{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.471956224350206,
  "eval_steps": 500,
  "global_step": 12000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.022799817601459188,
      "grad_norm": 2.5625,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 4.9225,
      "step": 50
    },
    {
      "epoch": 0.045599635202918376,
      "grad_norm": 3.546875,
      "learning_rate": 6.666666666666667e-05,
      "loss": 2.1478,
      "step": 100
    },
    {
      "epoch": 0.06839945280437756,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0001,
      "loss": 0.6361,
      "step": 150
    },
    {
      "epoch": 0.09119927040583675,
      "grad_norm": 1.46875,
      "learning_rate": 0.00013333333333333334,
      "loss": 0.467,
      "step": 200
    },
    {
      "epoch": 0.11399908800729594,
      "grad_norm": 2.15625,
      "learning_rate": 0.0001666666666666667,
      "loss": 0.4381,
      "step": 250
    },
    {
      "epoch": 0.13679890560875513,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0002,
      "loss": 0.3365,
      "step": 300
    },
    {
      "epoch": 0.15959872321021432,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00019988560970029743,
      "loss": 0.3255,
      "step": 350
    },
    {
      "epoch": 0.1823985408116735,
      "grad_norm": 1.40625,
      "learning_rate": 0.00019977121940059484,
      "loss": 0.2946,
      "step": 400
    },
    {
      "epoch": 0.2051983584131327,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00019965682910089226,
      "loss": 0.3023,
      "step": 450
    },
    {
      "epoch": 0.22799817601459188,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00019954243880118967,
      "loss": 0.2708,
      "step": 500
    },
    {
      "epoch": 0.2507979936160511,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001994280485014871,
      "loss": 0.3645,
      "step": 550
    },
    {
      "epoch": 0.27359781121751026,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00019931365820178448,
      "loss": 0.3072,
      "step": 600
    },
    {
      "epoch": 0.29639762881896947,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00019919926790208192,
      "loss": 0.2891,
      "step": 650
    },
    {
      "epoch": 0.31919744642042863,
      "grad_norm": 2.015625,
      "learning_rate": 0.00019908487760237934,
      "loss": 0.2763,
      "step": 700
    },
    {
      "epoch": 0.34199726402188785,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00019897048730267673,
      "loss": 0.2824,
      "step": 750
    },
    {
      "epoch": 0.364797081623347,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00019885609700297417,
      "loss": 0.3007,
      "step": 800
    },
    {
      "epoch": 0.3875968992248062,
      "grad_norm": 1.359375,
      "learning_rate": 0.00019874170670327156,
      "loss": 0.2955,
      "step": 850
    },
    {
      "epoch": 0.4103967168262654,
      "grad_norm": 1.078125,
      "learning_rate": 0.000198627316403569,
      "loss": 0.2385,
      "step": 900
    },
    {
      "epoch": 0.4331965344277246,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001985129261038664,
      "loss": 0.2461,
      "step": 950
    },
    {
      "epoch": 0.45599635202918376,
      "grad_norm": 1.421875,
      "learning_rate": 0.0001983985358041638,
      "loss": 0.2567,
      "step": 1000
    },
    {
      "epoch": 0.478796169630643,
      "grad_norm": 1.25,
      "learning_rate": 0.00019828414550446125,
      "loss": 0.2189,
      "step": 1050
    },
    {
      "epoch": 0.5015959872321022,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019816975520475864,
      "loss": 0.272,
      "step": 1100
    },
    {
      "epoch": 0.5243958048335613,
      "grad_norm": 1.78125,
      "learning_rate": 0.00019805536490505606,
      "loss": 0.2615,
      "step": 1150
    },
    {
      "epoch": 0.5471956224350205,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019794097460535347,
      "loss": 0.2476,
      "step": 1200
    },
    {
      "epoch": 0.5699954400364797,
      "grad_norm": 1.328125,
      "learning_rate": 0.0001978265843056509,
      "loss": 0.2327,
      "step": 1250
    },
    {
      "epoch": 0.5927952576379389,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001977121940059483,
      "loss": 0.2664,
      "step": 1300
    },
    {
      "epoch": 0.615595075239398,
      "grad_norm": 1.109375,
      "learning_rate": 0.00019759780370624572,
      "loss": 0.2266,
      "step": 1350
    },
    {
      "epoch": 0.6383948928408573,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00019748341340654314,
      "loss": 0.2234,
      "step": 1400
    },
    {
      "epoch": 0.6611947104423165,
      "grad_norm": 1.1875,
      "learning_rate": 0.00019736902310684055,
      "loss": 0.2463,
      "step": 1450
    },
    {
      "epoch": 0.6839945280437757,
      "grad_norm": 1.265625,
      "learning_rate": 0.00019725463280713797,
      "loss": 0.2311,
      "step": 1500
    },
    {
      "epoch": 0.7067943456452348,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019714024250743536,
      "loss": 0.2251,
      "step": 1550
    },
    {
      "epoch": 0.729594163246694,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001970258522077328,
      "loss": 0.2492,
      "step": 1600
    },
    {
      "epoch": 0.7523939808481532,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00019691146190803022,
      "loss": 0.2554,
      "step": 1650
    },
    {
      "epoch": 0.7751937984496124,
      "grad_norm": 0.8125,
      "learning_rate": 0.00019679707160832763,
      "loss": 0.2384,
      "step": 1700
    },
    {
      "epoch": 0.7979936160510716,
      "grad_norm": 1.3125,
      "learning_rate": 0.00019668268130862505,
      "loss": 0.2583,
      "step": 1750
    },
    {
      "epoch": 0.8207934336525308,
      "grad_norm": 0.625,
      "learning_rate": 0.00019656829100892244,
      "loss": 0.2343,
      "step": 1800
    },
    {
      "epoch": 0.84359325125399,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019645390070921988,
      "loss": 0.2193,
      "step": 1850
    },
    {
      "epoch": 0.8663930688554492,
      "grad_norm": 0.875,
      "learning_rate": 0.00019633951040951727,
      "loss": 0.2212,
      "step": 1900
    },
    {
      "epoch": 0.8891928864569083,
      "grad_norm": 1.578125,
      "learning_rate": 0.00019622512010981468,
      "loss": 0.2178,
      "step": 1950
    },
    {
      "epoch": 0.9119927040583675,
      "grad_norm": 0.8125,
      "learning_rate": 0.00019611072981011213,
      "loss": 0.2135,
      "step": 2000
    },
    {
      "epoch": 0.9347925216598267,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019599633951040952,
      "loss": 0.2522,
      "step": 2050
    },
    {
      "epoch": 0.957592339261286,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019588194921070696,
      "loss": 0