 0.0028,
+      "step": 605
+    },
+    {
+      "epoch": 5.020550458715596,
+      "grad_norm": 19.022951126098633,
+      "learning_rate": 5e-05,
+      "loss": 0.3606,
+      "step": 606
+    },
+    {
+      "epoch": 5.020917431192661,
+      "grad_norm": 0.06089720502495766,
+      "learning_rate": 5e-05,
+      "loss": 0.0027,
+      "step": 607
+    },
+    {
+      "epoch": 5.021284403669725,
+      "grad_norm": 0.06481636315584183,
+      "learning_rate": 5e-05,
+      "loss": 0.0028,
+      "step": 608
+    },
+    {
+      "epoch": 5.021651376146789,
+      "grad_norm": 0.17815494537353516,
+      "learning_rate": 5e-05,
+      "loss": 0.0024,
+      "step": 609
+    },
+    {
+      "epoch": 5.022018348623853,
+      "grad_norm": 0.5394445657730103,
+      "learning_rate": 5e-05,
+      "loss": 0.019,
+      "step": 610
+    },
+    {
+      "epoch": 5.022385321100917,
+      "grad_norm": 0.04871244728565216,
+      "learning_rate": 5e-05,
+      "loss": 0.0022,
+      "step": 611
+    },
+    {
+      "epoch": 5.022752293577982,
+      "grad_norm": 3.1479954719543457,
+      "learning_rate": 5e-05,
+      "loss": 0.0078,
+      "step": 612
+    },
+    {
+      "epoch": 5.023119266055046,
+      "grad_norm": 5.6066718101501465,
+      "learning_rate": 5e-05,
+      "loss": 0.5926,
+      "step": 613
+    },
+    {
+      "epoch": 5.02348623853211,
+      "grad_norm": 16.42903709411621,
+      "learning_rate": 5e-05,
+      "loss": 0.3563,
+      "step": 614
+    },
+    {
+      "epoch": 5.0238532110091745,
+      "grad_norm": 6.7477946281433105,
+      "learning_rate": 5e-05,
+      "loss": 0.0641,
+      "step": 615
+    },
+    {
+      "epoch": 5.024220183486238,
+      "grad_norm": 0.03492635115981102,
+      "learning_rate": 5e-05,
+      "loss": 0.0016,
+      "step": 616
+    },
+    {
+      "epoch": 5.024587155963303,
+      "grad_norm": 24.751249313354492,
+      "learning_rate": 5e-05,
+      "loss": 0.0657,
+      "step": 617
+    },
+    {
+      "epoch": 5.024954128440367,
+      "grad_norm": 0.059814803302288055,
+      "learning_rate": 5e-05,
+      "loss": 0.0025,
+      "step": 618
+    },
+    {
+      "epoch": 5.025321100917431,
+      "grad_norm": 0.0341038703918457,
+      "learning_rate": 5e-05,
+      "loss": 0.0016,
+      "step": 619
+    },
+    {
+      "epoch": 5.0256880733944955,
+      "grad_norm": 0.09134375303983688,
+      "learning_rate": 5e-05,
+      "loss": 0.0036,
+      "step": 620
+    },
+    {
+      "epoch": 5.026055045871559,
+      "grad_norm": 16.1168270111084,
+      "learning_rate": 5e-05,
+      "loss": 0.3515,
+      "step": 621
+    },
+    {
+      "epoch": 5.026422018348624,
+      "grad_norm": 7.758620262145996,
+      "learning_rate": 5e-05,
+      "loss": 0.0657,
+      "step": 622
+    },
+    {
+      "epoch": 5.026788990825688,
+      "grad_norm": 15.976603507995605,
+      "learning_rate": 5e-05,
+      "loss": 0.8287,
+      "step": 623
+    },
+    {
+      "epoch": 5.027155963302753,
+      "grad_norm": 0.08052210509777069,
+      "learning_rate": 5e-05,
+      "loss": 0.0034,
+      "step": 624
+    },
+    {
+      "epoch": 5.027522935779817,
+      "grad_norm": 2.033261775970459,
+      "learning_rate": 5e-05,
+      "loss": 0.0458,
+      "step": 625
+    },
+    {
+      "epoch": 5.02788990825688,
+      "grad_norm": 5.506579399108887,
+      "learning_rate": 5e-05,
+      "loss": 0.7548,
+      "step": 626
+    },
+    {
+      "epoch": 5.028256880733945,
+      "grad_norm": 0.1303037852048874,
+      "learning_rate": 5e-05,
+      "loss": 0.003,
+      "step": 627
+    },
+    {
+      "epoch": 5.028623853211009,
+      "grad_norm": 48.042755126953125,
+      "learning_rate": 5e-05,
+      "loss": 0.2461,
+      "step": 628
+    },
+    {
+      "epoch": 5.028990825688074,
+      "grad_norm": 0.7187135815620422,
+      "learning_rate": 5e-05,
+      "loss": 0.0121,
+      "step": 629
+    },
+    {
+      "epoch": 5.029357798165138,
+      "grad_norm": 0.13123072683811188,
+      "learning_rate": 5e-05,
+      "loss": 0.0034,
+      "step": 630
+    },
+    {
+      "epoch": 5.0297247706422015,
+      "grad_norm": 7.5857253074646,
+      "learning_rate": 5e-05,
+      "loss": 0.2771,
+      "step": 631
+    },
+    {
+      "epoch": 5.030091743119266,
+      "grad_norm": 24.300785064697266,
+      "learning_rate": 5e-05,
+      "loss": 1.1828,
+      "step": 632
+    },
+    {
+      "epoch": 5.03045871559633,
+      "grad_norm": 0.16587497293949127,
+      "learning_rate": 5e-05,
+      "loss": 0.0049,
+      "step": 633
+    },
+    {
+      "epoch": 5.030825688073395,
+      "grad_norm": 0.12736903131008148,
+      "learning_rate": 5e-05,
+      "loss": 0.0023,
+      "step": 634
+    },
+    {
+      "epoch": 5.031192660550459,
+      "grad_norm": 0.04495488107204437,
+      "learning_rate": 5e-05,
+      "loss": 0.002,
+      "step": 635
+    },
+    {
+      "epoch": 5.0315596330275225,
+      "grad_norm": 7.888302326202393,
+      "learning_rate": 5e-05,
+      "loss": 1.3279,
+      "step": 636
+    },
+    {
+      "epoch": 5.031926605504587,
+      "grad_norm": 0.026433052495121956,
+      "learning_rate": 5e-05,
+      "loss": 0.0012,
+      "step": 637
+    },
+    {
+      "epoch": 5.032293577981651,
+      "grad_norm": 0.043104901909828186,
+      "learning_rate": 5e-05,
+      "loss": 0.0019,
+      "step": 638
+    },
+    {
+      "epoch": 5.032660550458716,
+      "grad_norm": 3.0377540588378906,
+      "learning_rate": 5e-05,
+      "loss": 0.7431,
+      "step": 639
+    },
+    {
+      "epoch": 5.03302752293578,
+      "grad_norm": 0.06931830197572708,
+      "learning_rate": 5e-05,
+      "loss": 0.003,
+      "step": 640
+    },
+    {
+      "epoch": 5.0333944954128444,
+      "grad_norm": 0.08807898312807083,
+      "learning_rate": 5e-05,
+      "loss": 0.0042,
+      "step": 641
+    },
+    {
+      "epoch": 5.033761467889908,
+      "grad_norm": 0.03736937791109085,
+      "learning_rate": 5e-05,
+      "loss": 0.0017,
+      "step": 642
+    },
+    {
+      "epoch": 5.034128440366972,
+      "grad_norm": 0.09845203161239624,
+      "learning_rate": 5e-05,
+      "loss": 0.0042,
+      "step": 643
+    },
+    {
+      "epoch": 5.034495412844037,
+      "grad_norm": 0.14018844068050385,
+      "learning_rate": 5e-05,
+      "loss": 0.0057,
+      "step": 644
+    },
+    {
+      "epoch": 5.034862385321101,
+      "grad_norm": 2.9101014137268066,
+      "learning_rate": 5e-05,
+      "loss": 0.6584,
+      "step": 645
+    },
+    {
+      "epoch": 5.0352293577981655,
+      "grad_norm": 0.09265273809432983,
+      "learning_rate": 5e-05,
+      "loss": 0.0039,
+      "step": 646
+    },
+    {
+      "epoch": 5.035596330275229,
+      "grad_norm": 2.9551618099212646,
+      "learning_rate": 5e-05,
+      "loss": 0.6603,
+      "step": 647
+    },
+    {
+      "epoch": 5.035963302752293,
+      "grad_norm": 0.13875389099121094,
+      "learning_rate": 5e-05,
+      "loss": 0.0062,
+      "step": 648
+    },
+    {
+      "epoch": 5.036330275229358,
+      "grad_norm": 3.770094871520996,
+      "learning_rate": 5e-05,
+      "loss": 0.108,
+      "step": 649
+    },
+    {
+      "epoch": 5.036697247706422,
+      "grad_norm": 3.043125867843628,
+      "learning_rate": 5e-05,
+      "loss": 0.5091,
+      "step": 650
+    },
+    {
+      "epoch": 5.0370642201834865,
+      "grad_norm": 2.8449511528015137,
+      "learning_rate": 5e-05,
+      "loss": 0.4505,
+      "step": 651
+    },
+    {
+      "epoch": 5.03743119266055,
+      "grad_norm": 4.5402655601501465,
+      "learning_rate": 5e-05,
+      "loss": 0.0868,
+      "step": 652
+    },
+    {
+      "epoch": 5.037798165137614,
+      "grad_norm": 3.074416399002075,
+      "learning_rate": 5e-05,
+      "loss": 0.4921,
+      "step": 653
+    },
+    {
+      "epoch": 5.038165137614679,
+      "grad_norm": 3.197272539138794,
+      "learning_rate": 5e-05,
+      "loss": 0.0587,
+      "step": 654
+    },
+    {
+      "epoch": 5.038532110091743,
+      "grad_norm": 1.8000171184539795,
+      "learning_rate": 5e-05,
+      "loss": 0.0433,
+      "step": 655
+    },
+    {
+      "epoch": 5.038899082568808,
+      "grad_norm": 0.9409215450286865,
+      "learning_rate": 5e-05,
+      "loss": 0.039,
+      "step": 656
+    },
+    {
+      "epoch": 5.039266055045871,
+      "grad_norm": 0.7487952709197998,
+      "learning_rate": 5e-05,
+      "loss": 0.0322,
+      "step": 657
+    },
+    {
+      "epoch": 5.039633027522936,
+      "grad_norm": 0.6303098201751709,
+      "learning_rate": 5e-05,
+      "loss": 0.0266,
+      "step": 658
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 0.5054812431335449,
+      "learning_rate": 5e-05,
+      "loss": 0.0214,
+      "step": 659
+    },
+    {
+      "epoch": 5.040366972477064,
+      "grad_norm": 0.3537321984767914,
+      "learning_rate": 5e-05,
+      "loss": 0.0156,
+      "step": 660
+    },
+    {
+      "epoch": 5.040366972477064,
+      "eval_accuracy": 0.20414201183431951,
+      "eval_loss": 2.348670482635498,
+      "eval_runtime": 25.9774,
+      "eval_samples_per_second": 13.011,
+      "eval_steps_per_second": 1.655,
+      "step": 660
+    },
+    {
+      "epoch": 6.000366972477064,
+      "grad_norm": 2.816509962081909,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3477,
+      "step": 661
+    },
+    {
+      "epoch": 6.000733944954129,
+      "grad_norm": 0.7520211338996887,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0221,
+      "step": 662
+    },
+    {
+      "epoch": 6.0011009174311924,
+      "grad_norm": 0.31658926606178284,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0177,
+      "step": 663
+    },
+    {
+      "epoch": 6.001467889908257,
+      "grad_norm": 0.30918991565704346,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0141,
+      "step": 664
+    },
+    {
+      "epoch": 6.001834862385321,
+      "grad_norm": 0.465072900056839,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0201,
+      "step": 665
+    },
+    {
+      "epoch": 6.002201834862385,
+      "grad_norm": 0.20198267698287964,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0091,
+      "step": 666
+    },
+    {
+      "epoch": 6.00256880733945,
+      "grad_norm": 0.206312358379364,
+      "learning_rate": 2.5e-05,
+      "loss": 0.009,
+      "step": 667
+    },
+    {
+      "epoch": 6.0029357798165135,
+      "grad_norm": 0.7710257172584534,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0343,
+      "step": 668
+    },
+    {
+      "epoch": 6.003302752293578,
+      "grad_norm": 0.23595662415027618,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0104,
+      "step": 669
+    },
+    {
+      "epoch": 6.003669724770642,
+      "grad_norm": 0.2505616843700409,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0117,
+      "step": 670
+    },
+    {
+      "epoch": 6.004036697247707,
+      "grad_norm": 9.189667701721191,
+      "learning_rate": 2.5e-05,
+      "loss": 0.276,
+      "step": 671
+    },
+    {
+      "epoch": 6.004403669724771,
+      "grad_norm": 1.080255150794983,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0478,
+      "step": 672
+    },
+    {
+      "epoch": 6.0047706422018345,
+      "grad_norm": 0.3962284028530121,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0181,
+      "step": 673
+    },
+    {
+      "epoch": 6.005137614678899,
+      "grad_norm": 0.14254595339298248,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0062,
+      "step": 674
+    },
+    {
+      "epoch": 6.005504587155963,
+      "grad_norm": 0.09690891206264496,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0043,
+      "step": 675
+    },
+    {
+      "epoch": 6.005871559633028,
+      "grad_norm": 0.08758708834648132,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0038,
+      "step": 676
+    },
+    {
+      "epoch": 6.006238532110092,
+      "grad_norm": 0.1661381870508194,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0072,
+      "step": 677
+    },
+    {
+      "epoch": 6.006605504587156,
+      "grad_norm": 0.3507902920246124,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0155,
+      "step": 678
+    },
+    {
+      "epoch": 6.00697247706422,
+      "grad_norm": 3.394791841506958,
+      "learning_rate": 2.5e-05,
+      "loss": 0.503,
+      "step": 679
+    },
+    {
+      "epoch": 6.007339449541284,
+      "grad_norm": 0.17461171746253967,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0079,
+      "step": 680
+    },
+    {
+      "epoch": 6.007706422018349,
+      "grad_norm": 0.110337994992733,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0049,
+      "step": 681
+    },
+    {
+      "epoch": 6.008073394495413,
+      "grad_norm": 0.14993230998516083,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0064,
+      "step": 682
+    },
+    {
+      "epoch": 6.008440366972477,
+      "grad_norm": 0.24526509642601013,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0104,
+      "step": 683
+    },
+    {
+      "epoch": 6.008807339449541,
+      "grad_norm": 2.4464268684387207,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0388,
+      "step": 684
+    },
+    {
+      "epoch": 6.009174311926605,
+      "grad_norm": 0.11476343125104904,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0049,
+      "step": 685
+    },
+    {
+      "epoch": 6.00954128440367,
+      "grad_norm": 0.10482797771692276,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0046,
+      "step": 686
+    },
+    {
+      "epoch": 6.009908256880734,
+      "grad_norm": 0.09792350232601166,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0045,
+      "step": 687
+    },
+    {
+      "epoch": 6.010275229357799,
+      "grad_norm": 3.136338472366333,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3727,
+      "step": 688
+    },
+    {
+      "epoch": 6.010642201834862,
+      "grad_norm": 0.09327653050422668,
+      "learning_rate": 2.5e-05,
+      "loss": 0.004,
+      "step": 689
+    },
+    {
+      "epoch": 6.011009174311926,
+      "grad_norm": 0.13465583324432373,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0059,
+      "step": 690
+    },
+    {
+      "epoch": 6.011376146788991,
+      "grad_norm": 0.0664028450846672,
+      "learning_rate": 2.5e-05,
+      "loss": 0.003,
+      "step": 691
+    },
+    {
+      "epoch": 6.011743119266055,
+      "grad_norm": 0.10982511192560196,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0048,
+      "step": 692
+    },
+    {
+      "epoch": 6.01211009174312,
+      "grad_norm": 4.2351179122924805,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5469,
+      "step": 693
+    },
+    {
+      "epoch": 6.0124770642201835,
+      "grad_norm": 45.049598693847656,
+      "learning_rate": 2.5e-05,
+      "loss": 0.8746,
+      "step": 694
+    },
+    {
+      "epoch": 6.012844036697247,
+      "grad_norm": 0.1141386479139328,
+      "learning_rate": 2.5e-05,
+      "loss": 0.005,
+      "step": 695
+    },
+    {
+      "epoch": 6.013211009174312,
+      "grad_norm": 0.13219565153121948,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0059,
+      "step": 696
+    },
+    {
+      "epoch": 6.013577981651376,
+      "grad_norm": 0.1607501357793808,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0067,
+      "step": 697
+    },
+    {
+      "epoch": 6.013944954128441,
+      "grad_norm": 0.5861008167266846,
+      "learning_rate": 2.5e-05,
+      "loss": 0.018,
+      "step": 698
+    },
+    {
+      "epoch": 6.0143119266055045,
+      "grad_norm": 7.0651421546936035,
+      "learning_rate": 2.5e-05,
+      "loss": 1.133,
+      "step": 699
+    },
+    {
+      "epoch": 6.014678899082568,
+      "grad_norm": 1.4427745342254639,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0384,
+      "step": 700
+    },
+    {
+      "epoch": 6.015045871559633,
+      "grad_norm": 0.7147204279899597,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0213,
+      "step": 701
+    },
+    {
+      "epoch": 6.015412844036697,
+      "grad_norm": 0.048989418894052505,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0022,
+      "step": 702
+    },
+    {
+      "epoch": 6.015779816513762,
+      "grad_norm": 0.15756098926067352,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0065,
+      "step": 703
+    },
+    {
+      "epoch": 6.0161467889908256,
+      "grad_norm": 0.134988933801651,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0058,
+      "step": 704
+    },
+    {
+      "epoch": 6.01651376146789,
+      "grad_norm": 0.12361788749694824,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0054,
+      "step": 705
+    },
+    {
+      "epoch": 6.016880733944954,
+      "grad_norm": 8.780678749084473,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3185,
+      "step": 706
+    },
+    {
+      "epoch": 6.017247706422018,
+      "grad_norm": 0.07378777861595154,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0032,
+      "step": 707
+    },
+    {
+      "epoch": 6.017614678899083,
+      "grad_norm": 0.11033988744020462,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0049,
+      "step": 708
+    },
+    {
+      "epoch": 6.017981651376147,
+      "grad_norm": 11.8186616897583,
+      "learning_rate": 2.5e-05,
+      "loss": 0.8959,
+      "step": 709
+    },
+    {
+      "epoch": 6.018348623853211,
+      "grad_norm": 0.10605289787054062,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0047,
+      "step": 710
+    },
+    {
+      "epoch": 6.018715596330275,
+      "grad_norm": 5.60904598236084,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4189,
+      "step": 711
+    },
+    {
+      "epoch": 6.019082568807339,
+      "grad_norm": 3.181572914123535,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5731,
+      "step": 712
+    },
+    {
+      "epoch": 6.019449541284404,
+      "grad_norm": 0.3805999159812927,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0144,
+      "step": 713
+    },
+    {
+      "epoch": 6.019816513761468,
+      "grad_norm": 1.68703293800354,
+      "learning_rate": 2.5e-05,
+      "loss": 0.036,
+      "step": 714
+    },
+    {
+      "epoch": 6.020183486238532,
+      "grad_norm": 0.18995602428913116,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0067,
+      "step": 715
+    },
+    {
+      "epoch": 6.020550458715596,
+      "grad_norm": 0.3267180323600769,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0116,
+      "step": 716
+    },
+    {
+      "epoch": 6.020917431192661,
+      "grad_norm": 0.21672053635120392,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0091,
+      "step": 717
+    },
+    {
+      "epoch": 6.021284403669725,
+      "grad_norm": 0.33753469586372375,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0162,
+      "step": 718
+    },
+    {
+      "epoch": 6.021651376146789,
+      "grad_norm": 3.211364984512329,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3312,
+      "step": 719
+    },
+    {
+      "epoch": 6.022018348623853,
+      "grad_norm": 0.15013383328914642,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0052,
+      "step": 720
+    },
+    {
+      "epoch": 6.022385321100917,
+      "grad_norm": 2.8594703674316406,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3594,
+      "step": 721
+    },
+    {
+      "epoch": 6.022752293577982,
+      "grad_norm": 0.14795120060443878,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0064,
+      "step": 722
+    },
+    {
+      "epoch": 6.023119266055046,
+      "grad_norm": 0.32280442118644714,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0112,
+      "step": 723
+    },
+    {
+      "epoch": 6.02348623853211,
+      "grad_norm": 5.153697967529297,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5662,
+      "step": 724
+    },
+    {
+      "epoch": 6.0238532110091745,
+      "grad_norm": 0.14898444712162018,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0051,
+      "step": 725
+    },
+    {
+      "epoch": 6.024220183486238,
+      "grad_norm": 0.565102756023407,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0169,
+      "step": 726
+    },
+    {
+      "epoch": 6.024587155963303,
+      "grad_norm": 0.18367063999176025,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0072,
+      "step": 727
+    },
+    {
+      "epoch": 6.024954128440367,
+      "grad_norm": 0.1236872524023056,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0053,
+      "step": 728
+    },
+    {
+      "epoch": 6.025321100917431,
+      "grad_norm": 0.08912146091461182,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0037,
+      "step": 729
+    },
+    {
+      "epoch": 6.0256880733944955,
+      "grad_norm": 2.2519307136535645,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0817,
+      "step": 730
+    },
+    {
+      "epoch": 6.026055045871559,
+      "grad_norm": 3.1285550594329834,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5717,
+      "step": 731
+    },
+    {
+      "epoch": 6.026422018348624,
+      "grad_norm": 4.893245697021484,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1226,
+      "step": 732
+    },
+    {
+      "epoch": 6.026788990825688,
+      "grad_norm": 6.925151824951172,
+      "learning_rate": 2.5e-05,
+      "loss": 0.9939,
+      "step": 733
+    },
+    {
+      "epoch": 6.027155963302753,
+      "grad_norm": 0.1979040801525116,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0084,
+      "step": 734
+    },
+    {
+      "epoch": 6.027522935779817,
+      "grad_norm": 0.3071933686733246,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0116,
+      "step": 735
+    },
+    {
+      "epoch": 6.02788990825688,
+      "grad_norm": 0.3615158200263977,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0139,
+      "step": 736
+    },
+    {
+      "epoch": 6.028256880733945,
+      "grad_norm": 0.07616909593343735,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0035,
+      "step": 737
+    },
+    {
+      "epoch": 6.028623853211009,
+      "grad_norm": 0.23265725374221802,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0111,
+      "step": 738
+    },
+    {
+      "epoch": 6.028990825688074,
+      "grad_norm": 0.7352873086929321,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0312,
+      "step": 739
+    },
+    {
+      "epoch": 6.029357798165138,
+      "grad_norm": 0.17847123742103577,
+      "learning_rate": 2.5e-05,
+      "loss": 0.007,
+      "step": 740
+    },
+    {
+      "epoch": 6.0297247706422015,
+      "grad_norm": 4.23317289352417,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4461,
+      "step": 741
+    },
+    {
+      "epoch": 6.030091743119266,
+      "grad_norm": 0.38569653034210205,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0139,
+      "step": 742
+    },
+    {
+      "epoch": 6.03045871559633,
+      "grad_norm": 0.05369468405842781,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0022,
+      "step": 743
+    },
+    {
+      "epoch": 6.030825688073395,
+      "grad_norm": 6.890101909637451,
+      "learning_rate": 2.5e-05,
+      "loss": 0.8353,
+      "step": 744
+    },
+    {
+      "epoch": 6.031192660550459,
+      "grad_norm": 3.59909725189209,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5714,
+      "step": 745
+    },
+    {
+      "epoch": 6.0315596330275225,
+      "grad_norm": 0.3259468972682953,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0138,
+      "step": 746
+    },
+    {
+      "epoch": 6.031926605504587,
+      "grad_norm": 0.20248031616210938,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0086,
+      "step": 747
+    },
+    {
+      "epoch": 6.032293577981651,
+      "grad_norm": 0.31660619378089905,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0137,
+      "step": 748
+    },
+    {
+      "epoch": 6.032660550458716,
+      "grad_norm": 2.15812349319458,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0519,
+      "step": 749
+    },
+    {
+      "epoch": 6.03302752293578,
+      "grad_norm": 0.20733949542045593,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0087,
+      "step": 750
+    },
+    {
+      "epoch": 6.0333944954128444,
+      "grad_norm": 0.14728552103042603,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0063,
+      "step": 751
+    },
+    {
+      "epoch": 6.033761467889908,
+      "grad_norm": 9.244287490844727,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3503,
+      "step": 752
+    },
+    {
+      "epoch": 6.034128440366972,
+      "grad_norm": 4.256886005401611,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0575,
+      "step": 753
+    },
+    {
+      "epoch": 6.034495412844037,
+      "grad_norm": 0.11548680812120438,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0049,
+      "step": 754
+    },
+    {
+      "epoch": 6.034862385321101,
+      "grad_norm": 0.11757824569940567,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0051,
+      "step": 755
+    },
+    {
+      "epoch": 6.0352293577981655,
+      "grad_norm": 0.15447628498077393,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0062,
+      "step": 756
+    },
+    {
+      "epoch": 6.035596330275229,
+      "grad_norm": 0.18529382348060608,
+      "learning_rate": 2.5e-05,
+      "loss": 0.008,
+      "step": 757
+    },
+    {
+      "epoch": 6.035963302752293,
+      "grad_norm": 8.31997299194336,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4633,
+      "step": 758
+    },
+    {
+      "epoch": 6.036330275229358,
+      "grad_norm": 0.13505524396896362,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0058,
+      "step": 759
+    },
+    {
+      "epoch": 6.036697247706422,
+      "grad_norm": 0.09383295476436615,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0038,
+      "step": 760
+    },
+    {
+      "epoch": 6.0370642201834865,
+      "grad_norm": 0.10302843898534775,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0046,
+      "step": 761
+    },
+    {
+      "epoch": 6.03743119266055,
+      "grad_norm": 0.1828688234090805,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0074,
+      "step": 762
+    },
+    {
+      "epoch": 6.037798165137614,
+      "grad_norm": 0.29041588306427,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0114,
+      "step": 763
+    },
+    {
+      "epoch": 6.038165137614679,
+      "grad_norm": 0.6149030327796936,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0196,
+      "step": 764
+    },
+    {
+      "epoch": 6.038532110091743,
+      "grad_norm": 0.7241535782814026,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0136,
+      "step": 765
+    },
+    {
+      "epoch": 6.038899082568808,
+      "grad_norm": 0.19188551604747772,
+      "learning_rate": 2.5e-05,
+      "loss": 0.008,
+      "step": 766
+    },
+    {
+      "epoch": 6.039266055045871,
+      "grad_norm": 2.9104697704315186,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3503,
+      "step": 767
+    },
+    {
+      "epoch": 6.039633027522936,
+      "grad_norm": 0.10823194682598114,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0048,
+      "step": 768
+    },
+    {
+      "epoch": 6.04,
+      "grad_norm": 0.17793519794940948,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0071,
+      "step": 769
+    },
+    {
+      "epoch": 6.040366972477064,
+      "grad_norm": 0.07905784994363785,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0033,
+      "step": 770
+    },
+    {
+      "epoch": 6.040366972477064,
+      "eval_accuracy": 0.1952662721893491,
+      "eval_loss": 4.460123538970947,
+      "eval_runtime": 25.7503,
+      "eval_samples_per_second": 13.126,
+      "eval_steps_per_second": 1.67,
+      "step": 770
+    },
+    {
+      "epoch": 7.000366972477064,
+      "grad_norm": 0.11132709681987762,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0049,
+      "step": 771
+    },
+    {
+      "epoch": 7.000733944954129,
+      "grad_norm": 0.23977777361869812,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0092,
+      "step": 772
+    },
+    {
+      "epoch": 7.0011009174311924,
+      "grad_norm": 0.07466649264097214,
+      "learning_rate": 2.5e-05,
+      "loss": 0.003,
+      "step": 773
+    },
+    {
+      "epoch": 7.001467889908257,
+      "grad_norm": 0.1077701672911644,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0043,
+      "step": 774
+    },
+    {
+      "epoch": 7.001834862385321,
+      "grad_norm": 0.5482657551765442,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0205,
+      "step": 775
+    },
+    {
+      "epoch": 7.002201834862385,
+      "grad_norm": 0.5116673111915588,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0068,
+      "step": 776
+    },
+    {
+      "epoch": 7.00256880733945,
+      "grad_norm": 3.067056655883789,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0525,
+      "step": 777
+    },
+    {
+      "epoch": 7.0029357798165135,
+      "grad_norm": 0.05802005156874657,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0026,
+      "step": 778
+    },
+    {
+      "epoch": 7.003302752293578,
+      "grad_norm": 0.05125277489423752,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0021,
+      "step": 779
+    },
+    {
+      "epoch": 7.003669724770642,
+      "grad_norm": 0.05854889005422592,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0026,
+      "step": 780
+    },
+    {
+      "epoch": 7.004036697247707,
+      "grad_norm": 0.06536930799484253,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0029,
+      "step": 781
+    },
+    {
+      "epoch": 7.004403669724771,
+      "grad_norm": 9.508818626403809,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3109,
+      "step": 782
+    },
+    {
+      "epoch": 7.0047706422018345,
+      "grad_norm": 3.143989324569702,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6645,
+      "step": 783
+    },
+    {
+      "epoch": 7.005137614678899,
+      "grad_norm": 0.18746322393417358,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0073,
+      "step": 784
+    },
+    {
+      "epoch": 7.005504587155963,
+      "grad_norm": 0.06501258164644241,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0029,
+      "step": 785
+    },
+    {
+      "epoch": 7.005871559633028,
+      "grad_norm": 0.04979090020060539,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0022,
+      "step": 786
+    },
+    {
+      "epoch": 7.006238532110092,
+      "grad_norm": 0.07349596917629242,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0032,
+      "step": 787
+    },
+    {
+      "epoch": 7.006605504587156,
+      "grad_norm": 0.09369394183158875,
+      "learning_rate": 2.5e-05,
+      "loss": 0.004,
+      "step": 788
+    },
+    {
+      "epoch": 7.00697247706422,
+      "grad_norm": 0.06985606253147125,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0031,
+      "step": 789
+    },
+    {
+      "epoch": 7.007339449541284,
+      "grad_norm": 0.0891771912574768,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0037,
+      "step": 790
+    },
+    {
+      "epoch": 7.007706422018349,
+      "grad_norm": 4.187360763549805,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5207,
+      "step": 791
+    },
+    {
+      "epoch": 7.008073394495413,
+      "grad_norm": 0.040006790310144424,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0018,
+      "step": 792
+    },
+    {
+      "epoch": 7.008440366972477,
+      "grad_norm": 0.05395551398396492,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0024,
+      "step": 793
+    },
+    {
+      "epoch": 7.008807339449541,
+      "grad_norm": 0.061198245733976364,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0027,
+      "step": 794
+    },
+    {
+      "epoch": 7.009174311926605,
+      "grad_norm": 0.05776246637105942,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0026,
+      "step": 795
+    },
+    {
+      "epoch": 7.00954128440367,
+      "grad_norm": 0.027192378416657448,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0011,
+      "step": 796
+    },
+    {
+      "epoch": 7.009908256880734,
+      "grad_norm": 4.1108784675598145,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5843,
+      "step": 797
+    },
+    {
+      "epoch": 7.010275229357799,
+      "grad_norm": 0.08927580714225769,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0039,
+      "step": 798
+    },
+    {
+      "epoch": 7.010642201834862,
+      "grad_norm": 3.982842445373535,
+      "learning_rate": 2.5e-05,
+      "loss": 0.548,
+      "step": 799
+    },
+    {
+      "epoch": 7.011009174311926,
+      "grad_norm": 0.036366917192935944,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0012,
+      "step": 800
+    },
+    {
+      "epoch": 7.011376146788991,
+      "grad_norm": 8.185258865356445,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4844,
+      "step": 801
+    },
+    {
+      "epoch": 7.011743119266055,
+      "grad_norm": 6.049882411956787,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3414,
+      "step": 802
+    },
+    {
+      "epoch": 7.01211009174312,
+      "grad_norm": 0.21621595323085785,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0063,
+      "step": 803
+    },
+    {
+      "epoch": 7.0124770642201835,
+      "grad_norm": 21.75786781311035,
+      "learning_rate": 2.5e-05,
+      "loss": 0.2048,
+      "step": 804
+    },
+    {
+      "epoch": 7.012844036697247,
+      "grad_norm": 0.05727781355381012,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0023,
+      "step": 805
+    },
+    {
+      "epoch": 7.013211009174312,
+      "grad_norm": 0.12199017405509949,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0048,
+      "step": 806
+    },
+    {
+      "epoch": 7.013577981651376,
+      "grad_norm": 0.41428858041763306,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0106,
+      "step": 807
+    },
+    {
+      "epoch": 7.013944954128441,
+      "grad_norm": 6.190532207489014,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3096,
+      "step": 808
+    },
+    {
+      "epoch": 7.0143119266055045,
+      "grad_norm": 6.646399974822998,
+      "learning_rate": 2.5e-05,
+      "loss": 0.9978,
+      "step": 809
+    },
+    {
+      "epoch": 7.014678899082568,
+      "grad_norm": 14.746865272521973,
+      "learning_rate": 2.5e-05,
+      "loss": 0.2103,
+      "step": 810
+    },
+    {
+      "epoch": 7.015045871559633,
+      "grad_norm": 0.6672438979148865,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0158,
+      "step": 811
+    },
+    {
+      "epoch": 7.015412844036697,
+      "grad_norm": 0.1324150711297989,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0052,
+      "step": 812
+    },
+    {
+      "epoch": 7.015779816513762,
+      "grad_norm": 0.23772357404232025,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0085,
+      "step": 813
+    },
+    {
+      "epoch": 7.0161467889908256,
+      "grad_norm": 0.13693159818649292,
+      "learning_rate": 2.5e-05,
+      "loss": 0.005,
+      "step": 814
+    },
+    {
+      "epoch": 7.01651376146789,
+      "grad_norm": 0.15698811411857605,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0062,
+      "step": 815
+    },
+    {
+      "epoch": 7.016880733944954,
+      "grad_norm": 0.4931865632534027,
+      "learning_rate": 2.5e-05,
+      "loss": 0.019,
+      "step": 816
+    },
+    {
+      "epoch": 7.017247706422018,
+      "grad_norm": 0.16839449107646942,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0072,
+      "step": 817
+    },
+    {
+      "epoch": 7.017614678899083,
+      "grad_norm": 7.954697132110596,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3549,
+      "step": 818
+    },
+    {
+      "epoch": 7.017981651376147,
+      "grad_norm": 2.7547831535339355,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5665,
+      "step": 819
+    },
+    {
+      "epoch": 7.018348623853211,
+      "grad_norm": 0.11733569204807281,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0046,
+      "step": 820
+    },
+    {
+      "epoch": 7.018715596330275,
+      "grad_norm": 0.09013906121253967,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0033,
+      "step": 821
+    },
+    {
+      "epoch": 7.019082568807339,
+      "grad_norm": 11.25192928314209,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0589,
+      "step": 822
+    },
+    {
+      "epoch": 7.019449541284404,
+      "grad_norm": 0.57774418592453,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0133,
+      "step": 823
+    },
+    {
+      "epoch": 7.019816513761468,
+      "grad_norm": 0.2544844448566437,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0093,
+      "step": 824
+    },
+    {
+      "epoch": 7.020183486238532,
+      "grad_norm": 4.944797039031982,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0954,
+      "step": 825
+    },
+    {
+      "epoch": 7.020550458715596,
+      "grad_norm": 0.4089986979961395,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0152,
+      "step": 826
+    },
+    {
+      "epoch": 7.020917431192661,
+      "grad_norm": 6.0038909912109375,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5809,
+      "step": 827
+    },
+    {
+      "epoch": 7.021284403669725,
+      "grad_norm": 0.12247730046510696,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0049,
+      "step": 828
+    },
+    {
+      "epoch": 7.021651376146789,
+      "grad_norm": 0.8706518411636353,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0174,
+      "step": 829
+    },
+    {
+      "epoch": 7.022018348623853,
+      "grad_norm": 0.1190580427646637,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0047,
+      "step": 830
+    },
+    {
+      "epoch": 7.022385321100917,
+      "grad_norm": 0.08350613713264465,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0034,
+      "step": 831
+    },
+    {
+      "epoch": 7.022752293577982,
+      "grad_norm": 0.05365229398012161,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0022,
+      "step": 832
+    },
+    {
+      "epoch": 7.023119266055046,
+      "grad_norm": 0.3614063858985901,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0167,
+      "step": 833
+    },
+    {
+      "epoch": 7.02348623853211,
+      "grad_norm": 0.023435207083821297,
+      "learning_rate": 2.5e-05,
+      "loss": 0.001,
+      "step": 834
+    },
+    {
+      "epoch": 7.0238532110091745,
+      "grad_norm": 3.230663776397705,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5613,
+      "step": 835
+    },
+    {
+      "epoch": 7.024220183486238,
+      "grad_norm": 0.09338148683309555,
+      "learning_rate": 2.5e-05,
+      "loss": 0.004,
+      "step": 836
+    },
+    {
+      "epoch": 7.024587155963303,
+      "grad_norm": 0.08192034065723419,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0035,
+      "step": 837
+    },
+    {
+      "epoch": 7.024954128440367,
+      "grad_norm": 6.965814590454102,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3126,
+      "step": 838
+    },
+    {
+      "epoch": 7.025321100917431,
+      "grad_norm": 0.06162590906023979,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0025,
+      "step": 839
+    },
+    {
+      "epoch": 7.0256880733944955,
+      "grad_norm": 0.05617407336831093,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0023,
+      "step": 840
+    },
+    {
+      "epoch": 7.026055045871559,
+      "grad_norm": 0.09172047674655914,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0038,
+      "step": 841
+    },
+    {
+      "epoch": 7.026422018348624,
+      "grad_norm": 6.57578182220459,
+      "learning_rate": 2.5e-05,
+      "loss": 0.2358,
+      "step": 842
+    },
+    {
+      "epoch": 7.026788990825688,
+      "grad_norm": 0.1204962283372879,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0047,
+      "step": 843
+    },
+    {
+      "epoch": 7.027155963302753,
+      "grad_norm": 0.4741459786891937,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0107,
+      "step": 844
+    },
+    {
+      "epoch": 7.027522935779817,
+      "grad_norm": 0.305530309677124,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0124,
+      "step": 845
+    },
+    {
+      "epoch": 7.02788990825688,
+      "grad_norm": 0.11488713324069977,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0049,
+      "step": 846
+    },
+    {
+      "epoch": 7.028256880733945,
+      "grad_norm": 0.10418660193681717,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0044,
+      "step": 847
+    },
+    {
+      "epoch": 7.028623853211009,
+      "grad_norm": 1.4393413066864014,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0439,
+      "step": 848
+    },
+    {
+      "epoch": 7.028990825688074,
+      "grad_norm": 8.174779891967773,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3526,
+      "step": 849
+    },
+    {
+      "epoch": 7.029357798165138,
+      "grad_norm": 0.12033990025520325,
+      "learning_rate": 2.5e-05,
+      "loss": 0.005,
+      "step": 850
+    },
+    {
+      "epoch": 7.0297247706422015,
+      "grad_norm": 0.03736472502350807,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0015,
+      "step": 851
+    },
+    {
+      "epoch": 7.030091743119266,
+      "grad_norm": 0.06682645529508591,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0028,
+      "step": 852
+    },
+    {
+      "epoch": 7.03045871559633,
+      "grad_norm": 0.08293373137712479,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0036,
+      "step": 853
+    },
+    {
+      "epoch": 7.030825688073395,
+      "grad_norm": 1.926229476928711,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0644,
+      "step": 854
+    },
+    {
+      "epoch": 7.031192660550459,
+      "grad_norm": 0.9952171444892883,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0166,
+      "step": 855
+    },
+    {
+      "epoch": 7.0315596330275225,
+      "grad_norm": 1.4784842729568481,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0195,
+      "step": 856
+    },
+    {
+      "epoch": 7.031926605504587,
+      "grad_norm": 0.06602731347084045,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0029,
+      "step": 857
+    },
+    {
+      "epoch": 7.032293577981651,
+      "grad_norm": 0.018716471269726753,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0007,
+      "step": 858
+    },
+    {
+      "epoch": 7.032660550458716,
+      "grad_norm": 3.7852895259857178,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6403,
+      "step": 859
+    },
+    {
+      "epoch": 7.03302752293578,
+      "grad_norm": 8.50316333770752,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5141,
+      "step": 860
+    },
+    {
+      "epoch": 7.0333944954128444,
+      "grad_norm": 0.07294552773237228,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0031,
+      "step": 861
+    },
+    {
+      "epoch": 7.033761467889908,
+      "grad_norm": 0.06571441143751144,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0028,
+      "step": 862
+    },
+    {
+      "epoch": 7.034128440366972,
+      "grad_norm": 0.34902864694595337,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0133,
+      "step": 863
+    },
+    {
+      "epoch": 7.034495412844037,
+      "grad_norm": 0.056518327444791794,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0025,
+      "step": 864
+    },
+    {
+      "epoch": 7.034862385321101,
+      "grad_norm": 0.042712610214948654,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0017,
+      "step": 865
+    },
+    {
+      "epoch": 7.0352293577981655,
+      "grad_norm": 0.09082131087779999,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0039,
+      "step": 866
+    },
+    {
+      "epoch": 7.035596330275229,
+      "grad_norm": 0.14603590965270996,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0054,
+      "step": 867
+    },
+    {
+      "epoch": 7.035963302752293,
+      "grad_norm": 0.07532819360494614,
+      "learning_rate": 2.5e-05,
+      "loss": 0.003,
+      "step": 868
+    },
+    {
+      "epoch": 7.036330275229358,
+      "grad_norm": 0.046275947242975235,
+      "learning_rate": 2.5e-05,
+      "loss": 0.002,
+      "step": 869
+    },
+    {
+      "epoch": 7.036697247706422,
+      "grad_norm": 3.595900058746338,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6139,
+      "step": 870
+    },
+    {
+      "epoch": 7.0370642201834865,
+      "grad_norm": 0.10607864707708359,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0046,
+      "step": 871
+    },
+    {
+      "epoch": 7.03743119266055,
+      "grad_norm": 0.06544827669858932,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0026,
+      "step": 872
+    },
+    {
+      "epoch": 7.037798165137614,
+      "grad_norm": 0.5324288606643677,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0089,
+      "step": 873
+    },
+    {
+      "epoch": 7.038165137614679,
+      "grad_norm": 0.22051775455474854,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0063,
+      "step": 874
+    },
+    {
+      "epoch": 7.038532110091743,
+      "grad_norm": 3.7011406421661377,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6703,
+      "step": 875
+    },
+    {
+      "epoch": 7.038899082568808,
+      "grad_norm": 0.07121708244085312,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0029,
+      "step": 876
+    },
+    {
+      "epoch": 7.039266055045871,
+      "grad_norm": 3.431173324584961,
+      "learning_rate": 2.5e-05,
+      "loss": 0.2823,
+      "step": 877
+    },
+    {
+      "epoch": 7.039633027522936,
+      "grad_norm": 0.06873047351837158,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0036,
+      "step": 878
+    },
+    {
+      "epoch": 7.04,
+      "grad_norm": 0.12582124769687653,
+      "learning_rate": 2.5e-05,
+      "loss": 0.006,
+      "step": 879
+    },
+    {
+      "epoch": 7.040366972477064,
+      "grad_norm": 0.19980674982070923,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0071,
+      "step": 880
+    },
+    {
+      "epoch": 7.040366972477064,
+      "eval_accuracy": 0.09171597633136094,
+      "eval_loss": 4.60448694229126,
+      "eval_runtime": 25.7908,
+      "eval_samples_per_second": 13.105,
+      "eval_steps_per_second": 1.667,
+      "step": 880
+    },
+    {
+      "epoch": 8.000366972477064,
+      "grad_norm": 1.3397115468978882,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0272,
+      "step": 881
+    },
+    {
+      "epoch": 8.000733944954128,
+      "grad_norm": 0.2565714716911316,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0055,
+      "step": 882
+    },
+    {
+      "epoch": 8.001100917431193,
+      "grad_norm": 0.3067053258419037,
+      "learning_rate": 2.5e-05,
+      "loss": 0.011,
+      "step": 883
+    },
+    {
+      "epoch": 8.001467889908257,
+      "grad_norm": 0.22656728327274323,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0079,
+      "step": 884
+    },
+    {
+      "epoch": 8.001834862385321,
+      "grad_norm": 0.118247389793396,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0052,
+      "step": 885
+    },
+    {
+      "epoch": 8.002201834862385,
+      "grad_norm": 0.08141342550516129,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0034,
+      "step": 886
+    },
+    {
+      "epoch": 8.002568807339449,
+      "grad_norm": 17.343582153320312,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1088,
+      "step": 887
+    },
+    {
+      "epoch": 8.002935779816514,
+      "grad_norm": 0.0730825886130333,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0028,
+      "step": 888
+    },
+    {
+      "epoch": 8.003302752293578,
+      "grad_norm": 0.08300863206386566,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0036,
+      "step": 889
+    },
+    {
+      "epoch": 8.003669724770642,
+      "grad_norm": 0.11370746791362762,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0047,
+      "step": 890
+    },
+    {
+      "epoch": 8.004036697247706,
+      "grad_norm": 0.06766372174024582,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0028,
+      "step": 891
+    },
+    {
+      "epoch": 8.00440366972477,
+      "grad_norm": 4.366567611694336,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4738,
+      "step": 892
+    },
+    {
+      "epoch": 8.004770642201835,
+      "grad_norm": 0.04274710267782211,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0018,
+      "step": 893
+    },
+    {
+      "epoch": 8.0051376146789,
+      "grad_norm": 0.11026450246572495,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0049,
+      "step": 894
+    },
+    {
+      "epoch": 8.005504587155963,
+      "grad_norm": 0.08949928730726242,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0039,
+      "step": 895
+    },
+    {
+      "epoch": 8.005871559633027,
+      "grad_norm": 5.311772346496582,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5638,
+      "step": 896
+    },
+    {
+      "epoch": 8.006238532110093,
+      "grad_norm": 7.326998233795166,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5558,
+      "step": 897
+    },
+    {
+      "epoch": 8.006605504587156,
+      "grad_norm": 0.13751623034477234,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0054,
+      "step": 898
+    },
+    {
+      "epoch": 8.00697247706422,
+      "grad_norm": 9.853394508361816,
+      "learning_rate": 2.5e-05,
+      "loss": 0.9549,
+      "step": 899
+    },
+    {
+      "epoch": 8.007339449541284,
+      "grad_norm": 7.673292636871338,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4846,
+      "step": 900
+    },
+    {
+      "epoch": 8.007706422018348,
+      "grad_norm": 0.17064766585826874,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0066,
+      "step": 901
+    },
+    {
+      "epoch": 8.008073394495414,
+      "grad_norm": 0.04655308276414871,
+      "learning_rate": 2.5e-05,
+      "loss": 0.002,
+      "step": 902
+    },
+    {
+      "epoch": 8.008440366972478,
+      "grad_norm": 8.002827644348145,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0653,
+      "step": 903
+    },
+    {
+      "epoch": 8.008807339449541,
+      "grad_norm": 0.16547171771526337,
+      "learning_rate": 2.5e-05,
+      "loss": 0.006,
+      "step": 904
+    },
+    {
+      "epoch": 8.009174311926605,
+      "grad_norm": 0.2623595893383026,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0099,
+      "step": 905
+    },
+    {
+      "epoch": 8.009541284403669,
+      "grad_norm": 0.2325545996427536,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0087,
+      "step": 906
+    },
+    {
+      "epoch": 8.009908256880735,
+      "grad_norm": 0.28599950671195984,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0087,
+      "step": 907
+    },
+    {
+      "epoch": 8.010275229357799,
+      "grad_norm": 3.8705835342407227,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3834,
+      "step": 908
+    },
+    {
+      "epoch": 8.010642201834862,
+      "grad_norm": 0.11399892717599869,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0043,
+      "step": 909
+    },
+    {
+      "epoch": 8.011009174311926,
+      "grad_norm": 0.6602398753166199,
+      "learning_rate": 2.5e-05,
+      "loss": 0.021,
+      "step": 910
+    },
+    {
+      "epoch": 8.01137614678899,
+      "grad_norm": 4.2441864013671875,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0396,
+      "step": 911
+    },
+    {
+      "epoch": 8.011743119266056,
+      "grad_norm": 7.841331481933594,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0616,
+      "step": 912
+    },
+    {
+      "epoch": 8.01211009174312,
+      "grad_norm": 0.2644534409046173,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0108,
+      "step": 913
+    },
+    {
+      "epoch": 8.012477064220183,
+      "grad_norm": 0.13179442286491394,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0056,
+      "step": 914
+    },
+    {
+      "epoch": 8.012844036697247,
+      "grad_norm": 0.2211037576198578,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0055,
+      "step": 915
+    },
+    {
+      "epoch": 8.013211009174311,
+      "grad_norm": 0.36059483885765076,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0143,
+      "step": 916
+    },
+    {
+      "epoch": 8.013577981651377,
+      "grad_norm": 0.07135040313005447,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0028,
+      "step": 917
+    },
+    {
+      "epoch": 8.01394495412844,
+      "grad_norm": 0.07437378913164139,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0031,
+      "step": 918
+    },
+    {
+      "epoch": 8.014311926605505,
+      "grad_norm": 0.12937739491462708,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0048,
+      "step": 919
+    },
+    {
+      "epoch": 8.014678899082568,
+      "grad_norm": 0.06491874903440475,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0027,
+      "step": 920
+    },
+    {
+      "epoch": 8.015045871559632,
+      "grad_norm": 0.040153179317712784,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0017,
+      "step": 921
+    },
+    {
+      "epoch": 8.015412844036698,
+      "grad_norm": 0.24626868963241577,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0091,
+      "step": 922
+    },
+    {
+      "epoch": 8.015779816513762,
+      "grad_norm": 2.8731839656829834,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0221,
+      "step": 923
+    },
+    {
+      "epoch": 8.016146788990826,
+      "grad_norm": 0.025319838896393776,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0011,
+      "step": 924
+    },
+    {
+      "epoch": 8.01651376146789,
+      "grad_norm": 0.03755786269903183,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0017,
+      "step": 925
+    },
+    {
+      "epoch": 8.016880733944953,
+      "grad_norm": 0.04529043659567833,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0019,
+      "step": 926
+    },
+    {
+      "epoch": 8.017247706422019,
+      "grad_norm": 0.06085320562124252,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0026,
+      "step": 927
+    },
+    {
+      "epoch": 8.017614678899083,
+      "grad_norm": 0.060384154319763184,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0026,
+      "step": 928
+    },
+    {
+      "epoch": 8.017981651376147,
+      "grad_norm": 0.030146710574626923,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0012,
+      "step": 929
+    },
+    {
+      "epoch": 8.01834862385321,
+      "grad_norm": 0.018250251188874245,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0007,
+      "step": 930
+    },
+    {
+      "epoch": 8.018715596330276,
+      "grad_norm": 11.49174976348877,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3104,
+      "step": 931
+    },
+    {
+      "epoch": 8.01908256880734,
+      "grad_norm": 1.3130452632904053,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0406,
+      "step": 932
+    },
+    {
+      "epoch": 8.019449541284404,
+      "grad_norm": 0.05622019246220589,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0023,
+      "step": 933
+    },
+    {
+      "epoch": 8.019816513761468,
+      "grad_norm": 0.04962543025612831,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0021,
+      "step": 934
+    },
+    {
+      "epoch": 8.020183486238532,
+      "grad_norm": 0.04291941225528717,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0018,
+      "step": 935
+    },
+    {
+      "epoch": 8.020550458715597,
+      "grad_norm": 20.716293334960938,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6135,
+      "step": 936
+    },
+    {
+      "epoch": 8.020917431192661,
+      "grad_norm": 8.805044174194336,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4054,
+      "step": 937
+    },
+    {
+      "epoch": 8.021284403669725,
+      "grad_norm": 5.732924461364746,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0516,
+      "step": 938
+    },
+    {
+      "epoch": 8.021651376146789,
+      "grad_norm": 0.05633801966905594,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0024,
+      "step": 939
+    },
+    {
+      "epoch": 8.022018348623853,
+      "grad_norm": 0.1017167717218399,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0034,
+      "step": 940
+    },
+    {
+      "epoch": 8.022385321100918,
+      "grad_norm": 0.6590113639831543,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0197,
+      "step": 941
+    },
+    {
+      "epoch": 8.022752293577982,
+      "grad_norm": 0.5520896911621094,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0066,
+      "step": 942
+    },
+    {
+      "epoch": 8.023119266055046,
+      "grad_norm": 0.1482970416545868,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0041,
+      "step": 943
+    },
+    {
+      "epoch": 8.02348623853211,
+      "grad_norm": 0.04678082838654518,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0019,
+      "step": 944
+    },
+    {
+      "epoch": 8.023853211009174,
+      "grad_norm": 0.030649982392787933,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0013,
+      "step": 945
+    },
+    {
+      "epoch": 8.02422018348624,
+      "grad_norm": 0.05503760278224945,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0023,
+      "step": 946
+    },
+    {
+      "epoch": 8.024587155963303,
+      "grad_norm": 0.07822321355342865,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0023,
+      "step": 947
+    },
+    {
+      "epoch": 8.024954128440367,
+      "grad_norm": 12.399303436279297,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4956,
+      "step": 948
+    },
+    {
+      "epoch": 8.02532110091743,
+      "grad_norm": 0.11732669174671173,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0023,
+      "step": 949
+    },
+    {
+      "epoch": 8.025688073394495,
+      "grad_norm": 4.517271041870117,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6111,
+      "step": 950
+    },
+    {
+      "epoch": 8.02605504587156,
+      "grad_norm": 4.188869953155518,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3012,
+      "step": 951
+    },
+    {
+      "epoch": 8.026422018348624,
+      "grad_norm": 0.26586246490478516,
+      "learning_rate": 2.5e-05,
+      "loss": 0.005,
+      "step": 952
+    },
+    {
+      "epoch": 8.026788990825688,
+      "grad_norm": 4.440396785736084,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6191,
+      "step": 953
+    },
+    {
+      "epoch": 8.027155963302752,
+      "grad_norm": 4.659714698791504,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4867,
+      "step": 954
+    },
+    {
+      "epoch": 8.027522935779816,
+      "grad_norm": 8.86314582824707,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4874,
+      "step": 955
+    },
+    {
+      "epoch": 8.027889908256881,
+      "grad_norm": 0.19289849698543549,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0044,
+      "step": 956
+    },
+    {
+      "epoch": 8.028256880733945,
+      "grad_norm": 3.503230333328247,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0488,
+      "step": 957
+    },
+    {
+      "epoch": 8.028623853211009,
+      "grad_norm": 0.043053142726421356,
+      "learning_rate": 2.5e-05,
+      "loss": 0.002,
+      "step": 958
+    },
+    {
+      "epoch": 8.028990825688073,
+      "grad_norm": 0.041521914303302765,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0017,
+      "step": 959
+    },
+    {
+      "epoch": 8.029357798165137,
+      "grad_norm": 5.427301406860352,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5184,
+      "step": 960
+    },
+    {
+      "epoch": 8.029724770642202,
+      "grad_norm": 0.14201723039150238,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0053,
+      "step": 961
+    },
+    {
+      "epoch": 8.030091743119266,
+      "grad_norm": 4.047647953033447,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1712,
+      "step": 962
+    },
+    {
+      "epoch": 8.03045871559633,
+      "grad_norm": 0.33958500623703003,
+      "learning_rate": 2.5e-05,
+      "loss": 0.012,
+      "step": 963
+    },
+    {
+      "epoch": 8.030825688073394,
+      "grad_norm": 0.16152925789356232,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0054,
+      "step": 964
+    },
+    {
+      "epoch": 8.03119266055046,
+      "grad_norm": 0.09419584274291992,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0011,
+      "step": 965
+    },
+    {
+      "epoch": 8.031559633027523,
+      "grad_norm": 0.09287504106760025,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0032,
+      "step": 966
+    },
+    {
+      "epoch": 8.031926605504587,
+      "grad_norm": 0.039954692125320435,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0015,
+      "step": 967
+    },
+    {
+      "epoch": 8.032293577981651,
+      "grad_norm": 10.889971733093262,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0753,
+      "step": 968
+    },
+    {
+      "epoch": 8.032660550458715,
+      "grad_norm": 0.16728755831718445,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0063,
+      "step": 969
+    },
+    {
+      "epoch": 8.03302752293578,
+      "grad_norm": 0.2292383462190628,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0088,
+      "step": 970
+    },
+    {
+      "epoch": 8.033394495412844,
+      "grad_norm": 10.985915184020996,
+      "learning_rate": 2.5e-05,
+      "loss": 0.8021,
+      "step": 971
+    },
+    {
+      "epoch": 8.033761467889908,
+      "grad_norm": 0.1991981565952301,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0069,
+      "step": 972
+    },
+    {
+      "epoch": 8.034128440366972,
+      "grad_norm": 0.32555434107780457,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0079,
+      "step": 973
+    },
+    {
+      "epoch": 8.034495412844036,
+      "grad_norm": 15.41511058807373,
+      "learning_rate": 2.5e-05,
+      "loss": 0.181,
+      "step": 974
+    },
+    {
+      "epoch": 8.034862385321102,
+      "grad_norm": 0.05417482927441597,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0021,
+      "step": 975
+    },
+    {
+      "epoch": 8.035229357798165,
+      "grad_norm": 0.19286848604679108,
+      "learning_rate": 2.5e-05,
+      "loss": 0.007,
+      "step": 976
+    },
+    {
+      "epoch": 8.03559633027523,
+      "grad_norm": 3.3367908000946045,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0584,
+      "step": 977
+    },
+    {
+      "epoch": 8.035963302752293,
+      "grad_norm": 0.14648135006427765,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0052,
+      "step": 978
+    },
+    {
+      "epoch": 8.036330275229357,
+      "grad_norm": 5.295133113861084,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4871,
+      "step": 979
+    },
+    {
+      "epoch": 8.036697247706423,
+      "grad_norm": 0.10857419669628143,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0047,
+      "step": 980
+    },
+    {
+      "epoch": 8.037064220183487,
+      "grad_norm": 0.032847270369529724,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0014,
+      "step": 981
+    },
+    {
+      "epoch": 8.03743119266055,
+      "grad_norm": 0.17431005835533142,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0069,
+      "step": 982
+    },
+    {
+      "epoch": 8.037798165137614,
+      "grad_norm": 0.1771594136953354,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0055,
+      "step": 983
+    },
+    {
+      "epoch": 8.038165137614678,
+      "grad_norm": 0.15307791531085968,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0056,
+      "step": 984
+    },
+    {
+      "epoch": 8.038532110091744,
+      "grad_norm": 1.4514143466949463,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0155,
+      "step": 985
+    },
+    {
+      "epoch": 8.038899082568808,
+      "grad_norm": 0.05131838843226433,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0021,
+      "step": 986
+    },
+    {
+      "epoch": 8.039266055045871,
+      "grad_norm": 0.11886367201805115,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0049,
+      "step": 987
+    },
+    {
+      "epoch": 8.039633027522935,
+      "grad_norm": 0.1211148127913475,
+      "learning_rate": 2.5e-05,
+      "loss": 0.004,
+      "step": 988
+    },
+    {
+      "epoch": 8.04,
+      "grad_norm": 4.949779033660889,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5278,
+      "step": 989
+    },
+    {
+      "epoch": 8.040366972477065,
+      "grad_norm": 0.10737671703100204,
+      "learning_rate": 2.5e-05,
+      "loss": 0.004,
+      "step": 990
+    },
+    {
+      "epoch": 8.040366972477065,
+      "eval_accuracy": 0.40828402366863903,
+      "eval_loss": 3.4061784744262695,
+      "eval_runtime": 25.9914,
+      "eval_samples_per_second": 13.004,
+      "eval_steps_per_second": 1.654,
+      "step": 990
+    },
+    {
+      "epoch": 9.000366972477064,
+      "grad_norm": 5.277964115142822,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4738,
+      "step": 991
+    },
+    {
+      "epoch": 9.000733944954128,
+      "grad_norm": 21.515140533447266,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1582,
+      "step": 992
+    },
+    {
+      "epoch": 9.001100917431193,
+      "grad_norm": 0.09683515876531601,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0036,
+      "step": 993
+    },
+    {
+      "epoch": 9.001467889908257,
+      "grad_norm": 0.05067368224263191,
+      "learning_rate": 2.5e-05,
+      "loss": 0.002,
+      "step": 994
+    },
+    {
+      "epoch": 9.001834862385321,
+      "grad_norm": 12.914159774780273,
+      "learning_rate": 2.5e-05,
+      "loss": 0.8219,
+      "step": 995
+    },
+    {
+      "epoch": 9.002201834862385,
+      "grad_norm": 0.46718406677246094,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0079,
+      "step": 996
+    },
+    {
+      "epoch": 9.002568807339449,
+      "grad_norm": 0.5645943880081177,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0142,
+      "step": 997
+    },
+    {
+      "epoch": 9.002935779816514,
+      "grad_norm": 3.393951416015625,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5289,
+      "step": 998
+    },
+    {
+      "epoch": 9.003302752293578,
+      "grad_norm": 0.9267740249633789,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0238,
+      "step": 999
+    },
+    {
+      "epoch": 9.003669724770642,
+      "grad_norm": 0.11789031326770782,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0047,
+      "step": 1000
+    },
+    {
+      "epoch": 9.004036697247706,
+      "grad_norm": 0.06680285930633545,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0019,
+      "step": 1001
+    },
+    {
+      "epoch": 9.00440366972477,
+      "grad_norm": 0.6144696474075317,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0128,
+      "step": 1002
+    },
+    {
+      "epoch": 9.004770642201835,
+      "grad_norm": 3.328437328338623,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0534,
+      "step": 1003
+    },
+    {
+      "epoch": 9.0051376146789,
+      "grad_norm": 19.235515594482422,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1217,
+      "step": 1004
+    },
+    {
+      "epoch": 9.005504587155963,
+      "grad_norm": 0.21028536558151245,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0072,
+      "step": 1005
+    },
+    {
+      "epoch": 9.005871559633027,
+      "grad_norm": 0.14233331382274628,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0051,
+      "step": 1006
+    },
+    {
+      "epoch": 9.006238532110093,
+      "grad_norm": 0.27787259221076965,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0103,
+      "step": 1007
+    },
+    {
+      "epoch": 9.006605504587156,
+      "grad_norm": 7.600110054016113,
+      "learning_rate": 2.5e-05,
+      "loss": 0.303,
+      "step": 1008
+    },
+    {
+      "epoch": 9.00697247706422,
+      "grad_norm": 7.821009635925293,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6744,
+      "step": 1009
+    },
+    {
+      "epoch": 9.007339449541284,
+      "grad_norm": 0.3863763213157654,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0079,
+      "step": 1010
+    },
+    {
+      "epoch": 9.007706422018348,
+      "grad_norm": 0.16768227517604828,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0062,
+      "step": 1011
+    },
+    {
+      "epoch": 9.008073394495414,
+      "grad_norm": 0.0627545416355133,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0025,
+      "step": 1012
+    },
+    {
+      "epoch": 9.008440366972478,
+      "grad_norm": 0.08043862879276276,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0032,
+      "step": 1013
+    },
+    {
+      "epoch": 9.008807339449541,
+      "grad_norm": 30.170005798339844,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6787,
+      "step": 1014
+    },
+    {
+      "epoch": 9.009174311926605,
+      "grad_norm": 0.08431703597307205,
+      "learning_rate": 2.5e-05,
+      "loss": 0.003,
+      "step": 1015
+    },
+    {
+      "epoch": 9.009541284403669,
+      "grad_norm": 1.5005310773849487,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0504,
+      "step": 1016
+    },
+    {
+      "epoch": 9.009908256880735,
+      "grad_norm": 17.917016983032227,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1786,
+      "step": 1017
+    },
+    {
+      "epoch": 9.010275229357799,
+      "grad_norm": 0.03360790014266968,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0013,
+      "step": 1018
+    },
+    {
+      "epoch": 9.010642201834862,
+      "grad_norm": 0.09401139616966248,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0032,
+      "step": 1019
+    },
+    {
+      "epoch": 9.011009174311926,
+      "grad_norm": 1.6750723123550415,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0173,
+      "step": 1020
+    },
+    {
+      "epoch": 9.01137614678899,
+      "grad_norm": 0.06304969638586044,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0023,
+      "step": 1021
+    },
+    {
+      "epoch": 9.011743119266056,
+      "grad_norm": 0.0607866570353508,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0024,
+      "step": 1022
+    },
+    {
+      "epoch": 9.01211009174312,
+      "grad_norm": 0.1267731934785843,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0041,
+      "step": 1023
+    },
+    {
+      "epoch": 9.012477064220183,
+      "grad_norm": 4.302545070648193,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0427,
+      "step": 1024
+    },
+    {
+      "epoch": 9.012844036697247,
+      "grad_norm": 0.05810270830988884,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0021,
+      "step": 1025
+    },
+    {
+      "epoch": 9.013211009174311,
+      "grad_norm": 0.025547843426465988,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0011,
+      "step": 1026
+    },
+    {
+      "epoch": 9.013577981651377,
+      "grad_norm": 0.03760534152388573,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0014,
+      "step": 1027
+    },
+    {
+      "epoch": 9.01394495412844,
+      "grad_norm": 9.723259925842285,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1421,
+      "step": 1028
+    },
+    {
+      "epoch": 9.014311926605505,
+      "grad_norm": 8.889154434204102,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0977,
+      "step": 1029
+    },
+    {
+      "epoch": 9.014678899082568,
+      "grad_norm": 0.08079709112644196,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0027,
+      "step": 1030
+    },
+    {
+      "epoch": 9.015045871559632,
+      "grad_norm": 0.11405689269304276,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0031,
+      "step": 1031
+    },
+    {
+      "epoch": 9.015412844036698,
+      "grad_norm": 0.04440542683005333,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0009,
+      "step": 1032
+    },
+    {
+      "epoch": 9.015779816513762,
+      "grad_norm": 0.05163365602493286,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0018,
+      "step": 1033
+    },
+    {
+      "epoch": 9.016146788990826,
+      "grad_norm": 0.03253662958741188,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0013,
+      "step": 1034
+    },
+    {
+      "epoch": 9.01651376146789,
+      "grad_norm": 0.04269241914153099,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0017,
+      "step": 1035
+    },
+    {
+      "epoch": 9.016880733944953,
+      "grad_norm": 0.03373705595731735,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0013,
+      "step": 1036
+    },
+    {
+      "epoch": 9.017247706422019,
+      "grad_norm": 0.014124157838523388,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0006,
+      "step": 1037
+    },
+    {
+      "epoch": 9.017614678899083,
+      "grad_norm": 0.0489923320710659,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0021,
+      "step": 1038
+    },
+    {
+      "epoch": 9.017981651376147,
+      "grad_norm": 0.06219998374581337,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0022,
+      "step": 1039
+    },
+    {
+      "epoch": 9.01834862385321,
+      "grad_norm": 0.015159577131271362,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0006,
+      "step": 1040
+    },
+    {
+      "epoch": 9.018715596330276,
+      "grad_norm": 0.09099878370761871,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0034,
+      "step": 1041
+    },
+    {
+      "epoch": 9.01908256880734,
+      "grad_norm": 0.07552758604288101,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0024,
+      "step": 1042
+    },
+    {
+      "epoch": 9.019449541284404,
+      "grad_norm": 0.05425313487648964,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0022,
+      "step": 1043
+    },
+    {
+      "epoch": 9.019816513761468,
+      "grad_norm": 0.029285207390785217,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0013,
+      "step": 1044
+    },
+    {
+      "epoch": 9.020183486238532,
+      "grad_norm": 0.07954943925142288,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0026,
+      "step": 1045
+    },
+    {
+      "epoch": 9.020550458715597,
+      "grad_norm": 0.04610136151313782,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0015,
+      "step": 1046
+    },
+    {
+      "epoch": 9.020917431192661,
+      "grad_norm": 0.017000334337353706,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0006,
+      "step": 1047
+    },
+    {
+      "epoch": 9.021284403669725,
+      "grad_norm": 0.3730951249599457,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0091,
+      "step": 1048
+    },
+    {
+      "epoch": 9.021651376146789,
+      "grad_norm": 0.06654736399650574,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0025,
+      "step": 1049
+    },
+    {
+      "epoch": 9.022018348623853,
+      "grad_norm": 0.022919123992323875,
+      "learning_rate": 2.5e-05,
+      "loss": 0.001,
+      "step": 1050
+    },
+    {
+      "epoch": 9.022385321100918,
+      "grad_norm": 0.027949770912528038,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0011,
+      "step": 1051
+    },
+    {
+      "epoch": 9.022752293577982,
+      "grad_norm": 0.03364739194512367,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0014,
+      "step": 1052
+    },
+    {
+      "epoch": 9.023119266055046,
+      "grad_norm": 0.1934141367673874,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0047,
+      "step": 1053
+    },
+    {
+      "epoch": 9.02348623853211,
+      "grad_norm": 0.024477841332554817,
+      "learning_rate": 2.5e-05,
+      "loss": 0.001,
+      "step": 1054
+    },
+    {
+      "epoch": 9.023853211009174,
+      "grad_norm": 0.024517137557268143,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0009,
+      "step": 1055
+    },
+    {
+      "epoch": 9.02422018348624,
+      "grad_norm": 2.7035939693450928,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0446,
+      "step": 1056
+    },
+    {
+      "epoch": 9.024587155963303,
+      "grad_norm": 0.00924652349203825,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0004,
+      "step": 1057
+    },
+    {
+      "epoch": 9.024954128440367,
+      "grad_norm": 0.030514901503920555,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0009,
+      "step": 1058
+    },
+    {
+      "epoch": 9.02532110091743,
+      "grad_norm": 0.10835156589746475,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0018,
+      "step": 1059
+    },
+    {
+      "epoch": 9.025688073394495,
+      "grad_norm": 0.10482455044984818,
+      "learning_rate": 2.5e-05,
+      "loss": 0.002,
+      "step": 1060
+    },
+    {
+      "epoch": 9.02605504587156,
+      "grad_norm": 0.021525854244828224,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0009,
+      "step": 1061
+    },
+    {
+      "epoch": 9.026422018348624,
+      "grad_norm": 21.751087188720703,
+      "learning_rate": 2.5e-05,
+      "loss": 0.2537,
+      "step": 1062
+    },
+    {
+      "epoch": 9.026788990825688,
+      "grad_norm": 0.016646642237901688,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0006,
+      "step": 1063
+    },
+    {
+      "epoch": 9.027155963302752,
+      "grad_norm": 0.06881947815418243,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0016,
+      "step": 1064
+    },
+    {
+      "epoch": 9.027522935779816,
+      "grad_norm": 0.06905637681484222,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0014,
+      "step": 1065
+    },
+    {
+      "epoch": 9.027889908256881,
+      "grad_norm": 0.03242538124322891,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0011,
+      "step": 1066
+    },
+    {
+      "epoch": 9.028256880733945,
+      "grad_norm": 0.05881950631737709,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0016,
+      "step": 1067
+    },
+    {
+      "epoch": 9.028623853211009,
+      "grad_norm": 20.61574363708496,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1606,
+      "step": 1068
+    },
+    {
+      "epoch": 9.028990825688073,
+      "grad_norm": 26.697114944458008,
+      "learning_rate": 2.5e-05,
+      "loss": 0.9182,
+      "step": 1069
+    },
+    {
+      "epoch": 9.029357798165137,
+      "grad_norm": 0.07642514258623123,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0019,
+      "step": 1070
+    },
+    {
+      "epoch": 9.029724770642202,
+      "grad_norm": 0.024645140394568443,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0008,
+      "step": 1071
+    },
+    {
+      "epoch": 9.030091743119266,
+      "grad_norm": 7.698794364929199,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5413,
+      "step": 1072
+    },
+    {
+      "epoch": 9.03045871559633,
+      "grad_norm": 7.868088245391846,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6788,
+      "step": 1073
+    },
+    {
+      "epoch": 9.030825688073394,
+      "grad_norm": 0.06784772872924805,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0021,
+      "step": 1074
+    },
+    {
+      "epoch": 9.03119266055046,
+      "grad_norm": 0.013106757774949074,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0006,
+      "step": 1075
+    },
+    {
+      "epoch": 9.031559633027523,
+      "grad_norm": 0.030605705454945564,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0012,
+      "step": 1076
+    },
+    {
+      "epoch": 9.031926605504587,
+      "grad_norm": 4.397395133972168,
+      "learning_rate": 2.5e-05,
+      "loss": 0.681,
+      "step": 1077
+    },
+    {
+      "epoch": 9.032293577981651,
+      "grad_norm": 23.4879150390625,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5132,
+      "step": 1078
+    },
+    {
+      "epoch": 9.032660550458715,
+      "grad_norm": 0.025631044059991837,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0011,
+      "step": 1079
+    },
+    {
+      "epoch": 9.03302752293578,
+      "grad_norm": 0.021947994828224182,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0009,
+      "step": 1080
+    },
+    {
+      "epoch": 9.033394495412844,
+      "grad_norm": 4.538063049316406,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1523,
+      "step": 1081
+    },
+    {
+      "epoch": 9.033761467889908,
+      "grad_norm": 4.6625542640686035,
+      "learning_rate": 2.5e-05,
+      "loss": 0.7827,
+      "step": 1082
+    },
+    {
+      "epoch": 9.034128440366972,
+      "grad_norm": 4.8275628089904785,
+      "learning_rate": 2.5e-05,
+      "loss": 0.7244,
+      "step": 1083
+    },
+    {
+      "epoch": 9.034495412844036,
+      "grad_norm": 0.19174185395240784,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0026,
+      "step": 1084
+    },
+    {
+      "epoch": 9.034862385321102,
+      "grad_norm": 0.05068483203649521,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0019,
+      "step": 1085
+    },
+    {
+      "epoch": 9.035229357798165,
+      "grad_norm": 0.09837653487920761,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0023,
+      "step": 1086
+    },
+    {
+      "epoch": 9.03559633027523,
+      "grad_norm": 0.3001652956008911,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0051,
+      "step": 1087
+    },
+    {
+      "epoch": 9.035963302752293,
+      "grad_norm": 4.821220397949219,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6117,
+      "step": 1088
+    },
+    {
+      "epoch": 9.036330275229357,
+      "grad_norm": 0.1156148687005043,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0029,
+      "step": 1089
+    },
+    {
+      "epoch": 9.036697247706423,
+      "grad_norm": 0.1332520991563797,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0034,
+      "step": 1090
+    },
+    {
+      "epoch": 9.037064220183487,
+      "grad_norm": 0.4530578553676605,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0107,
+      "step": 1091
+    },
+    {
+      "epoch": 9.03743119266055,
+      "grad_norm": 0.17837747931480408,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0054,
+      "step": 1092
+    },
+    {
+      "epoch": 9.037798165137614,
+      "grad_norm": 0.08601999282836914,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0029,
+      "step": 1093
+    },
+    {
+      "epoch": 9.038165137614678,
+      "grad_norm": 1.0509358644485474,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0344,
+      "step": 1094
+    },
+    {
+      "epoch": 9.038532110091744,
+      "grad_norm": 0.08061433583498001,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0026,
+      "step": 1095
+    },
+    {
+      "epoch": 9.038899082568808,
+      "grad_norm": 0.07802833616733551,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0028,
+      "step": 1096
+    },
+    {
+      "epoch": 9.039266055045871,
+      "grad_norm": 0.24374306201934814,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0072,
+      "step": 1097
+    },
+    {
+      "epoch": 9.039633027522935,
+      "grad_norm": 3.3682096004486084,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0516,
+      "step": 1098
+    },
+    {
+      "epoch": 9.04,
+      "grad_norm": 0.06994065642356873,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0027,
+      "step": 1099
+    },
+    {
+      "epoch": 9.040366972477065,
+      "grad_norm": 0.06850399821996689,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0017,
+      "step": 1100
+    },
+    {
+      "epoch": 9.040366972477065,
+      "eval_accuracy": 0.4940828402366864,
+      "eval_loss": 2.496060609817505,
+      "eval_runtime": 25.956,
+      "eval_samples_per_second": 13.022,
+      "eval_steps_per_second": 1.657,
+      "step": 1100
+    },
+    {
+      "epoch": 10.000366972477064,
+      "grad_norm": 0.3944116234779358,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0099,
+      "step": 1101
+    },
+    {
+      "epoch": 10.000733944954128,
+      "grad_norm": 0.36435312032699585,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0118,
+      "step": 1102
+    },
+    {
+      "epoch": 10.001100917431193,
+      "grad_norm": 0.22293834388256073,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0062,
+      "step": 1103
+    },
+    {
+      "epoch": 10.001467889908257,
+      "grad_norm": 6.4955925941467285,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3832,
+      "step": 1104
+    },
+    {
+      "epoch": 10.001834862385321,
+      "grad_norm": 0.22054608166217804,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0052,
+      "step": 1105
+    },
+    {
+      "epoch": 10.002201834862385,
+      "grad_norm": 0.10869724303483963,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0025,
+      "step": 1106
+    },
+    {
+      "epoch": 10.002568807339449,
+      "grad_norm": 1.0462502241134644,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0366,
+      "step": 1107
+    },
+    {
+      "epoch": 10.002935779816514,
+      "grad_norm": 7.7954254150390625,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4315,
+      "step": 1108
+    },
+    {
+      "epoch": 10.003302752293578,
+      "grad_norm": 0.04257207736372948,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0017,
+      "step": 1109
+    },
+    {
+      "epoch": 10.003669724770642,
+      "grad_norm": 0.0709013044834137,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0024,
+      "step": 1110
+    },
+    {
+      "epoch": 10.004036697247706,
+      "grad_norm": 12.050844192504883,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3463,
+      "step": 1111
+    },
+    {
+      "epoch": 10.00440366972477,
+      "grad_norm": 0.5227145552635193,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0061,
+      "step": 1112
+    },
+    {
+      "epoch": 10.004770642201835,
+      "grad_norm": 0.05547614023089409,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0016,
+      "step": 1113
+    },
+    {
+      "epoch": 10.0051376146789,
+      "grad_norm": 0.03555474430322647,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0015,
+      "step": 1114
+    },
+    {
+      "epoch": 10.005504587155963,
+      "grad_norm": 0.08506189286708832,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0032,
+      "step": 1115
+    },
+    {
+      "epoch": 10.005871559633027,
+      "grad_norm": 0.04708164557814598,
+      "learning_rate": 2.5e-05,
+      "loss": 0.002,
+      "step": 1116
+    },
+    {
+      "epoch": 10.006238532110093,
+      "grad_norm": 5.031776428222656,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5939,
+      "step": 1117
+    },
+    {
+      "epoch": 10.006605504587156,
+      "grad_norm": 0.26807036995887756,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0054,
+      "step": 1118
+    },
+    {
+      "epoch": 10.00697247706422,
+      "grad_norm": 8.716951370239258,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0304,
+      "step": 1119
+    },
+    {
+      "epoch": 10.007339449541284,
+      "grad_norm": 0.061252180486917496,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0019,
+      "step": 1120
+    },
+    {
+      "epoch": 10.007706422018348,
+      "grad_norm": 0.3089843690395355,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0039,
+      "step": 1121
+    },
+    {
+      "epoch": 10.008073394495414,
+      "grad_norm": 0.0742259994149208,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0028,
+      "step": 1122
+    },
+    {
+      "epoch": 10.008440366972478,
+      "grad_norm": 0.011265672743320465,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0005,
+      "step": 1123
+    },
+    {
+      "epoch": 10.008807339449541,
+      "grad_norm": 1.0335417985916138,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0064,
+      "step": 1124
+    },
+    {
+      "epoch": 10.009174311926605,
+      "grad_norm": 0.026845503598451614,
+      "learning_rate": 2.5e-05,
+      "loss": 0.001,
+      "step": 1125
+    },
+    {
+      "epoch": 10.009541284403669,
+      "grad_norm": 4.867141246795654,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5306,
+      "step": 1126
+    },
+    {
+      "epoch": 10.009908256880735,
+      "grad_norm": 0.06371437013149261,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0019,
+      "step": 1127
+    },
+    {
+      "epoch": 10.010275229357799,
+      "grad_norm": 0.04716891422867775,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0016,
+      "step": 1128
+    },
+    {
+      "epoch": 10.010642201834862,
+      "grad_norm": 0.06217212975025177,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0019,
+      "step": 1129
+    },
+    {
+      "epoch": 10.011009174311926,
+      "grad_norm": 0.009243843145668507,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0004,
+      "step": 1130
+    },
+    {
+      "epoch": 10.01137614678899,
+      "grad_norm": 0.09474127739667892,
+      "learning_rate": 2.5e-05,
+      "loss": 0.003,
+      "step": 1131
+    },
+    {
+      "epoch": 10.011743119266056,
+      "grad_norm": 0.007971953600645065,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0003,
+      "step": 1132
+    },
+    {
+      "epoch": 10.01211009174312,
+      "grad_norm": 0.13930995762348175,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0041,
+      "step": 1133
+    },
+    {
+      "epoch": 10.012477064220183,
+      "grad_norm": 0.1084611788392067,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0034,
+      "step": 1134
+    },
+    {
+      "epoch": 10.012844036697247,
+      "grad_norm": 0.07797815650701523,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0029,
+      "step": 1135
+    },
+    {
+      "epoch": 10.013211009174311,
+      "grad_norm": 14.477725982666016,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4046,
+      "step": 1136
+    },
+    {
+      "epoch": 10.013577981651377,
+      "grad_norm": 43.18414306640625,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1724,
+      "step": 1137
+    },
+    {
+      "epoch": 10.01394495412844,
+      "grad_norm": 0.03222573921084404,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0012,
+      "step": 1138
+    },
+    {
+      "epoch": 10.014311926605505,
+      "grad_norm": 7.187230110168457,
+      "learning_rate": 2.5e-05,
+      "loss": 0.192,
+      "step": 1139
+    },
+    {
+      "epoch": 10.014678899082568,
+      "grad_norm": 0.04186500236392021,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0014,
+      "step": 1140
+    },
+    {
+      "epoch": 10.015045871559632,
+      "grad_norm": 1.9843147993087769,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0089,
+      "step": 1141
+    },
+    {
+      "epoch": 10.015412844036698,
+      "grad_norm": 0.3523111045360565,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0089,
+      "step": 1142
+    },
+    {
+      "epoch": 10.015779816513762,
+      "grad_norm": 0.10755079239606857,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0017,
+      "step": 1143
+    },
+    {
+      "epoch": 10.016146788990826,
+      "grad_norm": 1.0479165315628052,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0188,
+      "step": 1144
+    },
+    {
+      "epoch": 10.01651376146789,
+      "grad_norm": 0.036803312599658966,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0012,
+      "step": 1145
+    },
+    {
+      "epoch": 10.016880733944953,
+      "grad_norm": 0.09662558883428574,
+      "learning_rate": 2.5e-05,
+      "loss": 0.002,
+      "step": 1146
+    },
+    {
+      "epoch": 10.017247706422019,
+      "grad_norm": 0.12040849775075912,
+      "learning_rate": 2.5e-05,
+      "loss": 0.003,
+      "step": 1147
+    },
+    {
+      "epoch": 10.017614678899083,
+      "grad_norm": 0.1467740535736084,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0028,
+      "step": 1148
+    },
+    {
+      "epoch": 10.017981651376147,
+      "grad_norm": 5.5427961349487305,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3354,
+      "step": 1149
+    },
+    {
+      "epoch": 10.01834862385321,
+      "grad_norm": 0.042355697602033615,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0016,
+      "step": 1150
+    },
+    {
+      "epoch": 10.018715596330276,
+      "grad_norm": 0.031690895557403564,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0012,
+      "step": 1151
+    },
+    {
+      "epoch": 10.01908256880734,
+      "grad_norm": 0.06647089123725891,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0027,
+      "step": 1152
+    },
+    {
+      "epoch": 10.019449541284404,
+      "grad_norm": 0.021304456517100334,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0009,
+      "step": 1153
+    },
+    {
+      "epoch": 10.019816513761468,
+      "grad_norm": 7.925635814666748,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6061,
+      "step": 1154
+    },
+    {
+      "epoch": 10.020183486238532,
+      "grad_norm": 0.03811515122652054,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0014,
+      "step": 1155
+    },
+    {
+      "epoch": 10.020550458715597,
+      "grad_norm": 0.0280921533703804,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0011,
+      "step": 1156
+    },
+    {
+      "epoch": 10.020917431192661,
+      "grad_norm": 0.03383231535553932,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0013,
+      "step": 1157
+    },
+    {
+      "epoch": 10.021284403669725,
+      "grad_norm": 0.01860538125038147,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0007,
+      "step": 1158
+    },
+    {
+      "epoch": 10.021651376146789,
+      "grad_norm": 0.02626795694231987,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0009,
+      "step": 1159
+    },
+    {
+      "epoch": 10.022018348623853,
+      "grad_norm": 18.885425567626953,
+      "learning_rate": 2.5e-05,
+      "loss": 0.9345,
+      "step": 1160
+    },
+    {
+      "epoch": 10.022385321100918,
+      "grad_norm": 13.233046531677246,
+      "learning_rate": 2.5e-05,
+      "loss": 1.3814,
+      "step": 1161
+    },
+    {
+      "epoch": 10.022752293577982,
+      "grad_norm": 0.021473852917551994,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0008,
+      "step": 1162
+    },
+    {
+      "epoch": 10.023119266055046,
+      "grad_norm": 0.022396618500351906,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0009,
+      "step": 1163
+    },
+    {
+      "epoch": 10.02348623853211,
+      "grad_norm": 4.693010330200195,
+      "learning_rate": 2.5e-05,
+      "loss": 0.7723,
+      "step": 1164
+    },
+    {
+      "epoch": 10.023853211009174,
+      "grad_norm": 0.11017192155122757,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0027,
+      "step": 1165
+    },
+    {
+      "epoch": 10.02422018348624,
+      "grad_norm": 0.023172445595264435,
+      "learning_rate": 2.5e-05,
+      "loss": 0.001,
+      "step": 1166
+    },
+    {
+      "epoch": 10.024587155963303,
+      "grad_norm": 0.4965081214904785,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0079,
+      "step": 1167
+    },
+    {
+      "epoch": 10.024954128440367,
+      "grad_norm": 6.461441516876221,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1013,
+      "step": 1168
+    },
+    {
+      "epoch": 10.02532110091743,
+      "grad_norm": 1.1123604774475098,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0465,
+      "step": 1169
+    },
+    {
+      "epoch": 10.025688073394495,
+      "grad_norm": 3.4512081146240234,
+      "learning_rate": 2.5e-05,
+      "loss": 0.2249,
+      "step": 1170
+    },
+    {
+      "epoch": 10.02605504587156,
+      "grad_norm": 0.13902178406715393,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0046,
+      "step": 1171
+    },
+    {
+      "epoch": 10.026422018348624,
+      "grad_norm": 1.2499973773956299,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0115,
+      "step": 1172
+    },
+    {
+      "epoch": 10.026788990825688,
+      "grad_norm": 0.04520212858915329,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0016,
+      "step": 1173
+    },
+    {
+      "epoch": 10.027155963302752,
+      "grad_norm": 0.337846040725708,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0048,
+      "step": 1174
+    },
+    {
+      "epoch": 10.027522935779816,
+      "grad_norm": 25.42504119873047,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0685,
+      "step": 1175
+    },
+    {
+      "epoch": 10.027889908256881,
+      "grad_norm": 0.12650935351848602,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0043,
+      "step": 1176
+    },
+    {
+      "epoch": 10.028256880733945,
+      "grad_norm": 0.0293078925460577,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0011,
+      "step": 1177
+    },
+    {
+      "epoch": 10.028623853211009,
+      "grad_norm": 4.093510150909424,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6146,
+      "step": 1178
+    },
+    {
+      "epoch": 10.028990825688073,
+      "grad_norm": 0.2520609200000763,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0057,
+      "step": 1179
+    },
+    {
+      "epoch": 10.029357798165137,
+      "grad_norm": 0.0548003688454628,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0022,
+      "step": 1180
+    },
+    {
+      "epoch": 10.029724770642202,
+      "grad_norm": 0.21667346358299255,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0083,
+      "step": 1181
+    },
+    {
+      "epoch": 10.030091743119266,
+      "grad_norm": 0.07467496395111084,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0026,
+      "step": 1182
+    },
+    {
+      "epoch": 10.03045871559633,
+      "grad_norm": 0.06521061062812805,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0017,
+      "step": 1183
+    },
+    {
+      "epoch": 10.030825688073394,
+      "grad_norm": 0.06587977707386017,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0025,
+      "step": 1184
+    },
+    {
+      "epoch": 10.03119266055046,
+      "grad_norm": 0.03392113000154495,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0013,
+      "step": 1185
+    },
+    {
+      "epoch": 10.031559633027523,
+      "grad_norm": 0.021968001499772072,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0009,
+      "step": 1186
+    },
+    {
+      "epoch": 10.031926605504587,
+      "grad_norm": 37.58613586425781,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4458,
+      "step": 1187
+    },
+    {
+      "epoch": 10.032293577981651,
+      "grad_norm": 3.879697322845459,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1391,
+      "step": 1188
+    },
+    {
+      "epoch": 10.032660550458715,
+      "grad_norm": 0.10899520665407181,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0032,
+      "step": 1189
+    },
+    {
+      "epoch": 10.03302752293578,
+      "grad_norm": 0.019114593043923378,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0008,
+      "step": 1190
+    },
+    {
+      "epoch": 10.033394495412844,
+      "grad_norm": 0.07935284823179245,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0026,
+      "step": 1191
+    },
+    {
+      "epoch": 10.033761467889908,
+      "grad_norm": 0.08476589620113373,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0032,
+      "step": 1192
+    },
+    {
+      "epoch": 10.034128440366972,
+      "grad_norm": 4.668460369110107,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0901,
+      "step": 1193
+    },
+    {
+      "epoch": 10.034495412844036,
+      "grad_norm": 0.05402791500091553,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0019,
+      "step": 1194
+    },
+    {
+      "epoch": 10.034862385321102,
+      "grad_norm": 0.03557025268673897,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0015,
+      "step": 1195
+    },
+    {
+      "epoch": 10.035229357798165,
+      "grad_norm": 7.520567417144775,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5252,
+      "step": 1196
+    },
+    {
+      "epoch": 10.03559633027523,
+      "grad_norm": 6.0408124923706055,
+      "learning_rate": 2.5e-05,
+      "loss": 0.378,
+      "step": 1197
+    },
+    {
+      "epoch": 10.035963302752293,
+      "grad_norm": 0.05525198206305504,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0017,
+      "step": 1198
+    },
+    {
+      "epoch": 10.036330275229357,
+      "grad_norm": 0.042404692620038986,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0016,
+      "step": 1199
+    },
+    {
+      "epoch": 10.036697247706423,
+      "grad_norm": 1.2513283491134644,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0185,
+      "step": 1200
+    },
+    {
+      "epoch": 10.037064220183487,
+      "grad_norm": 0.8135076761245728,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0257,
+      "step": 1201
+    },
+    {
+      "epoch": 10.03743119266055,
+      "grad_norm": 20.929569244384766,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3572,
+      "step": 1202
+    },
+    {
+      "epoch": 10.037798165137614,
+      "grad_norm": 1.4500948190689087,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0133,
+      "step": 1203
+    },
+    {
+      "epoch": 10.038165137614678,
+      "grad_norm": 0.4335315227508545,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0071,
+      "step": 1204
+    },
+    {
+      "epoch": 10.038532110091744,
+      "grad_norm": 1.1914621591567993,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0152,
+      "step": 1205
+    },
+    {
+      "epoch": 10.038899082568808,
+      "grad_norm": 5.105964660644531,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4502,
+      "step": 1206
+    },
+    {
+      "epoch": 10.039266055045871,
+      "grad_norm": 0.8331950902938843,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0083,
+      "step": 1207
+    },
+    {
+      "epoch": 10.039633027522935,
+      "grad_norm": 15.225933074951172,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4174,
+      "step": 1208
+    },
+    {
+      "epoch": 10.04,
+      "grad_norm": 0.073668472468853,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0022,
+      "step": 1209
+    },
+    {
+      "epoch": 10.040366972477065,
+      "grad_norm": 4.010093688964844,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4934,
+      "step": 1210
+    },
+    {
+      "epoch": 10.040366972477065,
+      "eval_accuracy": 0.4940828402366864,
+      "eval_loss": 2.97847056388855,
+      "eval_runtime": 26.0578,
+      "eval_samples_per_second": 12.971,
+      "eval_steps_per_second": 1.65,
+      "step": 1210
+    },
+    {
+      "epoch": 11.000366972477064,
+      "grad_norm": 0.20556648075580597,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0036,
+      "step": 1211
+    },
+    {
+      "epoch": 11.000733944954128,
+      "grad_norm": 0.036546025425195694,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0012,
+      "step": 1212
+    },
+    {
+      "epoch": 11.001100917431193,
+      "grad_norm": 28.453876495361328,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3423,
+      "step": 1213
+    },
+    {
+      "epoch": 11.001467889908257,
+      "grad_norm": 0.5969877243041992,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0104,
+      "step": 1214
+    },
+    {
+      "epoch": 11.001834862385321,
+      "grad_norm": 0.17068693041801453,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0039,
+      "step": 1215
+    },
+    {
+      "epoch": 11.002201834862385,
+      "grad_norm": 0.010091293603181839,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0004,
+      "step": 1216
+    },
+    {
+      "epoch": 11.002568807339449,
+      "grad_norm": 3.6620140075683594,
+      "learning_rate": 2.5e-05,
+      "loss": 0.8042,
+      "step": 1217
+    },
+    {
+      "epoch": 11.002935779816514,
+      "grad_norm": 0.029740633442997932,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0012,
+      "step": 1218
+    },
+    {
+      "epoch": 11.003302752293578,
+      "grad_norm": 3.5253098011016846,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0136,
+      "step": 1219
+    },
+    {
+      "epoch": 11.003669724770642,
+      "grad_norm": 0.20947608351707458,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0064,
+      "step": 1220
+    },
+    {
+      "epoch": 11.004036697247706,
+      "grad_norm": 0.12806753814220428,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0046,
+      "step": 1221
+    },
+    {
+      "epoch": 11.00440366972477,
+      "grad_norm": 0.02363951876759529,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0009,
+      "step": 1222
+    },
+    {
+      "epoch": 11.004770642201835,
+      "grad_norm": 0.01555804442614317,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0006,
+      "step": 1223
+    },
+    {
+      "epoch": 11.0051376146789,
+      "grad_norm": 9.572732925415039,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6465,
+      "step": 1224
+    },
+    {
+      "epoch": 11.005504587155963,
+      "grad_norm": 0.02781197428703308,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0013,
+      "step": 1225
+    },
+    {
+      "epoch": 11.005871559633027,
+      "grad_norm": 0.051709454506635666,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0023,
+      "step": 1226
+    },
+    {
+      "epoch": 11.006238532110093,
+      "grad_norm": 7.341026782989502,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0524,
+      "step": 1227
+    },
+    {
+      "epoch": 11.006605504587156,
+      "grad_norm": 0.21090282499790192,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0051,
+      "step": 1228
+    },
+    {
+      "epoch": 11.00697247706422,
+      "grad_norm": 0.033602479845285416,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0012,
+      "step": 1229
+    },
+    {
+      "epoch": 11.007339449541284,
+      "grad_norm": 0.23407487571239471,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0067,
+      "step": 1230
+    },
+    {
+      "epoch": 11.007706422018348,
+      "grad_norm": 0.1046948954463005,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0029,
+      "step": 1231
+    },
+    {
+      "epoch": 11.008073394495414,
+      "grad_norm": 2.484178066253662,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0534,
+      "step": 1232
+    },
+    {
+      "epoch": 11.008440366972478,
+      "grad_norm": 0.258402556180954,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0058,
+      "step": 1233
+    },
+    {
+      "epoch": 11.008807339449541,
+      "grad_norm": 3.7868142127990723,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6497,
+      "step": 1234
+    },
+    {
+      "epoch": 11.009174311926605,
+      "grad_norm": 0.3882410228252411,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0125,
+      "step": 1235
+    },
+    {
+      "epoch": 11.009541284403669,
+      "grad_norm": 0.016743695363402367,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0008,
+      "step": 1236
+    },
+    {
+      "epoch": 11.009908256880735,
+      "grad_norm": 0.03597034513950348,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0015,
+      "step": 1237
+    },
+    {
+      "epoch": 11.010275229357799,
+      "grad_norm": 0.04201594367623329,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0016,
+      "step": 1238
+    },
+    {
+      "epoch": 11.010642201834862,
+      "grad_norm": 13.533677101135254,
+      "learning_rate": 2.5e-05,
+      "loss": 1.0923,
+      "step": 1239
+    },
+    {
+      "epoch": 11.011009174311926,
+      "grad_norm": 2.689098596572876,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0145,
+      "step": 1240
+    },
+    {
+      "epoch": 11.01137614678899,
+      "grad_norm": 0.02373415417969227,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0007,
+      "step": 1241
+    },
+    {
+      "epoch": 11.011743119266056,
+      "grad_norm": 0.10941294580698013,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0036,
+      "step": 1242
+    },
+    {
+      "epoch": 11.01211009174312,
+      "grad_norm": 0.06366754323244095,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0023,
+      "step": 1243
+    },
+    {
+      "epoch": 11.012477064220183,
+      "grad_norm": 0.07805350422859192,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0017,
+      "step": 1244
+    },
+    {
+      "epoch": 11.012844036697247,
+      "grad_norm": 7.122638702392578,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0393,
+      "step": 1245
+    },
+    {
+      "epoch": 11.013211009174311,
+      "grad_norm": 0.017953095957636833,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0008,
+      "step": 1246
+    },
+    {
+      "epoch": 11.013577981651377,
+      "grad_norm": 0.15503093600273132,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0071,
+      "step": 1247
+    },
+    {
+      "epoch": 11.01394495412844,
+      "grad_norm": 0.08123112469911575,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0026,
+      "step": 1248
+    },
+    {
+      "epoch": 11.014311926605505,
+      "grad_norm": 0.11489523947238922,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0034,
+      "step": 1249
+    },
+    {
+      "epoch": 11.014678899082568,
+      "grad_norm": 0.07481618225574493,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0023,
+      "step": 1250
+    },
+    {
+      "epoch": 11.015045871559632,
+      "grad_norm": 25.060256958007812,
+      "learning_rate": 2.5e-05,
+      "loss": 0.2635,
+      "step": 1251
+    },
+    {
+      "epoch": 11.015412844036698,
+      "grad_norm": 0.043068282306194305,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0017,
+      "step": 1252
+    },
+    {
+      "epoch": 11.015779816513762,
+      "grad_norm": 0.4010566473007202,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0059,
+      "step": 1253
+    },
+    {
+      "epoch": 11.016146788990826,
+      "grad_norm": 0.02619306556880474,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0011,
+      "step": 1254
+    },
+    {
+      "epoch": 11.01651376146789,
+      "grad_norm": 0.032633133232593536,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0012,
+      "step": 1255
+    },
+    {
+      "epoch": 11.016880733944953,
+      "grad_norm": 0.19353869557380676,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0051,
+      "step": 1256
+    },
+    {
+      "epoch": 11.017247706422019,
+      "grad_norm": 0.020986469462513924,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0009,
+      "step": 1257
+    },
+    {
+      "epoch": 11.017614678899083,
+      "grad_norm": 22.179920196533203,
+      "learning_rate": 2.5e-05,
+      "loss": 0.8498,
+      "step": 1258
+    },
+    {
+      "epoch": 11.017981651376147,
+      "grad_norm": 3.850023031234741,
+      "learning_rate": 2.5e-05,
+      "loss": 0.7925,
+      "step": 1259
+    },
+    {
+      "epoch": 11.01834862385321,
+      "grad_norm": 0.007171152159571648,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0002,
+      "step": 1260
+    },
+    {
+      "epoch": 11.018715596330276,
+      "grad_norm": 0.08312695473432541,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0027,
+      "step": 1261
+    },
+    {
+      "epoch": 11.01908256880734,
+      "grad_norm": 0.027183033525943756,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0012,
+      "step": 1262
+    },
+    {
+      "epoch": 11.019449541284404,
+      "grad_norm": 0.03362927958369255,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0014,
+      "step": 1263
+    },
+    {
+      "epoch": 11.019816513761468,
+      "grad_norm": 0.037589918822050095,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0015,
+      "step": 1264
+    },
+    {
+      "epoch": 11.020183486238532,
+      "grad_norm": 34.70879364013672,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4033,
+      "step": 1265
+    },
+    {
+      "epoch": 11.020550458715597,
+      "grad_norm": 0.031473688781261444,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0014,
+      "step": 1266
+    },
+    {
+      "epoch": 11.020917431192661,
+      "grad_norm": 0.049193017184734344,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0019,
+      "step": 1267
+    },
+    {
+      "epoch": 11.021284403669725,
+      "grad_norm": 0.03142306208610535,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0014,
+      "step": 1268
+    },
+    {
+      "epoch": 11.021651376146789,
+      "grad_norm": 0.040651679039001465,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0016,
+      "step": 1269
+    },
+    {
+      "epoch": 11.022018348623853,
+      "grad_norm": 0.06500530242919922,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0022,
+      "step": 1270
+    },
+    {
+      "epoch": 11.022385321100918,
+      "grad_norm": 0.05012776330113411,
+      "learning_rate": 2.5e-05,
+      "loss": 0.002,
+      "step": 1271
+    },
+    {
+      "epoch": 11.022752293577982,
+      "grad_norm": 0.17458389699459076,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0082,
+      "step": 1272
+    },
+    {
+      "epoch": 11.023119266055046,
+      "grad_norm": 0.06550407409667969,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0015,
+      "step": 1273
+    },
+    {
+      "epoch": 11.02348623853211,
+      "grad_norm": 0.06398274004459381,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0023,
+      "step": 1274
+    },
+    {
+      "epoch": 11.023853211009174,
+      "grad_norm": 17.040502548217773,
+      "learning_rate": 2.5e-05,
+      "loss": 0.733,
+      "step": 1275
+    },
+    {
+      "epoch": 11.02422018348624,
+      "grad_norm": 8.624760627746582,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5697,
+      "step": 1276
+    },
+    {
+      "epoch": 11.024587155963303,
+      "grad_norm": 0.13655667006969452,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0047,
+      "step": 1277
+    },
+    {
+      "epoch": 11.024954128440367,
+      "grad_norm": 0.049438733607530594,
+      "learning_rate": 2.5e-05,
+      "loss": 0.002,
+      "step": 1278
+    },
+    {
+      "epoch": 11.02532110091743,
+      "grad_norm": 0.11114528775215149,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0033,
+      "step": 1279
+    },
+    {
+      "epoch": 11.025688073394495,
+      "grad_norm": 0.5085391998291016,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0133,
+      "step": 1280
+    },
+    {
+      "epoch": 11.02605504587156,
+      "grad_norm": 0.1276930570602417,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0032,
+      "step": 1281
+    },
+    {
+      "epoch": 11.026422018348624,
+      "grad_norm": 0.08746213465929031,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0021,
+      "step": 1282
+    },
+    {
+      "epoch": 11.026788990825688,
+      "grad_norm": 0.010403720661997795,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0005,
+      "step": 1283
+    },
+    {
+      "epoch": 11.027155963302752,
+      "grad_norm": 0.06900820136070251,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0023,
+      "step": 1284
+    },
+    {
+      "epoch": 11.027522935779816,
+      "grad_norm": 37.267066955566406,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3008,
+      "step": 1285
+    },
+    {
+      "epoch": 11.027889908256881,
+      "grad_norm": 1.7005400657653809,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0042,
+      "step": 1286
+    },
+    {
+      "epoch": 11.028256880733945,
+      "grad_norm": 0.0621204636991024,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0026,
+      "step": 1287
+    },
+    {
+      "epoch": 11.028623853211009,
+      "grad_norm": 0.05243374779820442,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0019,
+      "step": 1288
+    },
+    {
+      "epoch": 11.028990825688073,
+      "grad_norm": 0.35005536675453186,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0043,
+      "step": 1289
+    },
+    {
+      "epoch": 11.029357798165137,
+      "grad_norm": 0.04229266941547394,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0018,
+      "step": 1290
+    },
+    {
+      "epoch": 11.029724770642202,
+      "grad_norm": 0.03818093612790108,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0014,
+      "step": 1291
+    },
+    {
+      "epoch": 11.030091743119266,
+      "grad_norm": 0.10495713353157043,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0029,
+      "step": 1292
+    },
+    {
+      "epoch": 11.03045871559633,
+      "grad_norm": 4.099051475524902,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5966,
+      "step": 1293
+    },
+    {
+      "epoch": 11.030825688073394,
+      "grad_norm": 0.049248624593019485,
+      "learning_rate": 2.5e-05,
+      "loss": 0.002,
+      "step": 1294
+    },
+    {
+      "epoch": 11.03119266055046,
+      "grad_norm": 0.07060599327087402,
+      "learning_rate": 2.5e-05,
+      "loss": 0.002,
+      "step": 1295
+    },
+    {
+      "epoch": 11.031559633027523,
+      "grad_norm": 1.1693766117095947,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0189,
+      "step": 1296
+    },
+    {
+      "epoch": 11.031926605504587,
+      "grad_norm": 17.03798484802246,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1523,
+      "step": 1297
+    },
+    {
+      "epoch": 11.032293577981651,
+      "grad_norm": 0.0490080900490284,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0016,
+      "step": 1298
+    },
+    {
+      "epoch": 11.032660550458715,
+      "grad_norm": 0.003829145571216941,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0001,
+      "step": 1299
+    },
+    {
+      "epoch": 11.03302752293578,
+      "grad_norm": 0.08711742609739304,
+      "learning_rate": 2.5e-05,
+      "loss": 0.004,
+      "step": 1300
+    },
+    {
+      "epoch": 11.033394495412844,
+      "grad_norm": 32.67654800415039,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3939,
+      "step": 1301
+    },
+    {
+      "epoch": 11.033761467889908,
+      "grad_norm": 0.16142717003822327,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0041,
+      "step": 1302
+    },
+    {
+      "epoch": 11.034128440366972,
+      "grad_norm": 0.03583717718720436,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0015,
+      "step": 1303
+    },
+    {
+      "epoch": 11.034495412844036,
+      "grad_norm": 0.024432364851236343,
+      "learning_rate": 2.5e-05,
+      "loss": 0.001,
+      "step": 1304
+    },
+    {
+      "epoch": 11.034862385321102,
+      "grad_norm": 3.8268659114837646,
+      "learning_rate": 2.5e-05,
+      "loss": 0.7207,
+      "step": 1305
+    },
+    {
+      "epoch": 11.035229357798165,
+      "grad_norm": 0.046738654375076294,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0015,
+      "step": 1306
+    },
+    {
+      "epoch": 11.03559633027523,
+      "grad_norm": 41.7634391784668,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1651,
+      "step": 1307
+    },
+    {
+      "epoch": 11.035963302752293,
+      "grad_norm": 0.020139550790190697,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0008,
+      "step": 1308
+    },
+    {
+      "epoch": 11.036330275229357,
+      "grad_norm": 0.08432772010564804,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0034,
+      "step": 1309
+    },
+    {
+      "epoch": 11.036697247706423,
+      "grad_norm": 0.7795783281326294,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0097,
+      "step": 1310
+    },
+    {
+      "epoch": 11.037064220183487,
+      "grad_norm": 10.122318267822266,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5012,
+      "step": 1311
+    },
+    {
+      "epoch": 11.03743119266055,
+      "grad_norm": 0.05448160320520401,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0022,
+      "step": 1312
+    },
+    {
+      "epoch": 11.037798165137614,
+      "grad_norm": 0.16079476475715637,
+      "learning_rate": 2.5e-05,
+      "loss": 0.005,
+      "step": 1313
+    },
+    {
+      "epoch": 11.038165137614678,
+      "grad_norm": 5.449344158172607,
+      "learning_rate": 2.5e-05,
+      "loss": 0.243,
+      "step": 1314
+    },
+    {
+      "epoch": 11.038532110091744,
+      "grad_norm": 0.10348948836326599,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0033,
+      "step": 1315
+    },
+    {
+      "epoch": 11.038899082568808,
+      "grad_norm": 0.09308164566755295,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0037,
+      "step": 1316
+    },
+    {
+      "epoch": 11.039266055045871,
+      "grad_norm": 0.20637206733226776,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0046,
+      "step": 1317
+    },
+    {
+      "epoch": 11.039633027522935,
+      "grad_norm": 0.22749949991703033,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0069,
+      "step": 1318
+    },
+    {
+      "epoch": 11.04,
+      "grad_norm": 0.4177914559841156,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0088,
+      "step": 1319
+    },
+    {
+      "epoch": 11.040366972477065,
+      "grad_norm": 31.309267044067383,
+      "learning_rate": 2.5e-05,
+      "loss": 0.43,
+      "step": 1320
+    },
+    {
+      "epoch": 11.040366972477065,
+      "eval_accuracy": 0.5207100591715976,
+      "eval_loss": 3.7030115127563477,
+      "eval_runtime": 26.0308,
+      "eval_samples_per_second": 12.985,
+      "eval_steps_per_second": 1.652,
+      "step": 1320
+    },
+    {
+      "epoch": 12.000366972477064,
+      "grad_norm": 0.9132550358772278,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0169,
+      "step": 1321
+    },
+    {
+      "epoch": 12.000733944954128,
+      "grad_norm": 0.07931016385555267,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0032,
+      "step": 1322
+    },
+    {
+      "epoch": 12.001100917431193,
+      "grad_norm": 0.2787172496318817,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0051,
+      "step": 1323
+    },
+    {
+      "epoch": 12.001467889908257,
+      "grad_norm": 0.17501048743724823,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0042,
+      "step": 1324
+    },
+    {
+      "epoch": 12.001834862385321,
+      "grad_norm": 0.22494573891162872,
+      "learning_rate": 1.25e-05,
+      "loss": 0.005,
+      "step": 1325
+    },
+    {
+      "epoch": 12.002201834862385,
+      "grad_norm": 0.06690148264169693,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0025,
+      "step": 1326
+    },
+    {
+      "epoch": 12.002568807339449,
+      "grad_norm": 0.10540340095758438,
+      "learning_rate": 1.25e-05,
+      "loss": 0.004,
+      "step": 1327
+    },
+    {
+      "epoch": 12.002935779816514,
+      "grad_norm": 1.202372431755066,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0245,
+      "step": 1328
+    },
+    {
+      "epoch": 12.003302752293578,
+      "grad_norm": 0.35300490260124207,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0065,
+      "step": 1329
+    },
+    {
+      "epoch": 12.003669724770642,
+      "grad_norm": 7.72408390045166,
+      "learning_rate": 1.25e-05,
+      "loss": 0.6703,
+      "step": 1330
+    },
+    {
+      "epoch": 12.004036697247706,
+      "grad_norm": 0.4010007083415985,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0076,
+      "step": 1331
+    },
+    {
+      "epoch": 12.00440366972477,
+      "grad_norm": 0.4644002318382263,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0071,
+      "step": 1332
+    },
+    {
+      "epoch": 12.004770642201835,
+      "grad_norm": 0.047453299164772034,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0017,
+      "step": 1333
+    },
+    {
+      "epoch": 12.0051376146789,
+      "grad_norm": 0.04737183451652527,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0019,
+      "step": 1334
+    },
+    {
+      "epoch": 12.005504587155963,
+      "grad_norm": 0.1000843495130539,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0031,
+      "step": 1335
+    },
+    {
+      "epoch": 12.005871559633027,
+      "grad_norm": 2.53250789642334,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0251,
+      "step": 1336
+    },
+    {
+      "epoch": 12.006238532110093,
+      "grad_norm": 0.07497711479663849,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0024,
+      "step": 1337
+    },
+    {
+      "epoch": 12.006605504587156,
+      "grad_norm": 0.017917612567543983,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0007,
+      "step": 1338
+    },
+    {
+      "epoch": 12.00697247706422,
+      "grad_norm": 3.9414963722229004,
+      "learning_rate": 1.25e-05,
+      "loss": 0.2549,
+      "step": 1339
+    },
+    {
+      "epoch": 12.007339449541284,
+      "grad_norm": 0.01251649297773838,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0005,
+      "step": 1340
+    },
+    {
+      "epoch": 12.007706422018348,
+      "grad_norm": 0.025953229516744614,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1341
+    },
+    {
+      "epoch": 12.008073394495414,
+      "grad_norm": 9.355483055114746,
+      "learning_rate": 1.25e-05,
+      "loss": 0.1583,
+      "step": 1342
+    },
+    {
+      "epoch": 12.008440366972478,
+      "grad_norm": 0.026226725429296494,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1343
+    },
+    {
+      "epoch": 12.008807339449541,
+      "grad_norm": 0.09587549418210983,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0019,
+      "step": 1344
+    },
+    {
+      "epoch": 12.009174311926605,
+      "grad_norm": 1.5006253719329834,
+      "learning_rate": 1.25e-05,
+      "loss": 0.007,
+      "step": 1345
+    },
+    {
+      "epoch": 12.009541284403669,
+      "grad_norm": 0.02712566778063774,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0012,
+      "step": 1346
+    },
+    {
+      "epoch": 12.009908256880735,
+      "grad_norm": 0.05939484387636185,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0023,
+      "step": 1347
+    },
+    {
+      "epoch": 12.010275229357799,
+      "grad_norm": 5.957181930541992,
+      "learning_rate": 1.25e-05,
+      "loss": 0.3697,
+      "step": 1348
+    },
+    {
+      "epoch": 12.010642201834862,
+      "grad_norm": 0.033889979124069214,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0015,
+      "step": 1349
+    },
+    {
+      "epoch": 12.011009174311926,
+      "grad_norm": 0.020186180248856544,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0007,
+      "step": 1350
+    },
+    {
+      "epoch": 12.01137614678899,
+      "grad_norm": 0.035561446100473404,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0015,
+      "step": 1351
+    },
+    {
+      "epoch": 12.011743119266056,
+      "grad_norm": 6.390066146850586,
+      "learning_rate": 1.25e-05,
+      "loss": 0.6462,
+      "step": 1352
+    },
+    {
+      "epoch": 12.01211009174312,
+      "grad_norm": 0.039618488401174545,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0011,
+      "step": 1353
+    },
+    {
+      "epoch": 12.012477064220183,
+      "grad_norm": 0.01931338757276535,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0008,
+      "step": 1354
+    },
+    {
+      "epoch": 12.012844036697247,
+      "grad_norm": 7.081968307495117,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0313,
+      "step": 1355
+    },
+    {
+      "epoch": 12.013211009174311,
+      "grad_norm": 0.014032003469765186,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0007,
+      "step": 1356
+    },
+    {
+      "epoch": 12.013577981651377,
+      "grad_norm": 0.02231457084417343,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1357
+    },
+    {
+      "epoch": 12.01394495412844,
+      "grad_norm": 0.2036891132593155,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0025,
+      "step": 1358
+    },
+    {
+      "epoch": 12.014311926605505,
+      "grad_norm": 0.027025874704122543,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0012,
+      "step": 1359
+    },
+    {
+      "epoch": 12.014678899082568,
+      "grad_norm": 21.988309860229492,
+      "learning_rate": 1.25e-05,
+      "loss": 0.4815,
+      "step": 1360
+    },
+    {
+      "epoch": 12.015045871559632,
+      "grad_norm": 0.023468362167477608,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1361
+    },
+    {
+      "epoch": 12.015412844036698,
+      "grad_norm": 0.03599802032113075,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0013,
+      "step": 1362
+    },
+    {
+      "epoch": 12.015779816513762,
+      "grad_norm": 1.0096933841705322,
+      "learning_rate": 1.25e-05,
+      "loss": 0.02,
+      "step": 1363
+    },
+    {
+      "epoch": 12.016146788990826,
+      "grad_norm": 13.640289306640625,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0497,
+      "step": 1364
+    },
+    {
+      "epoch": 12.01651376146789,
+      "grad_norm": 5.3907904624938965,
+      "learning_rate": 1.25e-05,
+      "loss": 0.6725,
+      "step": 1365
+    },
+    {
+      "epoch": 12.016880733944953,
+      "grad_norm": 3.9498846530914307,
+      "learning_rate": 1.25e-05,
+      "loss": 0.6231,
+      "step": 1366
+    },
+    {
+      "epoch": 12.017247706422019,
+      "grad_norm": 0.04329457879066467,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0018,
+      "step": 1367
+    },
+    {
+      "epoch": 12.017614678899083,
+      "grad_norm": 0.03312234953045845,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0014,
+      "step": 1368
+    },
+    {
+      "epoch": 12.017981651376147,
+      "grad_norm": 0.04505382478237152,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0018,
+      "step": 1369
+    },
+    {
+      "epoch": 12.01834862385321,
+      "grad_norm": 0.10698065161705017,
+      "learning_rate": 1.25e-05,
+      "loss": 0.003,
+      "step": 1370
+    },
+    {
+      "epoch": 12.018715596330276,
+      "grad_norm": 3.1098878383636475,
+      "learning_rate": 1.25e-05,
+      "loss": 0.8015,
+      "step": 1371
+    },
+    {
+      "epoch": 12.01908256880734,
+      "grad_norm": 0.023850951343774796,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1372
+    },
+    {
+      "epoch": 12.019449541284404,
+      "grad_norm": 0.09161610901355743,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0031,
+      "step": 1373
+    },
+    {
+      "epoch": 12.019816513761468,
+      "grad_norm": 3.158853530883789,
+      "learning_rate": 1.25e-05,
+      "loss": 0.6109,
+      "step": 1374
+    },
+    {
+      "epoch": 12.020183486238532,
+      "grad_norm": 0.09031084179878235,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0026,
+      "step": 1375
+    },
+    {
+      "epoch": 12.020550458715597,
+      "grad_norm": 0.07501880824565887,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0028,
+      "step": 1376
+    },
+    {
+      "epoch": 12.020917431192661,
+      "grad_norm": 0.027117209509015083,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0011,
+      "step": 1377
+    },
+    {
+      "epoch": 12.021284403669725,
+      "grad_norm": 0.09164215624332428,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0042,
+      "step": 1378
+    },
+    {
+      "epoch": 12.021651376146789,
+      "grad_norm": 6.086216449737549,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0297,
+      "step": 1379
+    },
+    {
+      "epoch": 12.022018348623853,
+      "grad_norm": 0.044853683561086655,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0017,
+      "step": 1380
+    },
+    {
+      "epoch": 12.022385321100918,
+      "grad_norm": 0.07254932820796967,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0029,
+      "step": 1381
+    },
+    {
+      "epoch": 12.022752293577982,
+      "grad_norm": 0.07669806480407715,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0029,
+      "step": 1382
+    },
+    {
+      "epoch": 12.023119266055046,
+      "grad_norm": 4.88906192779541,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0477,
+      "step": 1383
+    },
+    {
+      "epoch": 12.02348623853211,
+      "grad_norm": 0.05569980666041374,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0023,
+      "step": 1384
+    },
+    {
+      "epoch": 12.023853211009174,
+      "grad_norm": 9.062528610229492,
+      "learning_rate": 1.25e-05,
+      "loss": 0.4475,
+      "step": 1385
+    },
+    {
+      "epoch": 12.02422018348624,
+      "grad_norm": 0.06240231916308403,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0025,
+      "step": 1386
+    },
+    {
+      "epoch": 12.024587155963303,
+      "grad_norm": 0.07493241131305695,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0027,
+      "step": 1387
+    },
+    {
+      "epoch": 12.024954128440367,
+      "grad_norm": 0.0645071342587471,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0025,
+      "step": 1388
+    },
+    {
+      "epoch": 12.02532110091743,
+      "grad_norm": 7.570181846618652,
+      "learning_rate": 1.25e-05,
+      "loss": 0.4758,
+      "step": 1389
+    },
+    {
+      "epoch": 12.025688073394495,
+      "grad_norm": 0.058653607964515686,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0022,
+      "step": 1390
+    },
+    {
+      "epoch": 12.02605504587156,
+      "grad_norm": 0.08166255056858063,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0027,
+      "step": 1391
+    },
+    {
+      "epoch": 12.026422018348624,
+      "grad_norm": 11.00561809539795,
+      "learning_rate": 1.25e-05,
+      "loss": 0.3142,
+      "step": 1392
+    },
+    {
+      "epoch": 12.026788990825688,
+      "grad_norm": 0.05225876346230507,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0018,
+      "step": 1393
+    },
+    {
+      "epoch": 12.027155963302752,
+      "grad_norm": 4.933167457580566,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0574,
+      "step": 1394
+    },
+    {
+      "epoch": 12.027522935779816,
+      "grad_norm": 0.0769711285829544,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0025,
+      "step": 1395
+    },
+    {
+      "epoch": 12.027889908256881,
+      "grad_norm": 0.08121397346258163,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0033,
+      "step": 1396
+    },
+    {
+      "epoch": 12.028256880733945,
+      "grad_norm": 0.10960618406534195,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0042,
+      "step": 1397
+    },
+    {
+      "epoch": 12.028623853211009,
+      "grad_norm": 0.10151275247335434,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0035,
+      "step": 1398
+    },
+    {
+      "epoch": 12.028990825688073,
+      "grad_norm": 0.06118513271212578,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0022,
+      "step": 1399
+    },
+    {
+      "epoch": 12.029357798165137,
+      "grad_norm": 3.9730257987976074,
+      "learning_rate": 1.25e-05,
+      "loss": 0.1339,
+      "step": 1400
+    },
+    {
+      "epoch": 12.029724770642202,
+      "grad_norm": 0.20571595430374146,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0106,
+      "step": 1401
+    },
+    {
+      "epoch": 12.030091743119266,
+      "grad_norm": 0.23795028030872345,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0064,
+      "step": 1402
+    },
+    {
+      "epoch": 12.03045871559633,
+      "grad_norm": 0.014628360979259014,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0005,
+      "step": 1403
+    },
+    {
+      "epoch": 12.030825688073394,
+      "grad_norm": 0.027092652395367622,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0012,
+      "step": 1404
+    },
+    {
+      "epoch": 12.03119266055046,
+      "grad_norm": 2.102180242538452,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0238,
+      "step": 1405
+    },
+    {
+      "epoch": 12.031559633027523,
+      "grad_norm": 0.11743936687707901,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0031,
+      "step": 1406
+    },
+    {
+      "epoch": 12.031926605504587,
+      "grad_norm": 0.25206032395362854,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0103,
+      "step": 1407
+    },
+    {
+      "epoch": 12.032293577981651,
+      "grad_norm": 0.9687058925628662,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0184,
+      "step": 1408
+    },
+    {
+      "epoch": 12.032660550458715,
+      "grad_norm": 0.08827485889196396,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0034,
+      "step": 1409
+    },
+    {
+      "epoch": 12.03302752293578,
+      "grad_norm": 0.04332936927676201,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0018,
+      "step": 1410
+    },
+    {
+      "epoch": 12.033394495412844,
+      "grad_norm": 0.08677459508180618,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0032,
+      "step": 1411
+    },
+    {
+      "epoch": 12.033761467889908,
+      "grad_norm": 20.019454956054688,
+      "learning_rate": 1.25e-05,
+      "loss": 0.7313,
+      "step": 1412
+    },
+    {
+      "epoch": 12.034128440366972,
+      "grad_norm": 0.04310310631990433,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0014,
+      "step": 1413
+    },
+    {
+      "epoch": 12.034495412844036,
+      "grad_norm": 0.01615392416715622,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0007,
+      "step": 1414
+    },
+    {
+      "epoch": 12.034862385321102,
+      "grad_norm": 0.07221368700265884,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0025,
+      "step": 1415
+    },
+    {
+      "epoch": 12.035229357798165,
+      "grad_norm": 0.06981567293405533,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0031,
+      "step": 1416
+    },
+    {
+      "epoch": 12.03559633027523,
+      "grad_norm": 0.08338870108127594,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0031,
+      "step": 1417
+    },
+    {
+      "epoch": 12.035963302752293,
+      "grad_norm": 5.179531097412109,
+      "learning_rate": 1.25e-05,
+      "loss": 0.5468,
+      "step": 1418
+    },
+    {
+      "epoch": 12.036330275229357,
+      "grad_norm": 0.33509141206741333,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0037,
+      "step": 1419
+    },
+    {
+      "epoch": 12.036697247706423,
+      "grad_norm": 0.06253895163536072,
+      "learning_rate": 1.25e-05,
+      "loss": 0.002,
+      "step": 1420
+    },
+    {
+      "epoch": 12.037064220183487,
+      "grad_norm": 0.01912231557071209,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0008,
+      "step": 1421
+    },
+    {
+      "epoch": 12.03743119266055,
+      "grad_norm": 0.02545015700161457,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0011,
+      "step": 1422
+    },
+    {
+      "epoch": 12.037798165137614,
+      "grad_norm": 0.0938255786895752,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0029,
+      "step": 1423
+    },
+    {
+      "epoch": 12.038165137614678,
+      "grad_norm": 32.43787384033203,
+      "learning_rate": 1.25e-05,
+      "loss": 0.3269,
+      "step": 1424
+    },
+    {
+      "epoch": 12.038532110091744,
+      "grad_norm": 0.0437430813908577,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0016,
+      "step": 1425
+    },
+    {
+      "epoch": 12.038899082568808,
+      "grad_norm": 0.018908463418483734,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0007,
+      "step": 1426
+    },
+    {
+      "epoch": 12.039266055045871,
+      "grad_norm": 1.821930170059204,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0243,
+      "step": 1427
+    },
+    {
+      "epoch": 12.039633027522935,
+      "grad_norm": 0.043903157114982605,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0018,
+      "step": 1428
+    },
+    {
+      "epoch": 12.04,
+      "grad_norm": 4.110330581665039,
+      "learning_rate": 1.25e-05,
+      "loss": 0.593,
+      "step": 1429
+    },
+    {
+      "epoch": 12.040366972477065,
+      "grad_norm": 0.03760877251625061,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0014,
+      "step": 1430
+    },
+    {
+      "epoch": 12.040366972477065,
+      "eval_accuracy": 0.20118343195266272,
+      "eval_loss": 2.5479140281677246,
+      "eval_runtime": 25.8263,
+      "eval_samples_per_second": 13.087,
+      "eval_steps_per_second": 1.665,
+      "step": 1430
+    },
+    {
+      "epoch": 13.000366972477064,
+      "grad_norm": 2.3445279598236084,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0255,
+      "step": 1431
+    },
+    {
+      "epoch": 13.000733944954128,
+      "grad_norm": 0.07456192374229431,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0022,
+      "step": 1432
+    },
+    {
+      "epoch": 13.001100917431193,
+      "grad_norm": 0.049425944685935974,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0025,
+      "step": 1433
+    },
+    {
+      "epoch": 13.001467889908257,
+      "grad_norm": 0.30483201146125793,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0062,
+      "step": 1434
+    },
+    {
+      "epoch": 13.001834862385321,
+      "grad_norm": 1.3933831453323364,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0091,
+      "step": 1435
+    },
+    {
+      "epoch": 13.002201834862385,
+      "grad_norm": 0.060145534574985504,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0026,
+      "step": 1436
+    },
+    {
+      "epoch": 13.002568807339449,
+      "grad_norm": 0.025344733148813248,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1437
+    },
+    {
+      "epoch": 13.002935779816514,
+      "grad_norm": 0.8993895649909973,
+      "learning_rate": 1.25e-05,
+      "loss": 0.01,
+      "step": 1438
+    },
+    {
+      "epoch": 13.003302752293578,
+      "grad_norm": 0.03253284469246864,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0012,
+      "step": 1439
+    },
+    {
+      "epoch": 13.003669724770642,
+      "grad_norm": 0.13995416462421417,
+      "learning_rate": 1.25e-05,
+      "loss": 0.005,
+      "step": 1440
+    },
+    {
+      "epoch": 13.004036697247706,
+      "grad_norm": 5.188272476196289,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0307,
+      "step": 1441
+    },
+    {
+      "epoch": 13.00440366972477,
+      "grad_norm": 0.040029432624578476,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0017,
+      "step": 1442
+    },
+    {
+      "epoch": 13.004770642201835,
+      "grad_norm": 3.945310354232788,
+      "learning_rate": 1.25e-05,
+      "loss": 0.4687,
+      "step": 1443
+    },
+    {
+      "epoch": 13.0051376146789,
+      "grad_norm": 0.0805070698261261,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0029,
+      "step": 1444
+    },
+    {
+      "epoch": 13.005504587155963,
+      "grad_norm": 0.09031485766172409,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0032,
+      "step": 1445
+    },
+    {
+      "epoch": 13.005871559633027,
+      "grad_norm": 0.15356998145580292,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0021,
+      "step": 1446
+    },
+    {
+      "epoch": 13.006238532110093,
+      "grad_norm": 0.04531683400273323,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0019,
+      "step": 1447
+    },
+    {
+      "epoch": 13.006605504587156,
+      "grad_norm": 0.1082788035273552,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0029,
+      "step": 1448
+    },
+    {
+      "epoch": 13.00697247706422,
+      "grad_norm": 0.0885109230875969,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0043,
+      "step": 1449
+    },
+    {
+      "epoch": 13.007339449541284,
+      "grad_norm": 0.06119617447257042,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0022,
+      "step": 1450
+    },
+    {
+      "epoch": 13.007706422018348,
+      "grad_norm": 4.910311222076416,
+      "learning_rate": 1.25e-05,
+      "loss": 0.1462,
+      "step": 1451
+    },
+    {
+      "epoch": 13.008073394495414,
+      "grad_norm": 0.04216789826750755,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0013,
+      "step": 1452
+    },
+    {
+      "epoch": 13.008440366972478,
+      "grad_norm": 0.02794824168086052,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0012,
+      "step": 1453
+    },
+    {
+      "epoch": 13.008807339449541,
+      "grad_norm": 0.8641343116760254,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0119,
+      "step": 1454
+    },
+    {
+      "epoch": 13.009174311926605,
+      "grad_norm": 0.046625252813100815,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0018,
+      "step": 1455
+    },
+    {
+      "epoch": 13.009541284403669,
+      "grad_norm": 0.053742412477731705,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0022,
+      "step": 1456
+    },
+    {
+      "epoch": 13.009908256880735,
+      "grad_norm": 0.012518118135631084,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0005,
+      "step": 1457
+    },
+    {
+      "epoch": 13.010275229357799,
+      "grad_norm": 0.14832867681980133,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0045,
+      "step": 1458
+    },
+    {
+      "epoch": 13.010642201834862,
+      "grad_norm": 12.569035530090332,
+      "learning_rate": 1.25e-05,
+      "loss": 1.1552,
+      "step": 1459
+    },
+    {
+      "epoch": 13.011009174311926,
+      "grad_norm": 3.764392614364624,
+      "learning_rate": 1.25e-05,
+      "loss": 0.066,
+      "step": 1460
+    },
+    {
+      "epoch": 13.01137614678899,
+      "grad_norm": 0.0698130875825882,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0025,
+      "step": 1461
+    },
+    {
+      "epoch": 13.011743119266056,
+      "grad_norm": 0.03251948952674866,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0012,
+      "step": 1462
+    },
+    {
+      "epoch": 13.01211009174312,
+      "grad_norm": 0.526522696018219,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0135,
+      "step": 1463
+    },
+    {
+      "epoch": 13.012477064220183,
+      "grad_norm": 0.02679644338786602,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0012,
+      "step": 1464
+    },
+    {
+      "epoch": 13.012844036697247,
+      "grad_norm": 0.08167339861392975,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0022,
+      "step": 1465
+    },
+    {
+      "epoch": 13.013211009174311,
+      "grad_norm": 0.2470615953207016,
+      "learning_rate": 1.25e-05,
+      "loss": 0.01,
+      "step": 1466
+    },
+    {
+      "epoch": 13.013577981651377,
+      "grad_norm": 0.04037018120288849,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0015,
+      "step": 1467
+    },
+    {
+      "epoch": 13.01394495412844,
+      "grad_norm": 0.040535520762205124,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0015,
+      "step": 1468
+    },
+    {
+      "epoch": 13.014311926605505,
+      "grad_norm": 0.20592370629310608,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0044,
+      "step": 1469
+    },
+    {
+      "epoch": 13.014678899082568,
+      "grad_norm": 0.07645940780639648,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0019,
+      "step": 1470
+    },
+    {
+      "epoch": 13.015045871559632,
+      "grad_norm": 0.10262466967105865,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0057,
+      "step": 1471
+    },
+    {
+      "epoch": 13.015412844036698,
+      "grad_norm": 0.041500553488731384,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0016,
+      "step": 1472
+    },
+    {
+      "epoch": 13.015779816513762,
+      "grad_norm": 3.8759958744049072,
+      "learning_rate": 1.25e-05,
+      "loss": 0.4652,
+      "step": 1473
+    },
+    {
+      "epoch": 13.016146788990826,
+      "grad_norm": 0.5208672285079956,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0053,
+      "step": 1474
+    },
+    {
+      "epoch": 13.01651376146789,
+      "grad_norm": 0.11522316932678223,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0049,
+      "step": 1475
+    },
+    {
+      "epoch": 13.016880733944953,
+      "grad_norm": 0.0837157741189003,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0034,
+      "step": 1476
+    },
+    {
+      "epoch": 13.017247706422019,
+      "grad_norm": 0.06705445051193237,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0024,
+      "step": 1477
+    },
+    {
+      "epoch": 13.017614678899083,
+      "grad_norm": 0.038701269775629044,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0015,
+      "step": 1478
+    },
+    {
+      "epoch": 13.017981651376147,
+      "grad_norm": 0.046367544680833817,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0018,
+      "step": 1479
+    },
+    {
+      "epoch": 13.01834862385321,
+      "grad_norm": 0.03586876392364502,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0014,
+      "step": 1480
+    },
+    {
+      "epoch": 13.018715596330276,
+      "grad_norm": 5.215917110443115,
+      "learning_rate": 1.25e-05,
+      "loss": 0.5254,
+      "step": 1481
+    },
+    {
+      "epoch": 13.01908256880734,
+      "grad_norm": 0.0562765896320343,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0026,
+      "step": 1482
+    },
+    {
+      "epoch": 13.019449541284404,
+      "grad_norm": 2.776832342147827,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0105,
+      "step": 1483
+    },
+    {
+      "epoch": 13.019816513761468,
+      "grad_norm": 0.035216450691223145,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0013,
+      "step": 1484
+    },
+    {
+      "epoch": 13.020183486238532,
+      "grad_norm": 0.023580001667141914,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1485
+    },
+    {
+      "epoch": 13.020550458715597,
+      "grad_norm": 0.04757922515273094,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0017,
+      "step": 1486
+    },
+    {
+      "epoch": 13.020917431192661,
+      "grad_norm": 0.04171590134501457,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0015,
+      "step": 1487
+    },
+    {
+      "epoch": 13.021284403669725,
+      "grad_norm": 0.010839978232979774,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0005,
+      "step": 1488
+    },
+    {
+      "epoch": 13.021651376146789,
+      "grad_norm": 0.03581145405769348,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0007,
+      "step": 1489
+    },
+    {
+      "epoch": 13.022018348623853,
+      "grad_norm": 3.2871663570404053,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0248,
+      "step": 1490
+    },
+    {
+      "epoch": 13.022385321100918,
+      "grad_norm": 0.05126366764307022,
+      "learning_rate": 1.25e-05,
+      "loss": 0.002,
+      "step": 1491
+    },
+    {
+      "epoch": 13.022752293577982,
+      "grad_norm": 0.033572860062122345,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0012,
+      "step": 1492
+    },
+    {
+      "epoch": 13.023119266055046,
+      "grad_norm": 0.17035745084285736,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0037,
+      "step": 1493
+    },
+    {
+      "epoch": 13.02348623853211,
+      "grad_norm": 0.07110364735126495,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0025,
+      "step": 1494
+    },
+    {
+      "epoch": 13.023853211009174,
+      "grad_norm": 0.045812830328941345,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0017,
+      "step": 1495
+    },
+    {
+      "epoch": 13.02422018348624,
+      "grad_norm": 0.046104662120342255,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0018,
+      "step": 1496
+    },
+    {
+      "epoch": 13.024587155963303,
+      "grad_norm": 0.028975287452340126,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1497
+    },
+    {
+      "epoch": 13.024954128440367,
+      "grad_norm": 18.330602645874023,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0787,
+      "step": 1498
+    },
+    {
+      "epoch": 13.02532110091743,
+      "grad_norm": 0.05071160942316055,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0019,
+      "step": 1499
+    },
+    {
+      "epoch": 13.025688073394495,
+      "grad_norm": 0.2136876881122589,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0031,
+      "step": 1500
+    },
+    {
+      "epoch": 13.02605504587156,
+      "grad_norm": 0.0658545047044754,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0022,
+      "step": 1501
+    },
+    {
+      "epoch": 13.026422018348624,
+      "grad_norm": 0.29163017868995667,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0065,
+      "step": 1502
+    },
+    {
+      "epoch": 13.026788990825688,
+      "grad_norm": 0.15547268092632294,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0016,
+      "step": 1503
+    },
+    {
+      "epoch": 13.027155963302752,
+      "grad_norm": 0.07763663679361343,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0022,
+      "step": 1504
+    },
+    {
+      "epoch": 13.027522935779816,
+      "grad_norm": 0.04284100607037544,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0016,
+      "step": 1505
+    },
+    {
+      "epoch": 13.027889908256881,
+      "grad_norm": 0.11197897791862488,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0024,
+      "step": 1506
+    },
+    {
+      "epoch": 13.028256880733945,
+      "grad_norm": 0.09812965244054794,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0038,
+      "step": 1507
+    },
+    {
+      "epoch": 13.028623853211009,
+      "grad_norm": 0.03799255192279816,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0012,
+      "step": 1508
+    },
+    {
+      "epoch": 13.028990825688073,
+      "grad_norm": 2.0954830646514893,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0153,
+      "step": 1509
+    },
+    {
+      "epoch": 13.029357798165137,
+      "grad_norm": 12.482344627380371,
+      "learning_rate": 1.25e-05,
+      "loss": 0.5413,
+      "step": 1510
+    },
+    {
+      "epoch": 13.029724770642202,
+      "grad_norm": 0.019370924681425095,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0006,
+      "step": 1511
+    },
+    {
+      "epoch": 13.030091743119266,
+      "grad_norm": 0.32100096344947815,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0059,
+      "step": 1512
+    },
+    {
+      "epoch": 13.03045871559633,
+      "grad_norm": 0.03706175461411476,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0012,
+      "step": 1513
+    },
+    {
+      "epoch": 13.030825688073394,
+      "grad_norm": 0.04302698373794556,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0013,
+      "step": 1514
+    },
+    {
+      "epoch": 13.03119266055046,
+      "grad_norm": 0.020381907001137733,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0007,
+      "step": 1515
+    },
+    {
+      "epoch": 13.031559633027523,
+      "grad_norm": 0.005204925313591957,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0002,
+      "step": 1516
+    },
+    {
+      "epoch": 13.031926605504587,
+      "grad_norm": 0.019191399216651917,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0008,
+      "step": 1517
+    },
+    {
+      "epoch": 13.032293577981651,
+      "grad_norm": 0.07631263881921768,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0036,
+      "step": 1518
+    },
+    {
+      "epoch": 13.032660550458715,
+      "grad_norm": 0.05902083218097687,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0026,
+      "step": 1519
+    },
+    {
+      "epoch": 13.03302752293578,
+      "grad_norm": 0.4866631329059601,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0057,
+      "step": 1520
+    },
+    {
+      "epoch": 13.033394495412844,
+      "grad_norm": 0.04412684217095375,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0014,
+      "step": 1521
+    },
+    {
+      "epoch": 13.033761467889908,
+      "grad_norm": 2.679103136062622,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0232,
+      "step": 1522
+    },
+    {
+      "epoch": 13.034128440366972,
+      "grad_norm": 0.019163664430379868,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0008,
+      "step": 1523
+    },
+    {
+      "epoch": 13.034495412844036,
+      "grad_norm": 0.046621840447187424,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0017,
+      "step": 1524
+    },
+    {
+      "epoch": 13.034862385321102,
+      "grad_norm": 0.01070187333971262,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0005,
+      "step": 1525
+    },
+    {
+      "epoch": 13.035229357798165,
+      "grad_norm": 0.03362787514925003,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0012,
+      "step": 1526
+    },
+    {
+      "epoch": 13.03559633027523,
+      "grad_norm": 7.501707553863525,
+      "learning_rate": 1.25e-05,
+      "loss": 0.6223,
+      "step": 1527
+    },
+    {
+      "epoch": 13.035963302752293,
+      "grad_norm": 0.04660226032137871,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0017,
+      "step": 1528
+    },
+    {
+      "epoch": 13.036330275229357,
+      "grad_norm": 0.014180006459355354,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0006,
+      "step": 1529
+    },
+    {
+      "epoch": 13.036697247706423,
+      "grad_norm": 0.02152341976761818,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0007,
+      "step": 1530
+    },
+    {
+      "epoch": 13.037064220183487,
+      "grad_norm": 0.021999606862664223,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0008,
+      "step": 1531
+    },
+    {
+      "epoch": 13.03743119266055,
+      "grad_norm": 6.747068881988525,
+      "learning_rate": 1.25e-05,
+      "loss": 0.3213,
+      "step": 1532
+    },
+    {
+      "epoch": 13.037798165137614,
+      "grad_norm": 0.41011932492256165,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0087,
+      "step": 1533
+    },
+    {
+      "epoch": 13.038165137614678,
+      "grad_norm": 6.906198978424072,
+      "learning_rate": 1.25e-05,
+      "loss": 0.2488,
+      "step": 1534
+    },
+    {
+      "epoch": 13.038532110091744,
+      "grad_norm": 0.014847236685454845,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0007,
+      "step": 1535
+    },
+    {
+      "epoch": 13.038899082568808,
+      "grad_norm": 0.04215698689222336,
+      "learning_rate": 1.25e-05,
+      "loss": 0.002,
+      "step": 1536
+    },
+    {
+      "epoch": 13.039266055045871,
+      "grad_norm": 0.04047771543264389,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0014,
+      "step": 1537
+    },
+    {
+      "epoch": 13.039633027522935,
+      "grad_norm": 0.028728945180773735,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1538
+    },
+    {
+      "epoch": 13.04,
+      "grad_norm": 0.014704103581607342,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0006,
+      "step": 1539
+    },
+    {
+      "epoch": 13.040366972477065,
+      "grad_norm": 0.046350739896297455,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0021,
+      "step": 1540
+    },
+    {
+      "epoch": 13.040366972477065,
+      "eval_accuracy": 0.31952662721893493,
+      "eval_loss": 4.023542404174805,
+      "eval_runtime": 25.8288,
+      "eval_samples_per_second": 13.086,
+      "eval_steps_per_second": 1.665,
+      "step": 1540
+    },
+    {
+      "epoch": 14.000366972477064,
+      "grad_norm": 8.725737571716309,
+      "learning_rate": 1.25e-05,
+      "loss": 0.2431,
+      "step": 1541
+    },
+    {
+      "epoch": 14.000733944954128,
+      "grad_norm": 0.05007888004183769,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0023,
+      "step": 1542
+    },
+    {
+      "epoch": 14.001100917431193,
+      "grad_norm": 0.04863353446125984,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0014,
+      "step": 1543
+    },
+    {
+      "epoch": 14.001467889908257,
+      "grad_norm": 8.646441459655762,
+      "learning_rate": 1.25e-05,
+      "loss": 0.5961,
+      "step": 1544
+    },
+    {
+      "epoch": 14.001834862385321,
+      "grad_norm": 0.03763831779360771,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0018,
+      "step": 1545
+    },
+    {
+      "epoch": 14.002201834862385,
+      "grad_norm": 0.045137807726860046,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0016,
+      "step": 1546
+    },
+    {
+      "epoch": 14.002568807339449,
+      "grad_norm": 4.151777744293213,
+      "learning_rate": 1.25e-05,
+      "loss": 0.7937,
+      "step": 1547
+    },
+    {
+      "epoch": 14.002935779816514,
+      "grad_norm": 0.7858043909072876,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0219,
+      "step": 1548
+    },
+    {
+      "epoch": 14.003302752293578,
+      "grad_norm": 0.025060102343559265,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0008,
+      "step": 1549
+    },
+    {
+      "epoch": 14.003669724770642,
+      "grad_norm": 0.032970547676086426,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0016,
+      "step": 1550
+    },
+    {
+      "epoch": 14.004036697247706,
+      "grad_norm": 0.14705513417720795,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0039,
+      "step": 1551
+    },
+    {
+      "epoch": 14.00440366972477,
+      "grad_norm": 0.13219617307186127,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0039,
+      "step": 1552
+    },
+    {
+      "epoch": 14.004770642201835,
+      "grad_norm": 0.09093134105205536,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0034,
+      "step": 1553
+    },
+    {
+      "epoch": 14.0051376146789,
+      "grad_norm": 0.023366471752524376,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1554
+    },
+    {
+      "epoch": 14.005504587155963,
+      "grad_norm": 0.05955779179930687,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0019,
+      "step": 1555
+    },
+    {
+      "epoch": 14.005871559633027,
+      "grad_norm": 4.9263715744018555,
+      "learning_rate": 1.25e-05,
+      "loss": 0.3104,
+      "step": 1556
+    },
+    {
+      "epoch": 14.006238532110093,
+      "grad_norm": 0.10129030793905258,
+      "learning_rate": 1.25e-05,
+      "loss": 0.003,
+      "step": 1557
+    },
+    {
+      "epoch": 14.006605504587156,
+      "grad_norm": 0.2976308763027191,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0076,
+      "step": 1558
+    },
+    {
+      "epoch": 14.00697247706422,
+      "grad_norm": 0.007806284818798304,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0003,
+      "step": 1559
+    },
+    {
+      "epoch": 14.007339449541284,
+      "grad_norm": 0.03895338997244835,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0015,
+      "step": 1560
+    },
+    {
+      "epoch": 14.007706422018348,
+      "grad_norm": 0.01371223758906126,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0006,
+      "step": 1561
+    },
+    {
+      "epoch": 14.008073394495414,
+      "grad_norm": 0.01384967751801014,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0005,
+      "step": 1562
+    },
+    {
+      "epoch": 14.008440366972478,
+      "grad_norm": 0.02927769348025322,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1563
+    },
+    {
+      "epoch": 14.008807339449541,
+      "grad_norm": 0.07105225324630737,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0026,
+      "step": 1564
+    },
+    {
+      "epoch": 14.009174311926605,
+      "grad_norm": 0.033063165843486786,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0013,
+      "step": 1565
+    },
+    {
+      "epoch": 14.009541284403669,
+      "grad_norm": 0.03970335051417351,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0011,
+      "step": 1566
+    },
+    {
+      "epoch": 14.009908256880735,
+      "grad_norm": 0.0457913801074028,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0016,
+      "step": 1567
+    },
+    {
+      "epoch": 14.010275229357799,
+      "grad_norm": 0.05784710869193077,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0019,
+      "step": 1568
+    },
+    {
+      "epoch": 14.010642201834862,
+      "grad_norm": 35.38191223144531,
+      "learning_rate": 1.25e-05,
+      "loss": 0.4114,
+      "step": 1569
+    },
+    {
+      "epoch": 14.011009174311926,
+      "grad_norm": 0.022911233827471733,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0009,
+      "step": 1570
+    },
+    {
+      "epoch": 14.01137614678899,
+      "grad_norm": 0.05216791480779648,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0018,
+      "step": 1571
+    },
+    {
+      "epoch": 14.011743119266056,
+      "grad_norm": 0.037898238748311996,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0014,
+      "step": 1572
+    },
+    {
+      "epoch": 14.01211009174312,
+      "grad_norm": 0.5702415108680725,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0058,
+      "step": 1573
+    },
+    {
+      "epoch": 14.012477064220183,
+      "grad_norm": 0.42801639437675476,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0131,
+      "step": 1574
+    },
+    {
+      "epoch": 14.012844036697247,
+      "grad_norm": 0.06548340618610382,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0026,
+      "step": 1575
+    },
+    {
+      "epoch": 14.013211009174311,
+      "grad_norm": 0.025748850777745247,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1576
+    },
+    {
+      "epoch": 14.013577981651377,
+      "grad_norm": 0.02946171537041664,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0011,
+      "step": 1577
+    },
+    {
+      "epoch": 14.01394495412844,
+      "grad_norm": 0.04795168340206146,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0018,
+      "step": 1578
+    },
+    {
+      "epoch": 14.014311926605505,
+      "grad_norm": 0.30734822154045105,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0075,
+      "step": 1579
+    },
+    {
+      "epoch": 14.014678899082568,
+      "grad_norm": 0.01483188197016716,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0006,
+      "step": 1580
+    },
+    {
+      "epoch": 14.015045871559632,
+      "grad_norm": 0.03651377186179161,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0012,
+      "step": 1581
+    },
+    {
+      "epoch": 14.015412844036698,
+      "grad_norm": 0.01302403025329113,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0006,
+      "step": 1582
+    },
+    {
+      "epoch": 14.015779816513762,
+      "grad_norm": 41.251956939697266,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0593,
+      "step": 1583
+    },
+    {
+      "epoch": 14.016146788990826,
+      "grad_norm": 0.029474478214979172,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0012,
+      "step": 1584
+    },
+    {
+      "epoch": 14.01651376146789,
+      "grad_norm": 11.29187297821045,
+      "learning_rate": 1.25e-05,
+      "loss": 0.4693,
+      "step": 1585
+    },
+    {
+      "epoch": 14.016880733944953,
+      "grad_norm": 0.014432638883590698,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0006,
+      "step": 1586
+    },
+    {
+      "epoch": 14.017247706422019,
+      "grad_norm": 0.007910864427685738,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0004,
+      "step": 1587
+    },
+    {
+      "epoch": 14.017614678899083,
+      "grad_norm": 0.010554708540439606,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0005,
+      "step": 1588
+    },
+    {
+      "epoch": 14.017981651376147,
+      "grad_norm": 0.06227845326066017,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0022,
+      "step": 1589
+    },
+    {
+      "epoch": 14.01834862385321,
+      "grad_norm": 0.01888139359652996,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0007,
+      "step": 1590
+    },
+    {
+      "epoch": 14.018715596330276,
+      "grad_norm": 0.03110986202955246,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0011,
+      "step": 1591
+    },
+    {
+      "epoch": 14.01908256880734,
+      "grad_norm": 0.050834015011787415,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0017,
+      "step": 1592
+    },
+    {
+      "epoch": 14.019449541284404,
+      "grad_norm": 0.01743358187377453,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0007,
+      "step": 1593
+    },
+    {
+      "epoch": 14.019816513761468,
+      "grad_norm": 0.02412516064941883,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0008,
+      "step": 1594
+    },
+    {
+      "epoch": 14.020183486238532,
+      "grad_norm": 0.005441767629235983,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0002,
+      "step": 1595
+    },
+    {
+      "epoch": 14.020550458715597,
+      "grad_norm": 0.03822169452905655,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0013,
+      "step": 1596
+    },
+    {
+      "epoch": 14.020917431192661,
+      "grad_norm": 0.04043407738208771,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0012,
+      "step": 1597
+    },
+    {
+      "epoch": 14.021284403669725,
+      "grad_norm": 0.03905504196882248,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0014,
+      "step": 1598
+    },
+    {
+      "epoch": 14.021651376146789,
+      "grad_norm": 0.011848600581288338,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0005,
+      "step": 1599
+    },
+    {
+      "epoch": 14.022018348623853,
+      "grad_norm": 7.142123699188232,
+      "learning_rate": 1.25e-05,
+      "loss": 0.5574,
+      "step": 1600
+    },
+    {
+      "epoch": 14.022385321100918,
+      "grad_norm": 0.018463920801877975,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0007,
+      "step": 1601
+    },
+    {
+      "epoch": 14.022752293577982,
+      "grad_norm": 4.96084451675415,
+      "learning_rate": 1.25e-05,
+      "loss": 0.6916,
+      "step": 1602
+    },
+    {
+      "epoch": 14.023119266055046,
+      "grad_norm": 0.08990047872066498,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0023,
+      "step": 1603
+    },
+    {
+      "epoch": 14.02348623853211,
+      "grad_norm": 0.04759575054049492,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0017,
+      "step": 1604
+    },
+    {
+      "epoch": 14.023853211009174,
+      "grad_norm": 0.03422639146447182,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0014,
+      "step": 1605
+    },
+    {
+      "epoch": 14.02422018348624,
+      "grad_norm": 0.04419027268886566,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0014,
+      "step": 1606
+    },
+    {
+      "epoch": 14.024587155963303,
+      "grad_norm": 1.0535907745361328,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0074,
+      "step": 1607
+    },
+    {
+      "epoch": 14.024954128440367,
+      "grad_norm": 0.02783922292292118,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0009,
+      "step": 1608
+    },
+    {
+      "epoch": 14.02532110091743,
+      "grad_norm": 0.053749747574329376,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0029,
+      "step": 1609
+    },
+    {
+      "epoch": 14.025688073394495,
+      "grad_norm": 0.02862151712179184,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0012,
+      "step": 1610
+    },
+    {
+      "epoch": 14.02605504587156,
+      "grad_norm": 0.11665903776884079,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0044,
+      "step": 1611
+    },
+    {
+      "epoch": 14.026422018348624,
+      "grad_norm": 0.047582387924194336,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0017,
+      "step": 1612
+    },
+    {
+      "epoch": 14.026788990825688,
+      "grad_norm": 0.08930886536836624,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0025,
+      "step": 1613
+    },
+    {
+      "epoch": 14.027155963302752,
+      "grad_norm": 0.03473014011979103,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0009,
+      "step": 1614
+    },
+    {
+      "epoch": 14.027522935779816,
+      "grad_norm": 7.086911678314209,
+      "learning_rate": 1.25e-05,
+      "loss": 0.3994,
+      "step": 1615
+    },
+    {
+      "epoch": 14.027889908256881,
+      "grad_norm": 0.08884729444980621,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0028,
+      "step": 1616
+    },
+    {
+      "epoch": 14.028256880733945,
+      "grad_norm": 29.62557029724121,
+      "learning_rate": 1.25e-05,
+      "loss": 0.4454,
+      "step": 1617
+    },
+    {
+      "epoch": 14.028623853211009,
+      "grad_norm": 25.21485710144043,
+      "learning_rate": 1.25e-05,
+      "loss": 0.2325,
+      "step": 1618
+    },
+    {
+      "epoch": 14.028990825688073,
+      "grad_norm": 7.948330402374268,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0633,
+      "step": 1619
+    },
+    {
+      "epoch": 14.029357798165137,
+      "grad_norm": 9.7194242477417,
+      "learning_rate": 1.25e-05,
+      "loss": 0.4908,
+      "step": 1620
+    },
+    {
+      "epoch": 14.029724770642202,
+      "grad_norm": 0.24098999798297882,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0024,
+      "step": 1621
+    },
+    {
+      "epoch": 14.030091743119266,
+      "grad_norm": 19.98224449157715,
+      "learning_rate": 1.25e-05,
+      "loss": 0.4099,
+      "step": 1622
+    },
+    {
+      "epoch": 14.03045871559633,
+      "grad_norm": 0.46823570132255554,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0093,
+      "step": 1623
+    },
+    {
+      "epoch": 14.030825688073394,
+      "grad_norm": 0.03914785012602806,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0015,
+      "step": 1624
+    },
+    {
+      "epoch": 14.03119266055046,
+      "grad_norm": 0.09906239807605743,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0022,
+      "step": 1625
+    },
+    {
+      "epoch": 14.031559633027523,
+      "grad_norm": 0.1299586445093155,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0032,
+      "step": 1626
+    },
+    {
+      "epoch": 14.031926605504587,
+      "grad_norm": 0.12302394956350327,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0053,
+      "step": 1627
+    },
+    {
+      "epoch": 14.032293577981651,
+      "grad_norm": 0.016616681590676308,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0006,
+      "step": 1628
+    },
+    {
+      "epoch": 14.032660550458715,
+      "grad_norm": 0.1232086718082428,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0044,
+      "step": 1629
+    },
+    {
+      "epoch": 14.03302752293578,
+      "grad_norm": 0.0435197688639164,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0013,
+      "step": 1630
+    },
+    {
+      "epoch": 14.033394495412844,
+      "grad_norm": 0.12653155624866486,
+      "learning_rate": 1.25e-05,
+      "loss": 0.003,
+      "step": 1631
+    },
+    {
+      "epoch": 14.033761467889908,
+      "grad_norm": 0.13882018625736237,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0037,
+      "step": 1632
+    },
+    {
+      "epoch": 14.034128440366972,
+      "grad_norm": 0.045017603784799576,
+      "learning_rate": 1.25e-05,
+      "loss": 0.002,
+      "step": 1633
+    },
+    {
+      "epoch": 14.034495412844036,
+      "grad_norm": 0.08790068328380585,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0024,
+      "step": 1634
+    },
+    {
+      "epoch": 14.034862385321102,
+      "grad_norm": 0.01710103638470173,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0007,
+      "step": 1635
+    },
+    {
+      "epoch": 14.035229357798165,
+      "grad_norm": 3.546862840652466,
+      "learning_rate": 1.25e-05,
+      "loss": 0.5291,
+      "step": 1636
+    },
+    {
+      "epoch": 14.03559633027523,
+      "grad_norm": 0.038590770214796066,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0016,
+      "step": 1637
+    },
+    {
+      "epoch": 14.035963302752293,
+      "grad_norm": 0.02735825814306736,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1638
+    },
+    {
+      "epoch": 14.036330275229357,
+      "grad_norm": 0.04304742068052292,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1639
+    },
+    {
+      "epoch": 14.036697247706423,
+      "grad_norm": 0.12633277475833893,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0031,
+      "step": 1640
+    },
+    {
+      "epoch": 14.037064220183487,
+      "grad_norm": 5.850214004516602,
+      "learning_rate": 1.25e-05,
+      "loss": 0.4881,
+      "step": 1641
+    },
+    {
+      "epoch": 14.03743119266055,
+      "grad_norm": 0.051536548882722855,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0019,
+      "step": 1642
+    },
+    {
+      "epoch": 14.037798165137614,
+      "grad_norm": 0.26878494024276733,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0055,
+      "step": 1643
+    },
+    {
+      "epoch": 14.038165137614678,
+      "grad_norm": 0.09685373306274414,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0024,
+      "step": 1644
+    },
+    {
+      "epoch": 14.038532110091744,
+      "grad_norm": 1.9561368227005005,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0164,
+      "step": 1645
+    },
+    {
+      "epoch": 14.038899082568808,
+      "grad_norm": 0.02416694350540638,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1646
+    },
+    {
+      "epoch": 14.039266055045871,
+      "grad_norm": 0.13695059716701508,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0028,
+      "step": 1647
+    },
+    {
+      "epoch": 14.039633027522935,
+      "grad_norm": 0.037553418427705765,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0009,
+      "step": 1648
+    },
+    {
+      "epoch": 14.04,
+      "grad_norm": 0.08390378206968307,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0016,
+      "step": 1649
+    },
+    {
+      "epoch": 14.040366972477065,
+      "grad_norm": 37.212921142578125,
+      "learning_rate": 1.25e-05,
+      "loss": 0.2387,
+      "step": 1650
+    },
+    {
+      "epoch": 14.040366972477065,
+      "eval_accuracy": 0.23372781065088757,
+      "eval_loss": 4.604861259460449,
+      "eval_runtime": 25.9474,
+      "eval_samples_per_second": 13.026,
+      "eval_steps_per_second": 1.657,
+      "step": 1650
+    },
+    {
+      "epoch": 15.000366972477064,
+      "grad_norm": 0.04876839742064476,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0014,
+      "step": 1651
+    },
+    {
+      "epoch": 15.000733944954128,
+      "grad_norm": 0.05413774773478508,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0023,
+      "step": 1652
+    },
+    {
+      "epoch": 15.001100917431193,
+      "grad_norm": 0.2062683403491974,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0042,
+      "step": 1653
+    },
+    {
+      "epoch": 15.001467889908257,
+      "grad_norm": 0.08180626481771469,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0024,
+      "step": 1654
+    },
+    {
+      "epoch": 15.001834862385321,
+      "grad_norm": 0.05859791859984398,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0018,
+      "step": 1655
+    },
+    {
+      "epoch": 15.002201834862385,
+      "grad_norm": 1.017958164215088,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0164,
+      "step": 1656
+    },
+    {
+      "epoch": 15.002568807339449,
+      "grad_norm": 0.056351758539676666,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1657
+    },
+    {
+      "epoch": 15.002935779816514,
+      "grad_norm": 0.11059648543596268,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0039,
+      "step": 1658
+    },
+    {
+      "epoch": 15.003302752293578,
+      "grad_norm": 0.03513418138027191,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0008,
+      "step": 1659
+    },
+    {
+      "epoch": 15.003669724770642,
+      "grad_norm": 0.04445274919271469,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0013,
+      "step": 1660
+    },
+    {
+      "epoch": 15.004036697247706,
+      "grad_norm": 0.030361881479620934,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1661
+    },
+    {
+      "epoch": 15.00440366972477,
+      "grad_norm": 0.006681793835014105,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0003,
+      "step": 1662
+    },
+    {
+      "epoch": 15.004770642201835,
+      "grad_norm": 0.005998845212161541,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0003,
+      "step": 1663
+    },
+    {
+      "epoch": 15.0051376146789,
+      "grad_norm": 6.005819797515869,
+      "learning_rate": 1.25e-05,
+      "loss": 0.6991,
+      "step": 1664
+    },
+    {
+      "epoch": 15.005504587155963,
+      "grad_norm": 0.22628216445446014,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0053,
+      "step": 1665
+    },
+    {
+      "epoch": 15.005871559633027,
+      "grad_norm": 0.0831294059753418,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0028,
+      "step": 1666
+    },
+    {
+      "epoch": 15.006238532110093,
+      "grad_norm": 0.15850016474723816,
+      "learning_rate": 1.25e-05,
+      "loss": 0.003,
+      "step": 1667
+    },
+    {
+      "epoch": 15.006605504587156,
+      "grad_norm": 0.05295209586620331,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0015,
+      "step": 1668
+    },
+    {
+      "epoch": 15.00697247706422,
+      "grad_norm": 0.06930996477603912,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0013,
+      "step": 1669
+    },
+    {
+      "epoch": 15.007339449541284,
+      "grad_norm": 0.029055561870336533,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0011,
+      "step": 1670
+    },
+    {
+      "epoch": 15.007706422018348,
+      "grad_norm": 12.384570121765137,
+      "learning_rate": 1.25e-05,
+      "loss": 0.5192,
+      "step": 1671
+    },
+    {
+      "epoch": 15.008073394495414,
+      "grad_norm": 0.10765182226896286,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0029,
+      "step": 1672
+    },
+    {
+      "epoch": 15.008440366972478,
+      "grad_norm": 0.009517967700958252,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0004,
+      "step": 1673
+    },
+    {
+      "epoch": 15.008807339449541,
+      "grad_norm": 0.3372949957847595,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0045,
+      "step": 1674
+    },
+    {
+      "epoch": 15.009174311926605,
+      "grad_norm": 7.646303653717041,
+      "learning_rate": 1.25e-05,
+      "loss": 0.6101,
+      "step": 1675
+    },
+    {
+      "epoch": 15.009541284403669,
+      "grad_norm": 11.322826385498047,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0395,
+      "step": 1676
+    },
+    {
+      "epoch": 15.009908256880735,
+      "grad_norm": 0.06631970405578613,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0023,
+      "step": 1677
+    },
+    {
+      "epoch": 15.010275229357799,
+      "grad_norm": 0.06431559473276138,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0018,
+      "step": 1678
+    },
+    {
+      "epoch": 15.010642201834862,
+      "grad_norm": 0.04949197173118591,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0015,
+      "step": 1679
+    },
+    {
+      "epoch": 15.011009174311926,
+      "grad_norm": 0.012884916737675667,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0004,
+      "step": 1680
+    },
+    {
+      "epoch": 15.01137614678899,
+      "grad_norm": 5.443044662475586,
+      "learning_rate": 1.25e-05,
+      "loss": 0.5234,
+      "step": 1681
+    },
+    {
+      "epoch": 15.011743119266056,
+      "grad_norm": 9.226449012756348,
+      "learning_rate": 1.25e-05,
+      "loss": 1.204,
+      "step": 1682
+    },
+    {
+      "epoch": 15.01211009174312,
+      "grad_norm": 0.032926756888628006,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0008,
+      "step": 1683
+    },
+    {
+      "epoch": 15.012477064220183,
+      "grad_norm": 0.04610403999686241,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0016,
+      "step": 1684
+    },
+    {
+      "epoch": 15.012844036697247,
+      "grad_norm": 0.038929108530282974,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0017,
+      "step": 1685
+    },
+    {
+      "epoch": 15.013211009174311,
+      "grad_norm": 10.646738052368164,
+      "learning_rate": 1.25e-05,
+      "loss": 0.4167,
+      "step": 1686
+    },
+    {
+      "epoch": 15.013577981651377,
+      "grad_norm": 0.1464022845029831,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0049,
+      "step": 1687
+    },
+    {
+      "epoch": 15.01394495412844,
+      "grad_norm": 0.157272607088089,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0041,
+      "step": 1688
+    },
+    {
+      "epoch": 15.014311926605505,
+      "grad_norm": 0.2646397352218628,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0074,
+      "step": 1689
+    },
+    {
+      "epoch": 15.014678899082568,
+      "grad_norm": 7.77543830871582,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0791,
+      "step": 1690
+    },
+    {
+      "epoch": 15.015045871559632,
+      "grad_norm": 0.16761259734630585,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0052,
+      "step": 1691
+    },
+    {
+      "epoch": 15.015412844036698,
+      "grad_norm": 0.13270984590053558,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0041,
+      "step": 1692
+    },
+    {
+      "epoch": 15.015779816513762,
+      "grad_norm": 0.009066293947398663,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0003,
+      "step": 1693
+    },
+    {
+      "epoch": 15.016146788990826,
+      "grad_norm": 4.346078872680664,
+      "learning_rate": 1.25e-05,
+      "loss": 0.2096,
+      "step": 1694
+    },
+    {
+      "epoch": 15.01651376146789,
+      "grad_norm": 52.216983795166016,
+      "learning_rate": 1.25e-05,
+      "loss": 0.1166,
+      "step": 1695
+    },
+    {
+      "epoch": 15.016880733944953,
+      "grad_norm": 0.11610086262226105,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0031,
+      "step": 1696
+    },
+    {
+      "epoch": 15.017247706422019,
+      "grad_norm": 12.310649871826172,
+      "learning_rate": 1.25e-05,
+      "loss": 0.4254,
+      "step": 1697
+    },
+    {
+      "epoch": 15.017614678899083,
+      "grad_norm": 0.3123236298561096,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0045,
+      "step": 1698
+    },
+    {
+      "epoch": 15.017981651376147,
+      "grad_norm": 0.4322940707206726,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0076,
+      "step": 1699
+    },
+    {
+      "epoch": 15.01834862385321,
+      "grad_norm": 4.794229507446289,
+      "learning_rate": 1.25e-05,
+      "loss": 0.1177,
+      "step": 1700
+    },
+    {
+      "epoch": 15.018715596330276,
+      "grad_norm": 0.03519390523433685,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0008,
+      "step": 1701
+    },
+    {
+      "epoch": 15.01908256880734,
+      "grad_norm": 0.30617380142211914,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0052,
+      "step": 1702
+    },
+    {
+      "epoch": 15.019449541284404,
+      "grad_norm": 0.37854841351509094,
+      "learning_rate": 1.25e-05,
+      "loss": 0.006,
+      "step": 1703
+    },
+    {
+      "epoch": 15.019816513761468,
+      "grad_norm": 70.38347625732422,
+      "learning_rate": 1.25e-05,
+      "loss": 0.6283,
+      "step": 1704
+    },
+    {
+      "epoch": 15.020183486238532,
+      "grad_norm": 0.0575457438826561,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1705
+    },
+    {
+      "epoch": 15.020550458715597,
+      "grad_norm": 0.3456067144870758,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0041,
+      "step": 1706
+    },
+    {
+      "epoch": 15.020917431192661,
+      "grad_norm": 0.02876913733780384,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0012,
+      "step": 1707
+    },
+    {
+      "epoch": 15.021284403669725,
+      "grad_norm": 0.6320503354072571,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0071,
+      "step": 1708
+    },
+    {
+      "epoch": 15.021651376146789,
+      "grad_norm": 0.050501104444265366,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0016,
+      "step": 1709
+    },
+    {
+      "epoch": 15.022018348623853,
+      "grad_norm": 0.07113325595855713,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0018,
+      "step": 1710
+    },
+    {
+      "epoch": 15.022385321100918,
+      "grad_norm": 0.21116791665554047,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0042,
+      "step": 1711
+    },
+    {
+      "epoch": 15.022752293577982,
+      "grad_norm": 0.04356582462787628,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0014,
+      "step": 1712
+    },
+    {
+      "epoch": 15.023119266055046,
+      "grad_norm": 0.09786807000637054,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0024,
+      "step": 1713
+    },
+    {
+      "epoch": 15.02348623853211,
+      "grad_norm": 0.03016163408756256,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0011,
+      "step": 1714
+    },
+    {
+      "epoch": 15.023853211009174,
+      "grad_norm": 0.19040200114250183,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0041,
+      "step": 1715
+    },
+    {
+      "epoch": 15.02422018348624,
+      "grad_norm": 7.791591644287109,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0216,
+      "step": 1716
+    },
+    {
+      "epoch": 15.024587155963303,
+      "grad_norm": 0.121918685734272,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0021,
+      "step": 1717
+    },
+    {
+      "epoch": 15.024954128440367,
+      "grad_norm": 0.043672580271959305,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0013,
+      "step": 1718
+    },
+    {
+      "epoch": 15.02532110091743,
+      "grad_norm": 0.016822347417473793,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0007,
+      "step": 1719
+    },
+    {
+      "epoch": 15.025688073394495,
+      "grad_norm": 0.015447739511728287,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0006,
+      "step": 1720
+    },
+    {
+      "epoch": 15.02605504587156,
+      "grad_norm": 0.013137048110365868,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0005,
+      "step": 1721
+    },
+    {
+      "epoch": 15.026422018348624,
+      "grad_norm": 0.049040812999010086,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0018,
+      "step": 1722
+    },
+    {
+      "epoch": 15.026788990825688,
+      "grad_norm": 0.0286239180713892,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0008,
+      "step": 1723
+    },
+    {
+      "epoch": 15.027155963302752,
+      "grad_norm": 0.07446285337209702,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0017,
+      "step": 1724
+    },
+    {
+      "epoch": 15.027522935779816,
+      "grad_norm": 1.6416287422180176,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0057,
+      "step": 1725
+    },
+    {
+      "epoch": 15.027889908256881,
+      "grad_norm": 0.010669986717402935,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0005,
+      "step": 1726
+    },
+    {
+      "epoch": 15.028256880733945,
+      "grad_norm": 0.02003258280456066,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0006,
+      "step": 1727
+    },
+    {
+      "epoch": 15.028623853211009,
+      "grad_norm": 0.0687771663069725,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0019,
+      "step": 1728
+    },
+    {
+      "epoch": 15.028990825688073,
+      "grad_norm": 0.48718634247779846,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0027,
+      "step": 1729
+    },
+    {
+      "epoch": 15.029357798165137,
+      "grad_norm": 0.008882123045623302,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0004,
+      "step": 1730
+    },
+    {
+      "epoch": 15.029724770642202,
+      "grad_norm": 0.010564956814050674,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0003,
+      "step": 1731
+    },
+    {
+      "epoch": 15.030091743119266,
+      "grad_norm": 0.013637790456414223,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0005,
+      "step": 1732
+    },
+    {
+      "epoch": 15.03045871559633,
+      "grad_norm": 0.025979723781347275,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0009,
+      "step": 1733
+    },
+    {
+      "epoch": 15.030825688073394,
+      "grad_norm": 0.2182309478521347,
+      "learning_rate": 1.25e-05,
+      "loss": 0.006,
+      "step": 1734
+    },
+    {
+      "epoch": 15.03119266055046,
+      "grad_norm": 0.12560486793518066,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0023,
+      "step": 1735
+    },
+    {
+      "epoch": 15.031559633027523,
+      "grad_norm": 0.11862605065107346,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0026,
+      "step": 1736
+    },
+    {
+      "epoch": 15.031926605504587,
+      "grad_norm": 0.6057714819908142,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0081,
+      "step": 1737
+    },
+    {
+      "epoch": 15.032293577981651,
+      "grad_norm": 5.944339752197266,
+      "learning_rate": 1.25e-05,
+      "loss": 0.6378,
+      "step": 1738
+    },
+    {
+      "epoch": 15.032660550458715,
+      "grad_norm": 0.026245983317494392,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1739
+    },
+    {
+      "epoch": 15.03302752293578,
+      "grad_norm": 0.009206948801875114,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0004,
+      "step": 1740
+    },
+    {
+      "epoch": 15.033394495412844,
+      "grad_norm": 0.01043816190212965,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0005,
+      "step": 1741
+    },
+    {
+      "epoch": 15.033761467889908,
+      "grad_norm": 0.005614196881651878,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0003,
+      "step": 1742
+    },
+    {
+      "epoch": 15.034128440366972,
+      "grad_norm": 0.02476939745247364,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0011,
+      "step": 1743
+    },
+    {
+      "epoch": 15.034495412844036,
+      "grad_norm": 0.2144959419965744,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0039,
+      "step": 1744
+    },
+    {
+      "epoch": 15.034862385321102,
+      "grad_norm": 0.005240797530859709,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0002,
+      "step": 1745
+    },
+    {
+      "epoch": 15.035229357798165,
+      "grad_norm": 5.427676677703857,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0194,
+      "step": 1746
+    },
+    {
+      "epoch": 15.03559633027523,
+      "grad_norm": 12.132808685302734,
+      "learning_rate": 1.25e-05,
+      "loss": 0.5683,
+      "step": 1747
+    },
+    {
+      "epoch": 15.035963302752293,
+      "grad_norm": 0.01633720099925995,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0007,
+      "step": 1748
+    },
+    {
+      "epoch": 15.036330275229357,
+      "grad_norm": 0.04882543534040451,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0014,
+      "step": 1749
+    },
+    {
+      "epoch": 15.036697247706423,
+      "grad_norm": 0.009626666083931923,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0004,
+      "step": 1750
+    },
+    {
+      "epoch": 15.037064220183487,
+      "grad_norm": 0.044971004128456116,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0017,
+      "step": 1751
+    },
+    {
+      "epoch": 15.03743119266055,
+      "grad_norm": 0.07679823040962219,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0025,
+      "step": 1752
+    },
+    {
+      "epoch": 15.037798165137614,
+      "grad_norm": 0.02204957604408264,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0007,
+      "step": 1753
+    },
+    {
+      "epoch": 15.038165137614678,
+      "grad_norm": 0.4750242531299591,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0081,
+      "step": 1754
+    },
+    {
+      "epoch": 15.038532110091744,
+      "grad_norm": 0.016196759417653084,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0007,
+      "step": 1755
+    },
+    {
+      "epoch": 15.038899082568808,
+      "grad_norm": 0.027464808896183968,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0008,
+      "step": 1756
+    },
+    {
+      "epoch": 15.039266055045871,
+      "grad_norm": 0.014484725892543793,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0005,
+      "step": 1757
+    },
+    {
+      "epoch": 15.039633027522935,
+      "grad_norm": 5.9926958084106445,
+      "learning_rate": 1.25e-05,
+      "loss": 0.4185,
+      "step": 1758
+    },
+    {
+      "epoch": 15.04,
+      "grad_norm": 0.02854829840362072,
+      "learning_rate": 1.25e-05,
+      "loss": 0.001,
+      "step": 1759
+    },
+    {
+      "epoch": 15.040366972477065,
+      "grad_norm": 0.03185071423649788,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0009,
+      "step": 1760
+    },
+    {
+      "epoch": 15.040366972477065,
+      "eval_accuracy": 0.2485207100591716,
+      "eval_loss": 4.306981563568115,
+      "eval_runtime": 25.9253,
+      "eval_samples_per_second": 13.037,
+      "eval_steps_per_second": 1.659,
+      "step": 1760
+    },
+    {
+      "epoch": 16.000366972477064,
+      "grad_norm": 0.016486991196870804,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0007,
+      "step": 1761
+    },
+    {
+      "epoch": 16.000733944954128,
+      "grad_norm": 9.085463523864746,
+      "learning_rate": 6.25e-06,
+      "loss": 0.5225,
+      "step": 1762
+    },
+    {
+      "epoch": 16.00110091743119,
+      "grad_norm": 0.03098965622484684,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0011,
+      "step": 1763
+    },
+    {
+      "epoch": 16.001467889908255,
+      "grad_norm": 0.02275032363831997,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0008,
+      "step": 1764
+    },
+    {
+      "epoch": 16.001834862385323,
+      "grad_norm": 0.00906342826783657,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1765
+    },
+    {
+      "epoch": 16.002201834862387,
+      "grad_norm": 0.02275027707219124,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0011,
+      "step": 1766
+    },
+    {
+      "epoch": 16.00256880733945,
+      "grad_norm": 0.01161968894302845,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1767
+    },
+    {
+      "epoch": 16.002935779816514,
+      "grad_norm": 0.24440757930278778,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0055,
+      "step": 1768
+    },
+    {
+      "epoch": 16.00330275229358,
+      "grad_norm": 0.01558439526706934,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 1769
+    },
+    {
+      "epoch": 16.003669724770642,
+      "grad_norm": 0.11280570179224014,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0018,
+      "step": 1770
+    },
+    {
+      "epoch": 16.004036697247706,
+      "grad_norm": 0.028768660500645638,
+      "learning_rate": 6.25e-06,
+      "loss": 0.001,
+      "step": 1771
+    },
+    {
+      "epoch": 16.00440366972477,
+      "grad_norm": 0.021213380619883537,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 1772
+    },
+    {
+      "epoch": 16.004770642201834,
+      "grad_norm": 0.013801740482449532,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 1773
+    },
+    {
+      "epoch": 16.005137614678898,
+      "grad_norm": 0.023930173367261887,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0008,
+      "step": 1774
+    },
+    {
+      "epoch": 16.005504587155965,
+      "grad_norm": 0.09018443524837494,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0021,
+      "step": 1775
+    },
+    {
+      "epoch": 16.00587155963303,
+      "grad_norm": 0.8821747303009033,
+      "learning_rate": 6.25e-06,
+      "loss": 0.006,
+      "step": 1776
+    },
+    {
+      "epoch": 16.006238532110093,
+      "grad_norm": 4.763535022735596,
+      "learning_rate": 6.25e-06,
+      "loss": 0.2343,
+      "step": 1777
+    },
+    {
+      "epoch": 16.006605504587156,
+      "grad_norm": 0.02842329442501068,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0009,
+      "step": 1778
+    },
+    {
+      "epoch": 16.00697247706422,
+      "grad_norm": 0.04870988428592682,
+      "learning_rate": 6.25e-06,
+      "loss": 0.001,
+      "step": 1779
+    },
+    {
+      "epoch": 16.007339449541284,
+      "grad_norm": 5.3503923416137695,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0249,
+      "step": 1780
+    },
+    {
+      "epoch": 16.007706422018348,
+      "grad_norm": 0.11343327909708023,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0016,
+      "step": 1781
+    },
+    {
+      "epoch": 16.008073394495412,
+      "grad_norm": 0.06886930763721466,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0017,
+      "step": 1782
+    },
+    {
+      "epoch": 16.008440366972476,
+      "grad_norm": 0.008610495366156101,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1783
+    },
+    {
+      "epoch": 16.00880733944954,
+      "grad_norm": 12.235538482666016,
+      "learning_rate": 6.25e-06,
+      "loss": 0.411,
+      "step": 1784
+    },
+    {
+      "epoch": 16.009174311926607,
+      "grad_norm": 0.27386656403541565,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0022,
+      "step": 1785
+    },
+    {
+      "epoch": 16.00954128440367,
+      "grad_norm": 0.5720701217651367,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0034,
+      "step": 1786
+    },
+    {
+      "epoch": 16.009908256880735,
+      "grad_norm": 0.05653269588947296,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0018,
+      "step": 1787
+    },
+    {
+      "epoch": 16.0102752293578,
+      "grad_norm": 0.010955655947327614,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 1788
+    },
+    {
+      "epoch": 16.010642201834862,
+      "grad_norm": 0.03426629677414894,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0011,
+      "step": 1789
+    },
+    {
+      "epoch": 16.011009174311926,
+      "grad_norm": 0.19184109568595886,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0027,
+      "step": 1790
+    },
+    {
+      "epoch": 16.01137614678899,
+      "grad_norm": 0.02054670825600624,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0008,
+      "step": 1791
+    },
+    {
+      "epoch": 16.011743119266054,
+      "grad_norm": 0.007963222451508045,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 1792
+    },
+    {
+      "epoch": 16.012110091743118,
+      "grad_norm": 2.4637038707733154,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0232,
+      "step": 1793
+    },
+    {
+      "epoch": 16.012477064220185,
+      "grad_norm": 0.029488926753401756,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0007,
+      "step": 1794
+    },
+    {
+      "epoch": 16.01284403669725,
+      "grad_norm": 0.029634516686201096,
+      "learning_rate": 6.25e-06,
+      "loss": 0.001,
+      "step": 1795
+    },
+    {
+      "epoch": 16.013211009174313,
+      "grad_norm": 77.65727233886719,
+      "learning_rate": 6.25e-06,
+      "loss": 0.135,
+      "step": 1796
+    },
+    {
+      "epoch": 16.013577981651377,
+      "grad_norm": 0.01024566125124693,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1797
+    },
+    {
+      "epoch": 16.01394495412844,
+      "grad_norm": 0.01036208588629961,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1798
+    },
+    {
+      "epoch": 16.014311926605505,
+      "grad_norm": 0.029805362224578857,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0007,
+      "step": 1799
+    },
+    {
+      "epoch": 16.01467889908257,
+      "grad_norm": 0.047902271151542664,
+      "learning_rate": 6.25e-06,
+      "loss": 0.001,
+      "step": 1800
+    },
+    {
+      "epoch": 16.015045871559632,
+      "grad_norm": 0.016275888308882713,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 1801
+    },
+    {
+      "epoch": 16.015412844036696,
+      "grad_norm": 0.3012193739414215,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0049,
+      "step": 1802
+    },
+    {
+      "epoch": 16.01577981651376,
+      "grad_norm": 0.12183599174022675,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0021,
+      "step": 1803
+    },
+    {
+      "epoch": 16.016146788990827,
+      "grad_norm": 0.006971823982894421,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 1804
+    },
+    {
+      "epoch": 16.01651376146789,
+      "grad_norm": 0.014051520265638828,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0007,
+      "step": 1805
+    },
+    {
+      "epoch": 16.016880733944955,
+      "grad_norm": 20.228559494018555,
+      "learning_rate": 6.25e-06,
+      "loss": 0.6784,
+      "step": 1806
+    },
+    {
+      "epoch": 16.01724770642202,
+      "grad_norm": 0.1735542267560959,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0025,
+      "step": 1807
+    },
+    {
+      "epoch": 16.017614678899083,
+      "grad_norm": 0.032639406621456146,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0009,
+      "step": 1808
+    },
+    {
+      "epoch": 16.017981651376147,
+      "grad_norm": 0.01098865456879139,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 1809
+    },
+    {
+      "epoch": 16.01834862385321,
+      "grad_norm": 18.021305084228516,
+      "learning_rate": 6.25e-06,
+      "loss": 0.382,
+      "step": 1810
+    },
+    {
+      "epoch": 16.018715596330274,
+      "grad_norm": 0.013740317896008492,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 1811
+    },
+    {
+      "epoch": 16.019082568807338,
+      "grad_norm": 0.11425668746232986,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0026,
+      "step": 1812
+    },
+    {
+      "epoch": 16.019449541284402,
+      "grad_norm": 0.010876046493649483,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1813
+    },
+    {
+      "epoch": 16.01981651376147,
+      "grad_norm": 0.006368938367813826,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 1814
+    },
+    {
+      "epoch": 16.020183486238533,
+      "grad_norm": 0.032540351152420044,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0011,
+      "step": 1815
+    },
+    {
+      "epoch": 16.020550458715597,
+      "grad_norm": 0.015417372807860374,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 1816
+    },
+    {
+      "epoch": 16.02091743119266,
+      "grad_norm": 0.07452422380447388,
+      "learning_rate": 6.25e-06,
+      "loss": 0.002,
+      "step": 1817
+    },
+    {
+      "epoch": 16.021284403669725,
+      "grad_norm": 0.04269694909453392,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0012,
+      "step": 1818
+    },
+    {
+      "epoch": 16.02165137614679,
+      "grad_norm": 0.004409638699144125,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 1819
+    },
+    {
+      "epoch": 16.022018348623853,
+      "grad_norm": 0.005797105375677347,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 1820
+    },
+    {
+      "epoch": 16.022385321100916,
+      "grad_norm": 0.10794635862112045,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0035,
+      "step": 1821
+    },
+    {
+      "epoch": 16.02275229357798,
+      "grad_norm": 0.011360940523445606,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 1822
+    },
+    {
+      "epoch": 16.023119266055044,
+      "grad_norm": 0.006002004723995924,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 1823
+    },
+    {
+      "epoch": 16.02348623853211,
+      "grad_norm": 0.010427127592265606,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1824
+    },
+    {
+      "epoch": 16.023853211009175,
+      "grad_norm": 0.03412167355418205,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0007,
+      "step": 1825
+    },
+    {
+      "epoch": 16.02422018348624,
+      "grad_norm": 0.009403062053024769,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1826
+    },
+    {
+      "epoch": 16.024587155963303,
+      "grad_norm": 26.155256271362305,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0352,
+      "step": 1827
+    },
+    {
+      "epoch": 16.024954128440367,
+      "grad_norm": 12.478118896484375,
+      "learning_rate": 6.25e-06,
+      "loss": 0.1027,
+      "step": 1828
+    },
+    {
+      "epoch": 16.02532110091743,
+      "grad_norm": 0.003556481096893549,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 1829
+    },
+    {
+      "epoch": 16.025688073394495,
+      "grad_norm": 0.012528013437986374,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 1830
+    },
+    {
+      "epoch": 16.02605504587156,
+      "grad_norm": 11.9004545211792,
+      "learning_rate": 6.25e-06,
+      "loss": 0.5943,
+      "step": 1831
+    },
+    {
+      "epoch": 16.026422018348622,
+      "grad_norm": 0.014376720413565636,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 1832
+    },
+    {
+      "epoch": 16.02678899082569,
+      "grad_norm": 0.00960728619247675,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1833
+    },
+    {
+      "epoch": 16.027155963302754,
+      "grad_norm": 0.006929417606443167,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 1834
+    },
+    {
+      "epoch": 16.027522935779817,
+      "grad_norm": 8.788637161254883,
+      "learning_rate": 6.25e-06,
+      "loss": 0.7863,
+      "step": 1835
+    },
+    {
+      "epoch": 16.02788990825688,
+      "grad_norm": 0.11287093162536621,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0013,
+      "step": 1836
+    },
+    {
+      "epoch": 16.028256880733945,
+      "grad_norm": 0.01415779534727335,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 1837
+    },
+    {
+      "epoch": 16.02862385321101,
+      "grad_norm": 0.17339679598808289,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0027,
+      "step": 1838
+    },
+    {
+      "epoch": 16.028990825688073,
+      "grad_norm": 0.027837960049510002,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0007,
+      "step": 1839
+    },
+    {
+      "epoch": 16.029357798165137,
+      "grad_norm": 0.019132794812321663,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 1840
+    },
+    {
+      "epoch": 16.0297247706422,
+      "grad_norm": 12.72171688079834,
+      "learning_rate": 6.25e-06,
+      "loss": 0.5543,
+      "step": 1841
+    },
+    {
+      "epoch": 16.030091743119264,
+      "grad_norm": 0.07706566154956818,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0029,
+      "step": 1842
+    },
+    {
+      "epoch": 16.030458715596332,
+      "grad_norm": 0.010515131056308746,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1843
+    },
+    {
+      "epoch": 16.030825688073396,
+      "grad_norm": 0.05993041396141052,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0012,
+      "step": 1844
+    },
+    {
+      "epoch": 16.03119266055046,
+      "grad_norm": 0.30817094445228577,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0095,
+      "step": 1845
+    },
+    {
+      "epoch": 16.031559633027523,
+      "grad_norm": 8.6234769821167,
+      "learning_rate": 6.25e-06,
+      "loss": 0.4559,
+      "step": 1846
+    },
+    {
+      "epoch": 16.031926605504587,
+      "grad_norm": 0.01420559175312519,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 1847
+    },
+    {
+      "epoch": 16.03229357798165,
+      "grad_norm": 8.96253490447998,
+      "learning_rate": 6.25e-06,
+      "loss": 0.4554,
+      "step": 1848
+    },
+    {
+      "epoch": 16.032660550458715,
+      "grad_norm": 0.0339556485414505,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0011,
+      "step": 1849
+    },
+    {
+      "epoch": 16.03302752293578,
+      "grad_norm": 0.006203192286193371,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 1850
+    },
+    {
+      "epoch": 16.033394495412843,
+      "grad_norm": 0.009912838228046894,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1851
+    },
+    {
+      "epoch": 16.033761467889907,
+      "grad_norm": 13.58121109008789,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0272,
+      "step": 1852
+    },
+    {
+      "epoch": 16.034128440366974,
+      "grad_norm": 0.1386987864971161,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0021,
+      "step": 1853
+    },
+    {
+      "epoch": 16.034495412844038,
+      "grad_norm": 2.492140531539917,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0218,
+      "step": 1854
+    },
+    {
+      "epoch": 16.0348623853211,
+      "grad_norm": 0.007958345115184784,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 1855
+    },
+    {
+      "epoch": 16.035229357798165,
+      "grad_norm": 0.03618380427360535,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0013,
+      "step": 1856
+    },
+    {
+      "epoch": 16.03559633027523,
+      "grad_norm": 0.009199478663504124,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1857
+    },
+    {
+      "epoch": 16.035963302752293,
+      "grad_norm": 0.035442739725112915,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0008,
+      "step": 1858
+    },
+    {
+      "epoch": 16.036330275229357,
+      "grad_norm": 0.17309559881687164,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0052,
+      "step": 1859
+    },
+    {
+      "epoch": 16.03669724770642,
+      "grad_norm": 0.16191352903842926,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0023,
+      "step": 1860
+    },
+    {
+      "epoch": 16.037064220183485,
+      "grad_norm": 0.034038905054330826,
+      "learning_rate": 6.25e-06,
+      "loss": 0.001,
+      "step": 1861
+    },
+    {
+      "epoch": 16.037431192660552,
+      "grad_norm": 0.04002762958407402,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0012,
+      "step": 1862
+    },
+    {
+      "epoch": 16.037798165137616,
+      "grad_norm": 0.012818408198654652,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 1863
+    },
+    {
+      "epoch": 16.03816513761468,
+      "grad_norm": 0.044809773564338684,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0014,
+      "step": 1864
+    },
+    {
+      "epoch": 16.038532110091744,
+      "grad_norm": 0.053980398923158646,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0012,
+      "step": 1865
+    },
+    {
+      "epoch": 16.038899082568808,
+      "grad_norm": 0.025539973750710487,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0009,
+      "step": 1866
+    },
+    {
+      "epoch": 16.03926605504587,
+      "grad_norm": 0.09875453263521194,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0017,
+      "step": 1867
+    },
+    {
+      "epoch": 16.039633027522935,
+      "grad_norm": 0.032195813953876495,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0012,
+      "step": 1868
+    },
+    {
+      "epoch": 16.04,
+      "grad_norm": 44.63292694091797,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0799,
+      "step": 1869
+    },
+    {
+      "epoch": 16.040366972477063,
+      "grad_norm": 0.009903392754495144,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1870
+    },
+    {
+      "epoch": 16.040366972477063,
+      "eval_accuracy": 0.2514792899408284,
+      "eval_loss": 4.457294940948486,
+      "eval_runtime": 26.0015,
+      "eval_samples_per_second": 12.999,
+      "eval_steps_per_second": 1.654,
+      "step": 1870
+    },
+    {
+      "epoch": 17.000366972477064,
+      "grad_norm": 17.477916717529297,
+      "learning_rate": 6.25e-06,
+      "loss": 0.3298,
+      "step": 1871
+    },
+    {
+      "epoch": 17.000733944954128,
+      "grad_norm": 0.05367795377969742,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0012,
+      "step": 1872
+    },
+    {
+      "epoch": 17.00110091743119,
+      "grad_norm": 0.010049309581518173,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1873
+    },
+    {
+      "epoch": 17.001467889908255,
+      "grad_norm": 0.12345709651708603,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0018,
+      "step": 1874
+    },
+    {
+      "epoch": 17.001834862385323,
+      "grad_norm": 0.05371297523379326,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0019,
+      "step": 1875
+    },
+    {
+      "epoch": 17.002201834862387,
+      "grad_norm": 0.24640512466430664,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0041,
+      "step": 1876
+    },
+    {
+      "epoch": 17.00256880733945,
+      "grad_norm": 31.12327003479004,
+      "learning_rate": 6.25e-06,
+      "loss": 0.1457,
+      "step": 1877
+    },
+    {
+      "epoch": 17.002935779816514,
+      "grad_norm": 0.05583217367529869,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0014,
+      "step": 1878
+    },
+    {
+      "epoch": 17.00330275229358,
+      "grad_norm": 14.96981430053711,
+      "learning_rate": 6.25e-06,
+      "loss": 0.6899,
+      "step": 1879
+    },
+    {
+      "epoch": 17.003669724770642,
+      "grad_norm": 0.00852893851697445,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1880
+    },
+    {
+      "epoch": 17.004036697247706,
+      "grad_norm": 0.008971312083303928,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 1881
+    },
+    {
+      "epoch": 17.00440366972477,
+      "grad_norm": 1.217157006263733,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0098,
+      "step": 1882
+    },
+    {
+      "epoch": 17.004770642201834,
+      "grad_norm": 0.48853975534439087,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0054,
+      "step": 1883
+    },
+    {
+      "epoch": 17.005137614678898,
+      "grad_norm": 25.195314407348633,
+      "learning_rate": 6.25e-06,
+      "loss": 0.2755,
+      "step": 1884
+    },
+    {
+      "epoch": 17.005504587155965,
+      "grad_norm": 0.02359093725681305,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 1885
+    },
+    {
+      "epoch": 17.00587155963303,
+      "grad_norm": 0.019245833158493042,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 1886
+    },
+    {
+      "epoch": 17.006238532110093,
+      "grad_norm": 0.009344483725726604,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1887
+    },
+    {
+      "epoch": 17.006605504587156,
+      "grad_norm": 5.632843017578125,
+      "learning_rate": 6.25e-06,
+      "loss": 0.1113,
+      "step": 1888
+    },
+    {
+      "epoch": 17.00697247706422,
+      "grad_norm": 0.012568165548145771,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 1889
+    },
+    {
+      "epoch": 17.007339449541284,
+      "grad_norm": 0.009133757092058659,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1890
+    },
+    {
+      "epoch": 17.007706422018348,
+      "grad_norm": 0.05227641388773918,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0012,
+      "step": 1891
+    },
+    {
+      "epoch": 17.008073394495412,
+      "grad_norm": 0.06068197265267372,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0019,
+      "step": 1892
+    },
+    {
+      "epoch": 17.008440366972476,
+      "grad_norm": 0.02457628771662712,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 1893
+    },
+    {
+      "epoch": 17.00880733944954,
+      "grad_norm": 0.060813914984464645,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0012,
+      "step": 1894
+    },
+    {
+      "epoch": 17.009174311926607,
+      "grad_norm": 0.026891585439443588,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 1895
+    },
+    {
+      "epoch": 17.00954128440367,
+      "grad_norm": 0.347319096326828,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0034,
+      "step": 1896
+    },
+    {
+      "epoch": 17.009908256880735,
+      "grad_norm": 0.02671472169458866,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0007,
+      "step": 1897
+    },
+    {
+      "epoch": 17.0102752293578,
+      "grad_norm": 0.014482788741588593,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 1898
+    },
+    {
+      "epoch": 17.010642201834862,
+      "grad_norm": 0.053193967789411545,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0012,
+      "step": 1899
+    },
+    {
+      "epoch": 17.011009174311926,
+      "grad_norm": 0.014831092208623886,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 1900
+    },
+    {
+      "epoch": 17.01137614678899,
+      "grad_norm": 0.08786682784557343,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0031,
+      "step": 1901
+    },
+    {
+      "epoch": 17.011743119266054,
+      "grad_norm": 0.042599327862262726,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0009,
+      "step": 1902
+    },
+    {
+      "epoch": 17.012110091743118,
+      "grad_norm": 0.05343453586101532,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0015,
+      "step": 1903
+    },
+    {
+      "epoch": 17.012477064220185,
+      "grad_norm": 2.8523800373077393,
+      "learning_rate": 6.25e-06,
+      "loss": 0.3755,
+      "step": 1904
+    },
+    {
+      "epoch": 17.01284403669725,
+      "grad_norm": 24.88134002685547,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0539,
+      "step": 1905
+    },
+    {
+      "epoch": 17.013211009174313,
+      "grad_norm": 1.3844808340072632,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0167,
+      "step": 1906
+    },
+    {
+      "epoch": 17.013577981651377,
+      "grad_norm": 0.3920688033103943,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0059,
+      "step": 1907
+    },
+    {
+      "epoch": 17.01394495412844,
+      "grad_norm": 9.91474437713623,
+      "learning_rate": 6.25e-06,
+      "loss": 0.4594,
+      "step": 1908
+    },
+    {
+      "epoch": 17.014311926605505,
+      "grad_norm": 0.008450795896351337,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 1909
+    },
+    {
+      "epoch": 17.01467889908257,
+      "grad_norm": 0.4051934778690338,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0049,
+      "step": 1910
+    },
+    {
+      "epoch": 17.015045871559632,
+      "grad_norm": 0.5229806303977966,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0047,
+      "step": 1911
+    },
+    {
+      "epoch": 17.015412844036696,
+      "grad_norm": 0.011865312233567238,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 1912
+    },
+    {
+      "epoch": 17.01577981651376,
+      "grad_norm": 0.004810481332242489,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 1913
+    },
+    {
+      "epoch": 17.016146788990827,
+      "grad_norm": 0.042907748371362686,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0014,
+      "step": 1914
+    },
+    {
+      "epoch": 17.01651376146789,
+      "grad_norm": 12.105053901672363,
+      "learning_rate": 6.25e-06,
+      "loss": 0.4249,
+      "step": 1915
+    },
+    {
+      "epoch": 17.016880733944955,
+      "grad_norm": 0.14441271126270294,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0015,
+      "step": 1916
+    },
+    {
+      "epoch": 17.01724770642202,
+      "grad_norm": 4.462470531463623,
+      "learning_rate": 6.25e-06,
+      "loss": 0.7195,
+      "step": 1917
+    },
+    {
+      "epoch": 17.017614678899083,
+      "grad_norm": 0.032095350325107574,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0009,
+      "step": 1918
+    },
+    {
+      "epoch": 17.017981651376147,
+      "grad_norm": 4.893198013305664,
+      "learning_rate": 6.25e-06,
+      "loss": 0.1882,
+      "step": 1919
+    },
+    {
+      "epoch": 17.01834862385321,
+      "grad_norm": 0.01311525795608759,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 1920
+    },
+    {
+      "epoch": 17.018715596330274,
+      "grad_norm": 0.04030586779117584,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0008,
+      "step": 1921
+    },
+    {
+      "epoch": 17.019082568807338,
+      "grad_norm": 0.11734176427125931,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0023,
+      "step": 1922
+    },
+    {
+      "epoch": 17.019449541284402,
+      "grad_norm": 0.005079815164208412,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 1923
+    },
+    {
+      "epoch": 17.01981651376147,
+      "grad_norm": 0.05307978391647339,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0007,
+      "step": 1924
+    },
+    {
+      "epoch": 17.020183486238533,
+      "grad_norm": 0.02513750083744526,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0007,
+      "step": 1925
+    },
+    {
+      "epoch": 17.020550458715597,
+      "grad_norm": 0.01469388883560896,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 1926
+    },
+    {
+      "epoch": 17.02091743119266,
+      "grad_norm": 0.38227275013923645,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0053,
+      "step": 1927
+    },
+    {
+      "epoch": 17.021284403669725,
+      "grad_norm": 0.034598782658576965,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0011,
+      "step": 1928
+    },
+    {
+      "epoch": 17.02165137614679,
+      "grad_norm": 0.009877179749310017,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1929
+    },
+    {
+      "epoch": 17.022018348623853,
+      "grad_norm": 6.425004482269287,
+      "learning_rate": 6.25e-06,
+      "loss": 0.5444,
+      "step": 1930
+    },
+    {
+      "epoch": 17.022385321100916,
+      "grad_norm": 0.008717404678463936,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1931
+    },
+    {
+      "epoch": 17.02275229357798,
+      "grad_norm": 0.010533430613577366,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1932
+    },
+    {
+      "epoch": 17.023119266055044,
+      "grad_norm": 0.004461710341274738,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 1933
+    },
+    {
+      "epoch": 17.02348623853211,
+      "grad_norm": 0.1839299201965332,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0063,
+      "step": 1934
+    },
+    {
+      "epoch": 17.023853211009175,
+      "grad_norm": 0.4279024600982666,
+      "learning_rate": 6.25e-06,
+      "loss": 0.007,
+      "step": 1935
+    },
+    {
+      "epoch": 17.02422018348624,
+      "grad_norm": 0.6013566851615906,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0036,
+      "step": 1936
+    },
+    {
+      "epoch": 17.024587155963303,
+      "grad_norm": 0.006980085279792547,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 1937
+    },
+    {
+      "epoch": 17.024954128440367,
+      "grad_norm": 0.06412402540445328,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0013,
+      "step": 1938
+    },
+    {
+      "epoch": 17.02532110091743,
+      "grad_norm": 0.007534929551184177,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 1939
+    },
+    {
+      "epoch": 17.025688073394495,
+      "grad_norm": 0.024083394557237625,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0007,
+      "step": 1940
+    },
+    {
+      "epoch": 17.02605504587156,
+      "grad_norm": 0.05661863088607788,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0015,
+      "step": 1941
+    },
+    {
+      "epoch": 17.026422018348622,
+      "grad_norm": 0.001646261429414153,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0001,
+      "step": 1942
+    },
+    {
+      "epoch": 17.02678899082569,
+      "grad_norm": 0.016172723844647408,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 1943
+    },
+    {
+      "epoch": 17.027155963302754,
+      "grad_norm": 0.2106974571943283,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0035,
+      "step": 1944
+    },
+    {
+      "epoch": 17.027522935779817,
+      "grad_norm": 0.021005718037486076,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0008,
+      "step": 1945
+    },
+    {
+      "epoch": 17.02788990825688,
+      "grad_norm": 0.04612626135349274,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0012,
+      "step": 1946
+    },
+    {
+      "epoch": 17.028256880733945,
+      "grad_norm": 0.03742925450205803,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0012,
+      "step": 1947
+    },
+    {
+      "epoch": 17.02862385321101,
+      "grad_norm": 0.028907110914587975,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0009,
+      "step": 1948
+    },
+    {
+      "epoch": 17.028990825688073,
+      "grad_norm": 0.006026166956871748,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 1949
+    },
+    {
+      "epoch": 17.029357798165137,
+      "grad_norm": 0.01245788112282753,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 1950
+    },
+    {
+      "epoch": 17.0297247706422,
+      "grad_norm": 0.0025913205463439226,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0001,
+      "step": 1951
+    },
+    {
+      "epoch": 17.030091743119264,
+      "grad_norm": 0.016811618581414223,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1952
+    },
+    {
+      "epoch": 17.030458715596332,
+      "grad_norm": 0.0061055803671479225,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 1953
+    },
+    {
+      "epoch": 17.030825688073396,
+      "grad_norm": 0.02146347425878048,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0007,
+      "step": 1954
+    },
+    {
+      "epoch": 17.03119266055046,
+      "grad_norm": 9.06770133972168,
+      "learning_rate": 6.25e-06,
+      "loss": 0.6099,
+      "step": 1955
+    },
+    {
+      "epoch": 17.031559633027523,
+      "grad_norm": 0.070332832634449,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0018,
+      "step": 1956
+    },
+    {
+      "epoch": 17.031926605504587,
+      "grad_norm": 12.617301940917969,
+      "learning_rate": 6.25e-06,
+      "loss": 0.5954,
+      "step": 1957
+    },
+    {
+      "epoch": 17.03229357798165,
+      "grad_norm": 27.117158889770508,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0754,
+      "step": 1958
+    },
+    {
+      "epoch": 17.032660550458715,
+      "grad_norm": 0.05004006251692772,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0011,
+      "step": 1959
+    },
+    {
+      "epoch": 17.03302752293578,
+      "grad_norm": 0.010526631027460098,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 1960
+    },
+    {
+      "epoch": 17.033394495412843,
+      "grad_norm": 16.19788932800293,
+      "learning_rate": 6.25e-06,
+      "loss": 0.5387,
+      "step": 1961
+    },
+    {
+      "epoch": 17.033761467889907,
+      "grad_norm": 0.07324513792991638,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0022,
+      "step": 1962
+    },
+    {
+      "epoch": 17.034128440366974,
+      "grad_norm": 0.025474926456809044,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0011,
+      "step": 1963
+    },
+    {
+      "epoch": 17.034495412844038,
+      "grad_norm": 0.01056397333741188,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1964
+    },
+    {
+      "epoch": 17.0348623853211,
+      "grad_norm": 0.01814291439950466,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0008,
+      "step": 1965
+    },
+    {
+      "epoch": 17.035229357798165,
+      "grad_norm": 0.009125380776822567,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 1966
+    },
+    {
+      "epoch": 17.03559633027523,
+      "grad_norm": 0.0639096274971962,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0012,
+      "step": 1967
+    },
+    {
+      "epoch": 17.035963302752293,
+      "grad_norm": 0.16623161733150482,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0027,
+      "step": 1968
+    },
+    {
+      "epoch": 17.036330275229357,
+      "grad_norm": 0.013839509338140488,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 1969
+    },
+    {
+      "epoch": 17.03669724770642,
+      "grad_norm": 0.09220963716506958,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0015,
+      "step": 1970
+    },
+    {
+      "epoch": 17.037064220183485,
+      "grad_norm": 0.029545024037361145,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 1971
+    },
+    {
+      "epoch": 17.037431192660552,
+      "grad_norm": 0.1380985826253891,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0045,
+      "step": 1972
+    },
+    {
+      "epoch": 17.037798165137616,
+      "grad_norm": 0.008605247363448143,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 1973
+    },
+    {
+      "epoch": 17.03816513761468,
+      "grad_norm": 0.6541672945022583,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0063,
+      "step": 1974
+    },
+    {
+      "epoch": 17.038532110091744,
+      "grad_norm": 0.32922354340553284,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0055,
+      "step": 1975
+    },
+    {
+      "epoch": 17.038899082568808,
+      "grad_norm": 0.017893798649311066,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0008,
+      "step": 1976
+    },
+    {
+      "epoch": 17.03926605504587,
+      "grad_norm": 0.04146070405840874,
+      "learning_rate": 6.25e-06,
+      "loss": 0.001,
+      "step": 1977
+    },
+    {
+      "epoch": 17.039633027522935,
+      "grad_norm": 0.006382498424500227,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 1978
+    },
+    {
+      "epoch": 17.04,
+      "grad_norm": 0.011737913824617863,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1979
+    },
+    {
+      "epoch": 17.040366972477063,
+      "grad_norm": 13.5488920211792,
+      "learning_rate": 6.25e-06,
+      "loss": 0.5939,
+      "step": 1980
+    },
+    {
+      "epoch": 17.040366972477063,
+      "eval_accuracy": 0.35502958579881655,
+      "eval_loss": 4.342290878295898,
+      "eval_runtime": 25.9153,
+      "eval_samples_per_second": 13.042,
+      "eval_steps_per_second": 1.659,
+      "step": 1980
+    },
+    {
+      "epoch": 18.000366972477064,
+      "grad_norm": 0.24252568185329437,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0044,
+      "step": 1981
+    },
+    {
+      "epoch": 18.000733944954128,
+      "grad_norm": 0.005579877179116011,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 1982
+    },
+    {
+      "epoch": 18.00110091743119,
+      "grad_norm": 0.06243252009153366,
+      "learning_rate": 6.25e-06,
+      "loss": 0.001,
+      "step": 1983
+    },
+    {
+      "epoch": 18.001467889908255,
+      "grad_norm": 0.3555086553096771,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0078,
+      "step": 1984
+    },
+    {
+      "epoch": 18.001834862385323,
+      "grad_norm": 0.013176622800529003,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 1985
+    },
+    {
+      "epoch": 18.002201834862387,
+      "grad_norm": 0.025067154318094254,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0009,
+      "step": 1986
+    },
+    {
+      "epoch": 18.00256880733945,
+      "grad_norm": 0.11908164620399475,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0018,
+      "step": 1987
+    },
+    {
+      "epoch": 18.002935779816514,
+      "grad_norm": 0.12790387868881226,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0027,
+      "step": 1988
+    },
+    {
+      "epoch": 18.00330275229358,
+      "grad_norm": 0.05716130882501602,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0013,
+      "step": 1989
+    },
+    {
+      "epoch": 18.003669724770642,
+      "grad_norm": 0.009875811636447906,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 1990
+    },
+    {
+      "epoch": 18.004036697247706,
+      "grad_norm": 0.04722130298614502,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0019,
+      "step": 1991
+    },
+    {
+      "epoch": 18.00440366972477,
+      "grad_norm": 0.04684757813811302,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0009,
+      "step": 1992
+    },
+    {
+      "epoch": 18.004770642201834,
+      "grad_norm": 20.416101455688477,
+      "learning_rate": 6.25e-06,
+      "loss": 0.2395,
+      "step": 1993
+    },
+    {
+      "epoch": 18.005137614678898,
+      "grad_norm": 0.01941249892115593,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 1994
+    },
+    {
+      "epoch": 18.005504587155965,
+      "grad_norm": 0.2622072398662567,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0047,
+      "step": 1995
+    },
+    {
+      "epoch": 18.00587155963303,
+      "grad_norm": 0.026392275467514992,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0009,
+      "step": 1996
+    },
+    {
+      "epoch": 18.006238532110093,
+      "grad_norm": 0.16825400292873383,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0025,
+      "step": 1997
+    },
+    {
+      "epoch": 18.006605504587156,
+      "grad_norm": 11.92622184753418,
+      "learning_rate": 6.25e-06,
+      "loss": 0.373,
+      "step": 1998
+    },
+    {
+      "epoch": 18.00697247706422,
+      "grad_norm": 0.05778873339295387,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0011,
+      "step": 1999
+    },
+    {
+      "epoch": 18.007339449541284,
+      "grad_norm": 0.012871301732957363,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 2000
+    },
+    {
+      "epoch": 18.007706422018348,
+      "grad_norm": 0.03584280610084534,
+      "learning_rate": 6.25e-06,
+      "loss": 0.001,
+      "step": 2001
+    },
+    {
+      "epoch": 18.008073394495412,
+      "grad_norm": 0.08770672976970673,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0013,
+      "step": 2002
+    },
+    {
+      "epoch": 18.008440366972476,
+      "grad_norm": 0.014023813419044018,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 2003
+    },
+    {
+      "epoch": 18.00880733944954,
+      "grad_norm": 0.02106506936252117,
+      "learning_rate": 6.25e-06,
+      "loss": 0.001,
+      "step": 2004
+    },
+    {
+      "epoch": 18.009174311926607,
+      "grad_norm": 0.010105420835316181,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 2005
+    },
+    {
+      "epoch": 18.00954128440367,
+      "grad_norm": 50.41483688354492,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0804,
+      "step": 2006
+    },
+    {
+      "epoch": 18.009908256880735,
+      "grad_norm": 0.016915224492549896,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 2007
+    },
+    {
+      "epoch": 18.0102752293578,
+      "grad_norm": 0.014286397956311703,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 2008
+    },
+    {
+      "epoch": 18.010642201834862,
+      "grad_norm": 0.04319189116358757,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0011,
+      "step": 2009
+    },
+    {
+      "epoch": 18.011009174311926,
+      "grad_norm": 0.10475951433181763,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0018,
+      "step": 2010
+    },
+    {
+      "epoch": 18.01137614678899,
+      "grad_norm": 0.8170238733291626,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0046,
+      "step": 2011
+    },
+    {
+      "epoch": 18.011743119266054,
+      "grad_norm": 0.7766091227531433,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0076,
+      "step": 2012
+    },
+    {
+      "epoch": 18.012110091743118,
+      "grad_norm": 0.009043526835739613,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 2013
+    },
+    {
+      "epoch": 18.012477064220185,
+      "grad_norm": 0.10672438889741898,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0016,
+      "step": 2014
+    },
+    {
+      "epoch": 18.01284403669725,
+      "grad_norm": 0.011398933827877045,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 2015
+    },
+    {
+      "epoch": 18.013211009174313,
+      "grad_norm": 2.040708303451538,
+      "learning_rate": 6.25e-06,
+      "loss": 0.009,
+      "step": 2016
+    },
+    {
+      "epoch": 18.013577981651377,
+      "grad_norm": 0.03768087923526764,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0013,
+      "step": 2017
+    },
+    {
+      "epoch": 18.01394495412844,
+      "grad_norm": 8.129063606262207,
+      "learning_rate": 6.25e-06,
+      "loss": 0.6639,
+      "step": 2018
+    },
+    {
+      "epoch": 18.014311926605505,
+      "grad_norm": 0.018514199182391167,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0008,
+      "step": 2019
+    },
+    {
+      "epoch": 18.01467889908257,
+      "grad_norm": 0.008398096077144146,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2020
+    },
+    {
+      "epoch": 18.015045871559632,
+      "grad_norm": 0.19901345670223236,
+      "learning_rate": 6.25e-06,
+      "loss": 0.003,
+      "step": 2021
+    },
+    {
+      "epoch": 18.015412844036696,
+      "grad_norm": 0.009966458193957806,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 2022
+    },
+    {
+      "epoch": 18.01577981651376,
+      "grad_norm": 0.2510676085948944,
+      "learning_rate": 6.25e-06,
+      "loss": 0.003,
+      "step": 2023
+    },
+    {
+      "epoch": 18.016146788990827,
+      "grad_norm": 14.690150260925293,
+      "learning_rate": 6.25e-06,
+      "loss": 0.6954,
+      "step": 2024
+    },
+    {
+      "epoch": 18.01651376146789,
+      "grad_norm": 0.13395769894123077,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0018,
+      "step": 2025
+    },
+    {
+      "epoch": 18.016880733944955,
+      "grad_norm": 1.256048321723938,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0335,
+      "step": 2026
+    },
+    {
+      "epoch": 18.01724770642202,
+      "grad_norm": 0.019024115055799484,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0009,
+      "step": 2027
+    },
+    {
+      "epoch": 18.017614678899083,
+      "grad_norm": 0.0050199199467897415,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 2028
+    },
+    {
+      "epoch": 18.017981651376147,
+      "grad_norm": 0.1604446917772293,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0015,
+      "step": 2029
+    },
+    {
+      "epoch": 18.01834862385321,
+      "grad_norm": 21.672311782836914,
+      "learning_rate": 6.25e-06,
+      "loss": 0.39,
+      "step": 2030
+    },
+    {
+      "epoch": 18.018715596330274,
+      "grad_norm": 0.019483106210827827,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 2031
+    },
+    {
+      "epoch": 18.019082568807338,
+      "grad_norm": 19.6845703125,
+      "learning_rate": 6.25e-06,
+      "loss": 0.396,
+      "step": 2032
+    },
+    {
+      "epoch": 18.019449541284402,
+      "grad_norm": 0.013813750818371773,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 2033
+    },
+    {
+      "epoch": 18.01981651376147,
+      "grad_norm": 0.011385602876543999,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2034
+    },
+    {
+      "epoch": 18.020183486238533,
+      "grad_norm": 0.02313322201371193,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 2035
+    },
+    {
+      "epoch": 18.020550458715597,
+      "grad_norm": 0.006863860413432121,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2036
+    },
+    {
+      "epoch": 18.02091743119266,
+      "grad_norm": 0.020175714045763016,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0007,
+      "step": 2037
+    },
+    {
+      "epoch": 18.021284403669725,
+      "grad_norm": 0.012344634160399437,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 2038
+    },
+    {
+      "epoch": 18.02165137614679,
+      "grad_norm": 0.07676948606967926,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0022,
+      "step": 2039
+    },
+    {
+      "epoch": 18.022018348623853,
+      "grad_norm": 0.026113180443644524,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0008,
+      "step": 2040
+    },
+    {
+      "epoch": 18.022385321100916,
+      "grad_norm": 0.14478325843811035,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0015,
+      "step": 2041
+    },
+    {
+      "epoch": 18.02275229357798,
+      "grad_norm": 0.048560094088315964,
+      "learning_rate": 6.25e-06,
+      "loss": 0.001,
+      "step": 2042
+    },
+    {
+      "epoch": 18.023119266055044,
+      "grad_norm": 0.014162909239530563,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 2043
+    },
+    {
+      "epoch": 18.02348623853211,
+      "grad_norm": 0.005682117305696011,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2044
+    },
+    {
+      "epoch": 18.023853211009175,
+      "grad_norm": 0.041269391775131226,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0011,
+      "step": 2045
+    },
+    {
+      "epoch": 18.02422018348624,
+      "grad_norm": 9.198448181152344,
+      "learning_rate": 6.25e-06,
+      "loss": 0.5623,
+      "step": 2046
+    },
+    {
+      "epoch": 18.024587155963303,
+      "grad_norm": 0.29487547278404236,
+      "learning_rate": 6.25e-06,
+      "loss": 0.003,
+      "step": 2047
+    },
+    {
+      "epoch": 18.024954128440367,
+      "grad_norm": 0.17202623188495636,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0036,
+      "step": 2048
+    },
+    {
+      "epoch": 18.02532110091743,
+      "grad_norm": 0.041163764894008636,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0011,
+      "step": 2049
+    },
+    {
+      "epoch": 18.025688073394495,
+      "grad_norm": 0.04037086293101311,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0011,
+      "step": 2050
+    },
+    {
+      "epoch": 18.02605504587156,
+      "grad_norm": 6.995089530944824,
+      "learning_rate": 6.25e-06,
+      "loss": 0.2907,
+      "step": 2051
+    },
+    {
+      "epoch": 18.026422018348622,
+      "grad_norm": 0.06195395439863205,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0011,
+      "step": 2052
+    },
+    {
+      "epoch": 18.02678899082569,
+      "grad_norm": 0.008185009472072124,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2053
+    },
+    {
+      "epoch": 18.027155963302754,
+      "grad_norm": 0.05947251245379448,
+      "learning_rate": 6.25e-06,
+      "loss": 0.001,
+      "step": 2054
+    },
+    {
+      "epoch": 18.027522935779817,
+      "grad_norm": 0.016084305942058563,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 2055
+    },
+    {
+      "epoch": 18.02788990825688,
+      "grad_norm": 0.018039844930171967,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 2056
+    },
+    {
+      "epoch": 18.028256880733945,
+      "grad_norm": 0.020833930000662804,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 2057
+    },
+    {
+      "epoch": 18.02862385321101,
+      "grad_norm": 0.7317304015159607,
+      "learning_rate": 6.25e-06,
+      "loss": 0.008,
+      "step": 2058
+    },
+    {
+      "epoch": 18.028990825688073,
+      "grad_norm": 0.04000980779528618,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0018,
+      "step": 2059
+    },
+    {
+      "epoch": 18.029357798165137,
+      "grad_norm": 0.12100006639957428,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0013,
+      "step": 2060
+    },
+    {
+      "epoch": 18.0297247706422,
+      "grad_norm": 2.606705665588379,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0294,
+      "step": 2061
+    },
+    {
+      "epoch": 18.030091743119264,
+      "grad_norm": 13.80385971069336,
+      "learning_rate": 6.25e-06,
+      "loss": 0.6235,
+      "step": 2062
+    },
+    {
+      "epoch": 18.030458715596332,
+      "grad_norm": 0.0024406982120126486,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0001,
+      "step": 2063
+    },
+    {
+      "epoch": 18.030825688073396,
+      "grad_norm": 16.76881217956543,
+      "learning_rate": 6.25e-06,
+      "loss": 0.3475,
+      "step": 2064
+    },
+    {
+      "epoch": 18.03119266055046,
+      "grad_norm": 0.08073633164167404,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0013,
+      "step": 2065
+    },
+    {
+      "epoch": 18.031559633027523,
+      "grad_norm": 0.024951258674263954,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 2066
+    },
+    {
+      "epoch": 18.031926605504587,
+      "grad_norm": 0.0061537837609648705,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2067
+    },
+    {
+      "epoch": 18.03229357798165,
+      "grad_norm": 0.01806722767651081,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 2068
+    },
+    {
+      "epoch": 18.032660550458715,
+      "grad_norm": 0.3512369692325592,
+      "learning_rate": 6.25e-06,
+      "loss": 0.004,
+      "step": 2069
+    },
+    {
+      "epoch": 18.03302752293578,
+      "grad_norm": 0.19450858235359192,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0023,
+      "step": 2070
+    },
+    {
+      "epoch": 18.033394495412843,
+      "grad_norm": 0.007415194995701313,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2071
+    },
+    {
+      "epoch": 18.033761467889907,
+      "grad_norm": 0.15916714072227478,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0018,
+      "step": 2072
+    },
+    {
+      "epoch": 18.034128440366974,
+      "grad_norm": 0.015283467248082161,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 2073
+    },
+    {
+      "epoch": 18.034495412844038,
+      "grad_norm": 0.012823554687201977,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 2074
+    },
+    {
+      "epoch": 18.0348623853211,
+      "grad_norm": 27.923599243164062,
+      "learning_rate": 6.25e-06,
+      "loss": 0.1579,
+      "step": 2075
+    },
+    {
+      "epoch": 18.035229357798165,
+      "grad_norm": 0.10715489834547043,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0031,
+      "step": 2076
+    },
+    {
+      "epoch": 18.03559633027523,
+      "grad_norm": 0.43819621205329895,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0062,
+      "step": 2077
+    },
+    {
+      "epoch": 18.035963302752293,
+      "grad_norm": 4.0556440353393555,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0234,
+      "step": 2078
+    },
+    {
+      "epoch": 18.036330275229357,
+      "grad_norm": 0.01272510178387165,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 2079
+    },
+    {
+      "epoch": 18.03669724770642,
+      "grad_norm": 0.5048418045043945,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0065,
+      "step": 2080
+    },
+    {
+      "epoch": 18.037064220183485,
+      "grad_norm": 0.011538013815879822,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 2081
+    },
+    {
+      "epoch": 18.037431192660552,
+      "grad_norm": 3.383413553237915,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0156,
+      "step": 2082
+    },
+    {
+      "epoch": 18.037798165137616,
+      "grad_norm": 0.04888997599482536,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0011,
+      "step": 2083
+    },
+    {
+      "epoch": 18.03816513761468,
+      "grad_norm": 0.5483319163322449,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0083,
+      "step": 2084
+    },
+    {
+      "epoch": 18.038532110091744,
+      "grad_norm": 0.0180961936712265,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 2085
+    },
+    {
+      "epoch": 18.038899082568808,
+      "grad_norm": 0.13374537229537964,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0021,
+      "step": 2086
+    },
+    {
+      "epoch": 18.03926605504587,
+      "grad_norm": 0.037403788417577744,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0007,
+      "step": 2087
+    },
+    {
+      "epoch": 18.039633027522935,
+      "grad_norm": 0.01548128854483366,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 2088
+    },
+    {
+      "epoch": 18.04,
+      "grad_norm": 0.006623277440667152,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2089
+    },
+    {
+      "epoch": 18.040366972477063,
+      "grad_norm": 0.03926685079932213,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0013,
+      "step": 2090
+    },
+    {
+      "epoch": 18.040366972477063,
+      "eval_accuracy": 0.3047337278106509,
+      "eval_loss": 4.336460590362549,
+      "eval_runtime": 26.1068,
+      "eval_samples_per_second": 12.947,
+      "eval_steps_per_second": 1.647,
+      "step": 2090
+    },
+    {
+      "epoch": 19.000366972477064,
+      "grad_norm": 5.828792572021484,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0714,
+      "step": 2091
+    },
+    {
+      "epoch": 19.000733944954128,
+      "grad_norm": 0.32053354382514954,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0118,
+      "step": 2092
+    },
+    {
+      "epoch": 19.00110091743119,
+      "grad_norm": 43.80705642700195,
+      "learning_rate": 6.25e-06,
+      "loss": 0.1775,
+      "step": 2093
+    },
+    {
+      "epoch": 19.001467889908255,
+      "grad_norm": 0.01150493323802948,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 2094
+    },
+    {
+      "epoch": 19.001834862385323,
+      "grad_norm": 10.52509880065918,
+      "learning_rate": 6.25e-06,
+      "loss": 0.4943,
+      "step": 2095
+    },
+    {
+      "epoch": 19.002201834862387,
+      "grad_norm": 0.12844768166542053,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0034,
+      "step": 2096
+    },
+    {
+      "epoch": 19.00256880733945,
+      "grad_norm": 0.02460050955414772,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 2097
+    },
+    {
+      "epoch": 19.002935779816514,
+      "grad_norm": 0.031593263149261475,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0007,
+      "step": 2098
+    },
+    {
+      "epoch": 19.00330275229358,
+      "grad_norm": 0.018347637727856636,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 2099
+    },
+    {
+      "epoch": 19.003669724770642,
+      "grad_norm": 0.006747081410139799,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2100
+    },
+    {
+      "epoch": 19.004036697247706,
+      "grad_norm": 0.018978174775838852,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 2101
+    },
+    {
+      "epoch": 19.00440366972477,
+      "grad_norm": 0.0042154258117079735,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 2102
+    },
+    {
+      "epoch": 19.004770642201834,
+      "grad_norm": 0.10156916081905365,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0019,
+      "step": 2103
+    },
+    {
+      "epoch": 19.005137614678898,
+      "grad_norm": 0.011269387789070606,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 2104
+    },
+    {
+      "epoch": 19.005504587155965,
+      "grad_norm": 1.0194971561431885,
+      "learning_rate": 6.25e-06,
+      "loss": 0.005,
+      "step": 2105
+    },
+    {
+      "epoch": 19.00587155963303,
+      "grad_norm": 0.004952696152031422,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 2106
+    },
+    {
+      "epoch": 19.006238532110093,
+      "grad_norm": 0.032845549285411835,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0007,
+      "step": 2107
+    },
+    {
+      "epoch": 19.006605504587156,
+      "grad_norm": 14.156702995300293,
+      "learning_rate": 6.25e-06,
+      "loss": 0.4759,
+      "step": 2108
+    },
+    {
+      "epoch": 19.00697247706422,
+      "grad_norm": 0.05579781532287598,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0011,
+      "step": 2109
+    },
+    {
+      "epoch": 19.007339449541284,
+      "grad_norm": 0.5355889201164246,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0047,
+      "step": 2110
+    },
+    {
+      "epoch": 19.007706422018348,
+      "grad_norm": 0.11374775320291519,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0026,
+      "step": 2111
+    },
+    {
+      "epoch": 19.008073394495412,
+      "grad_norm": 0.006869818549603224,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2112
+    },
+    {
+      "epoch": 19.008440366972476,
+      "grad_norm": 0.0048071532510221004,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 2113
+    },
+    {
+      "epoch": 19.00880733944954,
+      "grad_norm": 0.0023122879210859537,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0001,
+      "step": 2114
+    },
+    {
+      "epoch": 19.009174311926607,
+      "grad_norm": 0.13460251688957214,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0018,
+      "step": 2115
+    },
+    {
+      "epoch": 19.00954128440367,
+      "grad_norm": 0.007274499628692865,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2116
+    },
+    {
+      "epoch": 19.009908256880735,
+      "grad_norm": 0.08446405827999115,
+      "learning_rate": 6.25e-06,
+      "loss": 0.001,
+      "step": 2117
+    },
+    {
+      "epoch": 19.0102752293578,
+      "grad_norm": 0.005403855815529823,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 2118
+    },
+    {
+      "epoch": 19.010642201834862,
+      "grad_norm": 0.0320221446454525,
+      "learning_rate": 6.25e-06,
+      "loss": 0.001,
+      "step": 2119
+    },
+    {
+      "epoch": 19.011009174311926,
+      "grad_norm": 0.0038279606960713863,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 2120
+    },
+    {
+      "epoch": 19.01137614678899,
+      "grad_norm": 0.1845453828573227,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0017,
+      "step": 2121
+    },
+    {
+      "epoch": 19.011743119266054,
+      "grad_norm": 30.10391616821289,
+      "learning_rate": 6.25e-06,
+      "loss": 0.3444,
+      "step": 2122
+    },
+    {
+      "epoch": 19.012110091743118,
+      "grad_norm": 0.00939994491636753,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2123
+    },
+    {
+      "epoch": 19.012477064220185,
+      "grad_norm": 0.058112144470214844,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0011,
+      "step": 2124
+    },
+    {
+      "epoch": 19.01284403669725,
+      "grad_norm": 0.01599031127989292,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 2125
+    },
+    {
+      "epoch": 19.013211009174313,
+      "grad_norm": 0.00823430810123682,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2126
+    },
+    {
+      "epoch": 19.013577981651377,
+      "grad_norm": 0.22702009975910187,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0044,
+      "step": 2127
+    },
+    {
+      "epoch": 19.01394495412844,
+      "grad_norm": 0.011800515465438366,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 2128
+    },
+    {
+      "epoch": 19.014311926605505,
+      "grad_norm": 0.024462571367621422,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 2129
+    },
+    {
+      "epoch": 19.01467889908257,
+      "grad_norm": 42.59157943725586,
+      "learning_rate": 6.25e-06,
+      "loss": 0.3904,
+      "step": 2130
+    },
+    {
+      "epoch": 19.015045871559632,
+      "grad_norm": 11.815838813781738,
+      "learning_rate": 6.25e-06,
+      "loss": 0.5384,
+      "step": 2131
+    },
+    {
+      "epoch": 19.015412844036696,
+      "grad_norm": 4.486865043640137,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0245,
+      "step": 2132
+    },
+    {
+      "epoch": 19.01577981651376,
+      "grad_norm": 0.09078022837638855,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0013,
+      "step": 2133
+    },
+    {
+      "epoch": 19.016146788990827,
+      "grad_norm": 0.018901098519563675,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 2134
+    },
+    {
+      "epoch": 19.01651376146789,
+      "grad_norm": 63.40033721923828,
+      "learning_rate": 6.25e-06,
+      "loss": 0.8218,
+      "step": 2135
+    },
+    {
+      "epoch": 19.016880733944955,
+      "grad_norm": 2.4428036212921143,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0236,
+      "step": 2136
+    },
+    {
+      "epoch": 19.01724770642202,
+      "grad_norm": 0.0049377023242414,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 2137
+    },
+    {
+      "epoch": 19.017614678899083,
+      "grad_norm": 0.023528199642896652,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0008,
+      "step": 2138
+    },
+    {
+      "epoch": 19.017981651376147,
+      "grad_norm": 26.496801376342773,
+      "learning_rate": 6.25e-06,
+      "loss": 0.1987,
+      "step": 2139
+    },
+    {
+      "epoch": 19.01834862385321,
+      "grad_norm": 0.05308035388588905,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0008,
+      "step": 2140
+    },
+    {
+      "epoch": 19.018715596330274,
+      "grad_norm": 13.511025428771973,
+      "learning_rate": 6.25e-06,
+      "loss": 0.5404,
+      "step": 2141
+    },
+    {
+      "epoch": 19.019082568807338,
+      "grad_norm": 0.004166475962847471,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 2142
+    },
+    {
+      "epoch": 19.019449541284402,
+      "grad_norm": 0.10278932005167007,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0014,
+      "step": 2143
+    },
+    {
+      "epoch": 19.01981651376147,
+      "grad_norm": 0.006754185073077679,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2144
+    },
+    {
+      "epoch": 19.020183486238533,
+      "grad_norm": 0.018132977187633514,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 2145
+    },
+    {
+      "epoch": 19.020550458715597,
+      "grad_norm": 0.015186520293354988,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 2146
+    },
+    {
+      "epoch": 19.02091743119266,
+      "grad_norm": 0.2977631688117981,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0022,
+      "step": 2147
+    },
+    {
+      "epoch": 19.021284403669725,
+      "grad_norm": 27.114315032958984,
+      "learning_rate": 6.25e-06,
+      "loss": 1.2159,
+      "step": 2148
+    },
+    {
+      "epoch": 19.02165137614679,
+      "grad_norm": 0.004629006143659353,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 2149
+    },
+    {
+      "epoch": 19.022018348623853,
+      "grad_norm": 0.006630502641201019,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2150
+    },
+    {
+      "epoch": 19.022385321100916,
+      "grad_norm": 0.2711504399776459,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0027,
+      "step": 2151
+    },
+    {
+      "epoch": 19.02275229357798,
+      "grad_norm": 0.019685599952936172,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0007,
+      "step": 2152
+    },
+    {
+      "epoch": 19.023119266055044,
+      "grad_norm": 14.860023498535156,
+      "learning_rate": 6.25e-06,
+      "loss": 0.4439,
+      "step": 2153
+    },
+    {
+      "epoch": 19.02348623853211,
+      "grad_norm": 0.009670207276940346,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 2154
+    },
+    {
+      "epoch": 19.023853211009175,
+      "grad_norm": 0.06338346749544144,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0014,
+      "step": 2155
+    },
+    {
+      "epoch": 19.02422018348624,
+      "grad_norm": 0.014472728595137596,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 2156
+    },
+    {
+      "epoch": 19.024587155963303,
+      "grad_norm": 0.08244498074054718,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0027,
+      "step": 2157
+    },
+    {
+      "epoch": 19.024954128440367,
+      "grad_norm": 0.006686851382255554,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2158
+    },
+    {
+      "epoch": 19.02532110091743,
+      "grad_norm": 0.013287510722875595,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 2159
+    },
+    {
+      "epoch": 19.025688073394495,
+      "grad_norm": 0.013549952767789364,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2160
+    },
+    {
+      "epoch": 19.02605504587156,
+      "grad_norm": 0.05713585764169693,
+      "learning_rate": 6.25e-06,
+      "loss": 0.001,
+      "step": 2161
+    },
+    {
+      "epoch": 19.026422018348622,
+      "grad_norm": 0.00891933124512434,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2162
+    },
+    {
+      "epoch": 19.02678899082569,
+      "grad_norm": 0.19058436155319214,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0059,
+      "step": 2163
+    },
+    {
+      "epoch": 19.027155963302754,
+      "grad_norm": 10.728453636169434,
+      "learning_rate": 6.25e-06,
+      "loss": 0.7289,
+      "step": 2164
+    },
+    {
+      "epoch": 19.027522935779817,
+      "grad_norm": 0.014145110733807087,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 2165
+    },
+    {
+      "epoch": 19.02788990825688,
+      "grad_norm": 0.01616099290549755,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 2166
+    },
+    {
+      "epoch": 19.028256880733945,
+      "grad_norm": 0.005085713230073452,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 2167
+    },
+    {
+      "epoch": 19.02862385321101,
+      "grad_norm": 6.216205596923828,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0372,
+      "step": 2168
+    },
+    {
+      "epoch": 19.028990825688073,
+      "grad_norm": 0.007044653408229351,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2169
+    },
+    {
+      "epoch": 19.029357798165137,
+      "grad_norm": 0.040171001106500626,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0007,
+      "step": 2170
+    },
+    {
+      "epoch": 19.0297247706422,
+      "grad_norm": 6.612220287322998,
+      "learning_rate": 6.25e-06,
+      "loss": 0.7196,
+      "step": 2171
+    },
+    {
+      "epoch": 19.030091743119264,
+      "grad_norm": 0.008035913109779358,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2172
+    },
+    {
+      "epoch": 19.030458715596332,
+      "grad_norm": 0.005456095561385155,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 2173
+    },
+    {
+      "epoch": 19.030825688073396,
+      "grad_norm": 0.014944036491215229,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0007,
+      "step": 2174
+    },
+    {
+      "epoch": 19.03119266055046,
+      "grad_norm": 0.00724659813567996,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2175
+    },
+    {
+      "epoch": 19.031559633027523,
+      "grad_norm": 0.020265886560082436,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0009,
+      "step": 2176
+    },
+    {
+      "epoch": 19.031926605504587,
+      "grad_norm": 0.0053767128847539425,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 2177
+    },
+    {
+      "epoch": 19.03229357798165,
+      "grad_norm": 0.004067923408001661,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 2178
+    },
+    {
+      "epoch": 19.032660550458715,
+      "grad_norm": 0.006404847837984562,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2179
+    },
+    {
+      "epoch": 19.03302752293578,
+      "grad_norm": 0.012277035042643547,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 2180
+    },
+    {
+      "epoch": 19.033394495412843,
+      "grad_norm": 29.482219696044922,
+      "learning_rate": 6.25e-06,
+      "loss": 0.3153,
+      "step": 2181
+    },
+    {
+      "epoch": 19.033761467889907,
+      "grad_norm": 0.005679774563759565,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 2182
+    },
+    {
+      "epoch": 19.034128440366974,
+      "grad_norm": 0.013896224088966846,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 2183
+    },
+    {
+      "epoch": 19.034495412844038,
+      "grad_norm": 19.799856185913086,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0316,
+      "step": 2184
+    },
+    {
+      "epoch": 19.0348623853211,
+      "grad_norm": 10.89291763305664,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0198,
+      "step": 2185
+    },
+    {
+      "epoch": 19.035229357798165,
+      "grad_norm": 0.011448165401816368,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 2186
+    },
+    {
+      "epoch": 19.03559633027523,
+      "grad_norm": 0.00795645173639059,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2187
+    },
+    {
+      "epoch": 19.035963302752293,
+      "grad_norm": 0.006039537489414215,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0002,
+      "step": 2188
+    },
+    {
+      "epoch": 19.036330275229357,
+      "grad_norm": 0.013683117926120758,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0005,
+      "step": 2189
+    },
+    {
+      "epoch": 19.03669724770642,
+      "grad_norm": 0.008376816287636757,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 2190
+    },
+    {
+      "epoch": 19.037064220183485,
+      "grad_norm": 0.3954543173313141,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0009,
+      "step": 2191
+    },
+    {
+      "epoch": 19.037431192660552,
+      "grad_norm": 0.007542344741523266,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2192
+    },
+    {
+      "epoch": 19.037798165137616,
+      "grad_norm": 0.006368572358042002,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0003,
+      "step": 2193
+    },
+    {
+      "epoch": 19.03816513761468,
+      "grad_norm": 0.0201836246997118,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 2194
+    },
+    {
+      "epoch": 19.038532110091744,
+      "grad_norm": 3.723318576812744,
+      "learning_rate": 6.25e-06,
+      "loss": 0.6602,
+      "step": 2195
+    },
+    {
+      "epoch": 19.038899082568808,
+      "grad_norm": 0.01571609452366829,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0006,
+      "step": 2196
+    },
+    {
+      "epoch": 19.03926605504587,
+      "grad_norm": 0.0682874545454979,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0014,
+      "step": 2197
+    },
+    {
+      "epoch": 19.039633027522935,
+      "grad_norm": 0.11792181432247162,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0016,
+      "step": 2198
+    },
+    {
+      "epoch": 19.04,
+      "grad_norm": 0.20089532434940338,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0028,
+      "step": 2199
+    },
+    {
+      "epoch": 19.040366972477063,
+      "grad_norm": 0.07066699862480164,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0015,
+      "step": 2200
+    },
+    {
+      "epoch": 19.040366972477063,
+      "eval_accuracy": 0.24260355029585798,
+      "eval_loss": 4.0963568687438965,
+      "eval_runtime": 26.0641,
+      "eval_samples_per_second": 12.968,
+      "eval_steps_per_second": 1.65,
+      "step": 2200
+    },
+    {
+      "epoch": 20.000366972477064,
+      "grad_norm": 0.015317992307245731,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2201
+    },
+    {
+      "epoch": 20.000733944954128,
+      "grad_norm": 0.004152167122811079,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0002,
+      "step": 2202
+    },
+    {
+      "epoch": 20.00110091743119,
+      "grad_norm": 0.012887159362435341,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2203
+    },
+    {
+      "epoch": 20.001467889908255,
+      "grad_norm": 0.005667123943567276,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0002,
+      "step": 2204
+    },
+    {
+      "epoch": 20.001834862385323,
+      "grad_norm": 0.017617488279938698,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2205
+    },
+    {
+      "epoch": 20.002201834862387,
+      "grad_norm": 0.02631082944571972,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0011,
+      "step": 2206
+    },
+    {
+      "epoch": 20.00256880733945,
+      "grad_norm": 0.002945562591776252,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0001,
+      "step": 2207
+    },
+    {
+      "epoch": 20.002935779816514,
+      "grad_norm": 16.195940017700195,
+      "learning_rate": 3.125e-06,
+      "loss": 0.1129,
+      "step": 2208
+    },
+    {
+      "epoch": 20.00330275229358,
+      "grad_norm": 0.11761868745088577,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0019,
+      "step": 2209
+    },
+    {
+      "epoch": 20.003669724770642,
+      "grad_norm": 0.017874015495181084,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2210
+    },
+    {
+      "epoch": 20.004036697247706,
+      "grad_norm": 0.007360911928117275,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2211
+    },
+    {
+      "epoch": 20.00440366972477,
+      "grad_norm": 0.008144790306687355,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2212
+    },
+    {
+      "epoch": 20.004770642201834,
+      "grad_norm": 6.975256443023682,
+      "learning_rate": 3.125e-06,
+      "loss": 0.837,
+      "step": 2213
+    },
+    {
+      "epoch": 20.005137614678898,
+      "grad_norm": 0.010705436579883099,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2214
+    },
+    {
+      "epoch": 20.005504587155965,
+      "grad_norm": 0.017650488764047623,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0008,
+      "step": 2215
+    },
+    {
+      "epoch": 20.00587155963303,
+      "grad_norm": 0.025242948904633522,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0006,
+      "step": 2216
+    },
+    {
+      "epoch": 20.006238532110093,
+      "grad_norm": 0.007139777764678001,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2217
+    },
+    {
+      "epoch": 20.006605504587156,
+      "grad_norm": 37.2630500793457,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0756,
+      "step": 2218
+    },
+    {
+      "epoch": 20.00697247706422,
+      "grad_norm": 0.008383985608816147,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2219
+    },
+    {
+      "epoch": 20.007339449541284,
+      "grad_norm": 0.013497921638190746,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2220
+    },
+    {
+      "epoch": 20.007706422018348,
+      "grad_norm": 0.022555328905582428,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2221
+    },
+    {
+      "epoch": 20.008073394495412,
+      "grad_norm": 0.03195679560303688,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0006,
+      "step": 2222
+    },
+    {
+      "epoch": 20.008440366972476,
+      "grad_norm": 0.012581287883222103,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2223
+    },
+    {
+      "epoch": 20.00880733944954,
+      "grad_norm": 0.007420066278427839,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2224
+    },
+    {
+      "epoch": 20.009174311926607,
+      "grad_norm": 77.09980010986328,
+      "learning_rate": 3.125e-06,
+      "loss": 0.1997,
+      "step": 2225
+    },
+    {
+      "epoch": 20.00954128440367,
+      "grad_norm": 0.025901569053530693,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0011,
+      "step": 2226
+    },
+    {
+      "epoch": 20.009908256880735,
+      "grad_norm": 0.08973561972379684,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0018,
+      "step": 2227
+    },
+    {
+      "epoch": 20.0102752293578,
+      "grad_norm": 0.056077077984809875,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0007,
+      "step": 2228
+    },
+    {
+      "epoch": 20.010642201834862,
+      "grad_norm": 10.603231430053711,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0682,
+      "step": 2229
+    },
+    {
+      "epoch": 20.011009174311926,
+      "grad_norm": 0.00524101871997118,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0002,
+      "step": 2230
+    },
+    {
+      "epoch": 20.01137614678899,
+      "grad_norm": 0.03746435418725014,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0008,
+      "step": 2231
+    },
+    {
+      "epoch": 20.011743119266054,
+      "grad_norm": 0.02348695695400238,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0009,
+      "step": 2232
+    },
+    {
+      "epoch": 20.012110091743118,
+      "grad_norm": 0.023266052827239037,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2233
+    },
+    {
+      "epoch": 20.012477064220185,
+      "grad_norm": 0.01092996820807457,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2234
+    },
+    {
+      "epoch": 20.01284403669725,
+      "grad_norm": 4.052068710327148,
+      "learning_rate": 3.125e-06,
+      "loss": 0.9208,
+      "step": 2235
+    },
+    {
+      "epoch": 20.013211009174313,
+      "grad_norm": 0.046622615307569504,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0009,
+      "step": 2236
+    },
+    {
+      "epoch": 20.013577981651377,
+      "grad_norm": 4.312165260314941,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0586,
+      "step": 2237
+    },
+    {
+      "epoch": 20.01394495412844,
+      "grad_norm": 0.00927566085010767,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2238
+    },
+    {
+      "epoch": 20.014311926605505,
+      "grad_norm": 0.007015680428594351,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2239
+    },
+    {
+      "epoch": 20.01467889908257,
+      "grad_norm": 0.05858108773827553,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0018,
+      "step": 2240
+    },
+    {
+      "epoch": 20.015045871559632,
+      "grad_norm": 0.02790078893303871,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0008,
+      "step": 2241
+    },
+    {
+      "epoch": 20.015412844036696,
+      "grad_norm": 0.008722535334527493,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2242
+    },
+    {
+      "epoch": 20.01577981651376,
+      "grad_norm": 0.05083201453089714,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0012,
+      "step": 2243
+    },
+    {
+      "epoch": 20.016146788990827,
+      "grad_norm": 0.0782344788312912,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0011,
+      "step": 2244
+    },
+    {
+      "epoch": 20.01651376146789,
+      "grad_norm": 0.0048544323071837425,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0002,
+      "step": 2245
+    },
+    {
+      "epoch": 20.016880733944955,
+      "grad_norm": 0.12359505891799927,
+      "learning_rate": 3.125e-06,
+      "loss": 0.002,
+      "step": 2246
+    },
+    {
+      "epoch": 20.01724770642202,
+      "grad_norm": 0.08342505991458893,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0016,
+      "step": 2247
+    },
+    {
+      "epoch": 20.017614678899083,
+      "grad_norm": 0.015932219102978706,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0006,
+      "step": 2248
+    },
+    {
+      "epoch": 20.017981651376147,
+      "grad_norm": 0.03922446072101593,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0011,
+      "step": 2249
+    },
+    {
+      "epoch": 20.01834862385321,
+      "grad_norm": 18.72905731201172,
+      "learning_rate": 3.125e-06,
+      "loss": 0.9872,
+      "step": 2250
+    },
+    {
+      "epoch": 20.018715596330274,
+      "grad_norm": 8.922704696655273,
+      "learning_rate": 3.125e-06,
+      "loss": 0.6939,
+      "step": 2251
+    },
+    {
+      "epoch": 20.019082568807338,
+      "grad_norm": 16.7719669342041,
+      "learning_rate": 3.125e-06,
+      "loss": 0.4258,
+      "step": 2252
+    },
+    {
+      "epoch": 20.019449541284402,
+      "grad_norm": 0.011279359459877014,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2253
+    },
+    {
+      "epoch": 20.01981651376147,
+      "grad_norm": 0.00902683287858963,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2254
+    },
+    {
+      "epoch": 20.020183486238533,
+      "grad_norm": 0.01984677091240883,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0006,
+      "step": 2255
+    },
+    {
+      "epoch": 20.020550458715597,
+      "grad_norm": 10.078396797180176,
+      "learning_rate": 3.125e-06,
+      "loss": 0.3291,
+      "step": 2256
+    },
+    {
+      "epoch": 20.02091743119266,
+      "grad_norm": 0.020954180508852005,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0006,
+      "step": 2257
+    },
+    {
+      "epoch": 20.021284403669725,
+      "grad_norm": 0.7649989724159241,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0043,
+      "step": 2258
+    },
+    {
+      "epoch": 20.02165137614679,
+      "grad_norm": 0.18087172508239746,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0025,
+      "step": 2259
+    },
+    {
+      "epoch": 20.022018348623853,
+      "grad_norm": 0.20101061463356018,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0033,
+      "step": 2260
+    },
+    {
+      "epoch": 20.022385321100916,
+      "grad_norm": 17.27302360534668,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0998,
+      "step": 2261
+    },
+    {
+      "epoch": 20.02275229357798,
+      "grad_norm": 0.04164685681462288,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0018,
+      "step": 2262
+    },
+    {
+      "epoch": 20.023119266055044,
+      "grad_norm": 14.77628231048584,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0546,
+      "step": 2263
+    },
+    {
+      "epoch": 20.02348623853211,
+      "grad_norm": 0.5247074365615845,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0066,
+      "step": 2264
+    },
+    {
+      "epoch": 20.023853211009175,
+      "grad_norm": 0.027248669415712357,
+      "learning_rate": 3.125e-06,
+      "loss": 0.001,
+      "step": 2265
+    },
+    {
+      "epoch": 20.02422018348624,
+      "grad_norm": 0.09920032322406769,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0011,
+      "step": 2266
+    },
+    {
+      "epoch": 20.024587155963303,
+      "grad_norm": 0.013501379638910294,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2267
+    },
+    {
+      "epoch": 20.024954128440367,
+      "grad_norm": 0.012602981179952621,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2268
+    },
+    {
+      "epoch": 20.02532110091743,
+      "grad_norm": 0.2184375822544098,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0028,
+      "step": 2269
+    },
+    {
+      "epoch": 20.025688073394495,
+      "grad_norm": 0.01530243456363678,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2270
+    },
+    {
+      "epoch": 20.02605504587156,
+      "grad_norm": 4.020350933074951,
+      "learning_rate": 3.125e-06,
+      "loss": 0.66,
+      "step": 2271
+    },
+    {
+      "epoch": 20.026422018348622,
+      "grad_norm": 0.028863701969385147,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0006,
+      "step": 2272
+    },
+    {
+      "epoch": 20.02678899082569,
+      "grad_norm": 0.007671784143894911,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2273
+    },
+    {
+      "epoch": 20.027155963302754,
+      "grad_norm": 0.03159390389919281,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2274
+    },
+    {
+      "epoch": 20.027522935779817,
+      "grad_norm": 0.004407484550029039,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0002,
+      "step": 2275
+    },
+    {
+      "epoch": 20.02788990825688,
+      "grad_norm": 0.014130991883575916,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2276
+    },
+    {
+      "epoch": 20.028256880733945,
+      "grad_norm": 0.04020540043711662,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0014,
+      "step": 2277
+    },
+    {
+      "epoch": 20.02862385321101,
+      "grad_norm": 0.10188324004411697,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0017,
+      "step": 2278
+    },
+    {
+      "epoch": 20.028990825688073,
+      "grad_norm": 1.178019642829895,
+      "learning_rate": 3.125e-06,
+      "loss": 0.012,
+      "step": 2279
+    },
+    {
+      "epoch": 20.029357798165137,
+      "grad_norm": 0.013761299662292004,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2280
+    },
+    {
+      "epoch": 20.0297247706422,
+      "grad_norm": 0.12450944632291794,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0036,
+      "step": 2281
+    },
+    {
+      "epoch": 20.030091743119264,
+      "grad_norm": 0.06028060242533684,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0012,
+      "step": 2282
+    },
+    {
+      "epoch": 20.030458715596332,
+      "grad_norm": 4.20280122756958,
+      "learning_rate": 3.125e-06,
+      "loss": 0.6289,
+      "step": 2283
+    },
+    {
+      "epoch": 20.030825688073396,
+      "grad_norm": 0.03613436594605446,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2284
+    },
+    {
+      "epoch": 20.03119266055046,
+      "grad_norm": 1.6654659509658813,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0135,
+      "step": 2285
+    },
+    {
+      "epoch": 20.031559633027523,
+      "grad_norm": 29.003992080688477,
+      "learning_rate": 3.125e-06,
+      "loss": 0.1189,
+      "step": 2286
+    },
+    {
+      "epoch": 20.031926605504587,
+      "grad_norm": 0.017878105863928795,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2287
+    },
+    {
+      "epoch": 20.03229357798165,
+      "grad_norm": 25.51042366027832,
+      "learning_rate": 3.125e-06,
+      "loss": 0.1581,
+      "step": 2288
+    },
+    {
+      "epoch": 20.032660550458715,
+      "grad_norm": 24.344806671142578,
+      "learning_rate": 3.125e-06,
+      "loss": 0.3504,
+      "step": 2289
+    },
+    {
+      "epoch": 20.03302752293578,
+      "grad_norm": 0.016436632722616196,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2290
+    },
+    {
+      "epoch": 20.033394495412843,
+      "grad_norm": 0.016158806160092354,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0006,
+      "step": 2291
+    },
+    {
+      "epoch": 20.033761467889907,
+      "grad_norm": 0.03189454227685928,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0009,
+      "step": 2292
+    },
+    {
+      "epoch": 20.034128440366974,
+      "grad_norm": 19.3234920501709,
+      "learning_rate": 3.125e-06,
+      "loss": 0.3106,
+      "step": 2293
+    },
+    {
+      "epoch": 20.034495412844038,
+      "grad_norm": 0.013577709905803204,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2294
+    },
+    {
+      "epoch": 20.0348623853211,
+      "grad_norm": 0.008525817655026913,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2295
+    },
+    {
+      "epoch": 20.035229357798165,
+      "grad_norm": 0.005974762607365847,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2296
+    },
+    {
+      "epoch": 20.03559633027523,
+      "grad_norm": 0.010409766808152199,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2297
+    },
+    {
+      "epoch": 20.035963302752293,
+      "grad_norm": 0.04464712738990784,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0007,
+      "step": 2298
+    },
+    {
+      "epoch": 20.036330275229357,
+      "grad_norm": 0.12531344592571259,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0017,
+      "step": 2299
+    },
+    {
+      "epoch": 20.03669724770642,
+      "grad_norm": 0.005685556214302778,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0002,
+      "step": 2300
+    },
+    {
+      "epoch": 20.037064220183485,
+      "grad_norm": 5.803192615509033,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0162,
+      "step": 2301
+    },
+    {
+      "epoch": 20.037431192660552,
+      "grad_norm": 0.1924077719449997,
+      "learning_rate": 3.125e-06,
+      "loss": 0.002,
+      "step": 2302
+    },
+    {
+      "epoch": 20.037798165137616,
+      "grad_norm": 0.017278678715229034,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2303
+    },
+    {
+      "epoch": 20.03816513761468,
+      "grad_norm": 0.023630280047655106,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0006,
+      "step": 2304
+    },
+    {
+      "epoch": 20.038532110091744,
+      "grad_norm": 0.16078397631645203,
+      "learning_rate": 3.125e-06,
+      "loss": 0.004,
+      "step": 2305
+    },
+    {
+      "epoch": 20.038899082568808,
+      "grad_norm": 0.5470094084739685,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0066,
+      "step": 2306
+    },
+    {
+      "epoch": 20.03926605504587,
+      "grad_norm": 0.012693501077592373,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2307
+    },
+    {
+      "epoch": 20.039633027522935,
+      "grad_norm": 4.726579189300537,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0209,
+      "step": 2308
+    },
+    {
+      "epoch": 20.04,
+      "grad_norm": 0.008185852319002151,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2309
+    },
+    {
+      "epoch": 20.040366972477063,
+      "grad_norm": 0.2242475152015686,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0032,
+      "step": 2310
+    },
+    {
+      "epoch": 20.040366972477063,
+      "eval_accuracy": 0.2988165680473373,
+      "eval_loss": 4.179484844207764,
+      "eval_runtime": 26.374,
+      "eval_samples_per_second": 12.816,
+      "eval_steps_per_second": 1.63,
+      "step": 2310
+    },
+    {
+      "epoch": 21.000366972477064,
+      "grad_norm": 0.32316264510154724,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0038,
+      "step": 2311
+    },
+    {
+      "epoch": 21.000733944954128,
+      "grad_norm": 0.24924685060977936,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0025,
+      "step": 2312
+    },
+    {
+      "epoch": 21.00110091743119,
+      "grad_norm": 0.010748972184956074,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2313
+    },
+    {
+      "epoch": 21.001467889908255,
+      "grad_norm": 0.059258315712213516,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0011,
+      "step": 2314
+    },
+    {
+      "epoch": 21.001834862385323,
+      "grad_norm": 0.02295132726430893,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0007,
+      "step": 2315
+    },
+    {
+      "epoch": 21.002201834862387,
+      "grad_norm": 32.10356521606445,
+      "learning_rate": 3.125e-06,
+      "loss": 0.1904,
+      "step": 2316
+    },
+    {
+      "epoch": 21.00256880733945,
+      "grad_norm": 0.039253003895282745,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0015,
+      "step": 2317
+    },
+    {
+      "epoch": 21.002935779816514,
+      "grad_norm": 0.021100062876939774,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0009,
+      "step": 2318
+    },
+    {
+      "epoch": 21.00330275229358,
+      "grad_norm": 0.00806365441530943,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2319
+    },
+    {
+      "epoch": 21.003669724770642,
+      "grad_norm": 10.04539680480957,
+      "learning_rate": 3.125e-06,
+      "loss": 0.627,
+      "step": 2320
+    },
+    {
+      "epoch": 21.004036697247706,
+      "grad_norm": 0.0954883024096489,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0032,
+      "step": 2321
+    },
+    {
+      "epoch": 21.00440366972477,
+      "grad_norm": 0.010831850580871105,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2322
+    },
+    {
+      "epoch": 21.004770642201834,
+      "grad_norm": 1.697843074798584,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0087,
+      "step": 2323
+    },
+    {
+      "epoch": 21.005137614678898,
+      "grad_norm": 1.969703197479248,
+      "learning_rate": 3.125e-06,
+      "loss": 0.024,
+      "step": 2324
+    },
+    {
+      "epoch": 21.005504587155965,
+      "grad_norm": 0.011999388225376606,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2325
+    },
+    {
+      "epoch": 21.00587155963303,
+      "grad_norm": 0.005680130794644356,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2326
+    },
+    {
+      "epoch": 21.006238532110093,
+      "grad_norm": 0.013913823291659355,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2327
+    },
+    {
+      "epoch": 21.006605504587156,
+      "grad_norm": 0.00876206811517477,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2328
+    },
+    {
+      "epoch": 21.00697247706422,
+      "grad_norm": 0.16155822575092316,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0021,
+      "step": 2329
+    },
+    {
+      "epoch": 21.007339449541284,
+      "grad_norm": 0.01608956791460514,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2330
+    },
+    {
+      "epoch": 21.007706422018348,
+      "grad_norm": 0.009029203094542027,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2331
+    },
+    {
+      "epoch": 21.008073394495412,
+      "grad_norm": 0.04165363311767578,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2332
+    },
+    {
+      "epoch": 21.008440366972476,
+      "grad_norm": 14.51718521118164,
+      "learning_rate": 3.125e-06,
+      "loss": 0.2787,
+      "step": 2333
+    },
+    {
+      "epoch": 21.00880733944954,
+      "grad_norm": 0.02146836742758751,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0007,
+      "step": 2334
+    },
+    {
+      "epoch": 21.009174311926607,
+      "grad_norm": 0.011820188723504543,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2335
+    },
+    {
+      "epoch": 21.00954128440367,
+      "grad_norm": 0.41864272952079773,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0054,
+      "step": 2336
+    },
+    {
+      "epoch": 21.009908256880735,
+      "grad_norm": 0.03901532292366028,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0007,
+      "step": 2337
+    },
+    {
+      "epoch": 21.0102752293578,
+      "grad_norm": 0.5663343667984009,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0046,
+      "step": 2338
+    },
+    {
+      "epoch": 21.010642201834862,
+      "grad_norm": 0.013743524439632893,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0006,
+      "step": 2339
+    },
+    {
+      "epoch": 21.011009174311926,
+      "grad_norm": 0.030926352366805077,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0011,
+      "step": 2340
+    },
+    {
+      "epoch": 21.01137614678899,
+      "grad_norm": 0.03486631065607071,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0006,
+      "step": 2341
+    },
+    {
+      "epoch": 21.011743119266054,
+      "grad_norm": 0.049033861607313156,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0011,
+      "step": 2342
+    },
+    {
+      "epoch": 21.012110091743118,
+      "grad_norm": 0.008756055496633053,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2343
+    },
+    {
+      "epoch": 21.012477064220185,
+      "grad_norm": 0.05314197763800621,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0012,
+      "step": 2344
+    },
+    {
+      "epoch": 21.01284403669725,
+      "grad_norm": 0.1176389828324318,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0027,
+      "step": 2345
+    },
+    {
+      "epoch": 21.013211009174313,
+      "grad_norm": 0.008223060518503189,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2346
+    },
+    {
+      "epoch": 21.013577981651377,
+      "grad_norm": 0.02039586752653122,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2347
+    },
+    {
+      "epoch": 21.01394495412844,
+      "grad_norm": 0.008805946446955204,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2348
+    },
+    {
+      "epoch": 21.014311926605505,
+      "grad_norm": 0.004297166597098112,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0002,
+      "step": 2349
+    },
+    {
+      "epoch": 21.01467889908257,
+      "grad_norm": 0.006547034252434969,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2350
+    },
+    {
+      "epoch": 21.015045871559632,
+      "grad_norm": 0.015044969506561756,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2351
+    },
+    {
+      "epoch": 21.015412844036696,
+      "grad_norm": 0.010934513993561268,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2352
+    },
+    {
+      "epoch": 21.01577981651376,
+      "grad_norm": 0.019999558106064796,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0006,
+      "step": 2353
+    },
+    {
+      "epoch": 21.016146788990827,
+      "grad_norm": 4.653348922729492,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0109,
+      "step": 2354
+    },
+    {
+      "epoch": 21.01651376146789,
+      "grad_norm": 0.5765218138694763,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0054,
+      "step": 2355
+    },
+    {
+      "epoch": 21.016880733944955,
+      "grad_norm": 164.21286010742188,
+      "learning_rate": 3.125e-06,
+      "loss": 0.3312,
+      "step": 2356
+    },
+    {
+      "epoch": 21.01724770642202,
+      "grad_norm": 0.004099734593182802,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0002,
+      "step": 2357
+    },
+    {
+      "epoch": 21.017614678899083,
+      "grad_norm": 4.578912734985352,
+      "learning_rate": 3.125e-06,
+      "loss": 0.6551,
+      "step": 2358
+    },
+    {
+      "epoch": 21.017981651376147,
+      "grad_norm": 0.051868364214897156,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0016,
+      "step": 2359
+    },
+    {
+      "epoch": 21.01834862385321,
+      "grad_norm": 0.05945153906941414,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0021,
+      "step": 2360
+    },
+    {
+      "epoch": 21.018715596330274,
+      "grad_norm": 0.06603698432445526,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0019,
+      "step": 2361
+    },
+    {
+      "epoch": 21.019082568807338,
+      "grad_norm": 0.19105669856071472,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0037,
+      "step": 2362
+    },
+    {
+      "epoch": 21.019449541284402,
+      "grad_norm": 0.020595677196979523,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0009,
+      "step": 2363
+    },
+    {
+      "epoch": 21.01981651376147,
+      "grad_norm": 0.08562837541103363,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0029,
+      "step": 2364
+    },
+    {
+      "epoch": 21.020183486238533,
+      "grad_norm": 0.02559935487806797,
+      "learning_rate": 3.125e-06,
+      "loss": 0.001,
+      "step": 2365
+    },
+    {
+      "epoch": 21.020550458715597,
+      "grad_norm": 0.0038835015147924423,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0002,
+      "step": 2366
+    },
+    {
+      "epoch": 21.02091743119266,
+      "grad_norm": 0.030159931629896164,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0012,
+      "step": 2367
+    },
+    {
+      "epoch": 21.021284403669725,
+      "grad_norm": 14.253634452819824,
+      "learning_rate": 3.125e-06,
+      "loss": 0.5473,
+      "step": 2368
+    },
+    {
+      "epoch": 21.02165137614679,
+      "grad_norm": 0.0158444382250309,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2369
+    },
+    {
+      "epoch": 21.022018348623853,
+      "grad_norm": 0.569360077381134,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0058,
+      "step": 2370
+    },
+    {
+      "epoch": 21.022385321100916,
+      "grad_norm": 0.06315754354000092,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0011,
+      "step": 2371
+    },
+    {
+      "epoch": 21.02275229357798,
+      "grad_norm": 0.01388737466186285,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2372
+    },
+    {
+      "epoch": 21.023119266055044,
+      "grad_norm": 0.062148209661245346,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0011,
+      "step": 2373
+    },
+    {
+      "epoch": 21.02348623853211,
+      "grad_norm": 0.006622489541769028,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2374
+    },
+    {
+      "epoch": 21.023853211009175,
+      "grad_norm": 28.1719913482666,
+      "learning_rate": 3.125e-06,
+      "loss": 0.3438,
+      "step": 2375
+    },
+    {
+      "epoch": 21.02422018348624,
+      "grad_norm": 0.013942455872893333,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2376
+    },
+    {
+      "epoch": 21.024587155963303,
+      "grad_norm": 0.07166176289319992,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0014,
+      "step": 2377
+    },
+    {
+      "epoch": 21.024954128440367,
+      "grad_norm": 0.008936182595789433,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2378
+    },
+    {
+      "epoch": 21.02532110091743,
+      "grad_norm": 0.06209037825465202,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0018,
+      "step": 2379
+    },
+    {
+      "epoch": 21.025688073394495,
+      "grad_norm": 0.005039326846599579,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0002,
+      "step": 2380
+    },
+    {
+      "epoch": 21.02605504587156,
+      "grad_norm": 0.005936936009675264,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0003,
+      "step": 2381
+    },
+    {
+      "epoch": 21.026422018348622,
+      "grad_norm": 0.0180437620729208,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0006,
+      "step": 2382
+    },
+    {
+      "epoch": 21.02678899082569,
+      "grad_norm": 0.026367709040641785,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2383
+    },
+    {
+      "epoch": 21.027155963302754,
+      "grad_norm": 0.013204758055508137,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2384
+    },
+    {
+      "epoch": 21.027522935779817,
+      "grad_norm": 0.056551527231931686,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0013,
+      "step": 2385
+    },
+    {
+      "epoch": 21.02788990825688,
+      "grad_norm": 0.009397230111062527,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2386
+    },
+    {
+      "epoch": 21.028256880733945,
+      "grad_norm": 0.007460306398570538,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0002,
+      "step": 2387
+    },
+    {
+      "epoch": 21.02862385321101,
+      "grad_norm": 0.01718973182141781,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2388
+    },
+    {
+      "epoch": 21.028990825688073,
+      "grad_norm": 0.08031223714351654,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0013,
+      "step": 2389
+    },
+    {
+      "epoch": 21.029357798165137,
+      "grad_norm": 0.09328863024711609,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0028,
+      "step": 2390
+    },
+    {
+      "epoch": 21.0297247706422,
+      "grad_norm": 0.010707106441259384,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2391
+    },
+    {
+      "epoch": 21.030091743119264,
+      "grad_norm": 0.33766600489616394,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0061,
+      "step": 2392
+    },
+    {
+      "epoch": 21.030458715596332,
+      "grad_norm": 0.004179957788437605,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0002,
+      "step": 2393
+    },
+    {
+      "epoch": 21.030825688073396,
+      "grad_norm": 0.02968457154929638,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0006,
+      "step": 2394
+    },
+    {
+      "epoch": 21.03119266055046,
+      "grad_norm": 0.011578693985939026,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2395
+    },
+    {
+      "epoch": 21.031559633027523,
+      "grad_norm": 0.011557463556528091,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2396
+    },
+    {
+      "epoch": 21.031926605504587,
+      "grad_norm": 19.778005599975586,
+      "learning_rate": 3.125e-06,
+      "loss": 0.5286,
+      "step": 2397
+    },
+    {
+      "epoch": 21.03229357798165,
+      "grad_norm": 0.059479694813489914,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0012,
+      "step": 2398
+    },
+    {
+      "epoch": 21.032660550458715,
+      "grad_norm": 0.017460443079471588,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2399
+    },
+    {
+      "epoch": 21.03302752293578,
+      "grad_norm": 0.02237357385456562,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0006,
+      "step": 2400
+    },
+    {
+      "epoch": 21.033394495412843,
+      "grad_norm": 0.004192739725112915,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0002,
+      "step": 2401
+    },
+    {
+      "epoch": 21.033761467889907,
+      "grad_norm": 0.04200279340147972,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0011,
+      "step": 2402
+    },
+    {
+      "epoch": 21.034128440366974,
+      "grad_norm": 0.1609215885400772,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0035,
+      "step": 2403
+    },
+    {
+      "epoch": 21.034495412844038,
+      "grad_norm": 0.4650096297264099,
+      "learning_rate": 3.125e-06,
+      "loss": 0.005,
+      "step": 2404
+    },
+    {
+      "epoch": 21.0348623853211,
+      "grad_norm": 16.532907485961914,
+      "learning_rate": 3.125e-06,
+      "loss": 0.192,
+      "step": 2405
+    },
+    {
+      "epoch": 21.035229357798165,
+      "grad_norm": 2.1189827919006348,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0126,
+      "step": 2406
+    },
+    {
+      "epoch": 21.03559633027523,
+      "grad_norm": 0.0033829419407993555,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0002,
+      "step": 2407
+    },
+    {
+      "epoch": 21.035963302752293,
+      "grad_norm": 0.03969793766736984,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0007,
+      "step": 2408
+    },
+    {
+      "epoch": 21.036330275229357,
+      "grad_norm": 0.014873363077640533,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2409
+    },
+    {
+      "epoch": 21.03669724770642,
+      "grad_norm": 0.04782862961292267,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0009,
+      "step": 2410
+    },
+    {
+      "epoch": 21.037064220183485,
+      "grad_norm": 0.010236002504825592,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0004,
+      "step": 2411
+    },
+    {
+      "epoch": 21.037431192660552,
+      "grad_norm": 0.011951657012104988,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0005,
+      "step": 2412
+    },
+    {
+      "epoch": 21.037798165137616,
+      "grad_norm": 0.07350239902734756,
+      "learning_rate": 3.125e-06,
+      "loss": 0.001,
+      "step": 2413
+    },
+    {
+      "epoch": 21.03816513761468,
+      "grad_norm": 0.02717510797083378,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0006,
+      "step": 2414
+    },
+    {
+      "epoch": 21.038532110091744,
+      "grad_norm": 0.03147394210100174,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0007,
+      "step": 2415
+    },
+    {
+      "epoch": 21.038899082568808,
+      "grad_norm": 0.03789057210087776,
+      "learning_rate": 3.125e-06,
+      "loss": 0.001,
+      "step": 2416
+    },
+    {
+      "epoch": 21.03926605504587,
+      "grad_norm": 0.06214606761932373,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0015,
+      "step": 2417
+    },
+    {
+      "epoch": 21.039633027522935,
+      "grad_norm": 40.12753677368164,
+      "learning_rate": 3.125e-06,
+      "loss": 0.1812,
+      "step": 2418
+    },
+    {
+      "epoch": 21.04,
+      "grad_norm": 0.017769725993275642,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0007,
+      "step": 2419
+    },
+    {
+      "epoch": 21.040366972477063,
+      "grad_norm": 0.017028233036398888,
+      "learning_rate": 3.125e-06,
+      "loss": 0.0006,
+      "step": 2420
+    },
+    {
+      "epoch": 21.040366972477063,
+      "eval_accuracy": 0.3136094674556213,
+      "eval_loss": 4.161154270172119,
+      "eval_runtime": 26.8749,
+      "eval_samples_per_second": 12.577,
+      "eval_steps_per_second": 1.6,
+      "step": 2420
+    },
+    {
+      "epoch": 21.040366972477063,
+      "step": 2420,
+      "total_flos": 2.4096397031204585e+19,
+      "train_loss": 0.11396398439318252,
+      "train_runtime": 4301.4449,
+      "train_samples_per_second": 5.068,
+      "train_steps_per_second": 0.634
+    },
+    {
+      "epoch": 21.040366972477063,
+      "eval_accuracy": 0.4913294797687861,
+      "eval_loss": 4.264735698699951,
+      "eval_runtime": 27.0002,
+      "eval_samples_per_second": 12.815,
+      "eval_steps_per_second": 1.63,
+      "step": 2420
+    },
+    {
+      "epoch": 21.040366972477063,
+      "eval_accuracy": 0.4913294797687861,
+      "eval_loss": 4.264735221862793,
+      "eval_runtime": 26.3705,
+      "eval_samples_per_second": 13.121,
+      "eval_steps_per_second": 1.669,
+      "step": 2420
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 2725,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 10,
+        "early_stopping_threshold": 0.0
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 10
+      }
+    },
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.4096397031204585e+19,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}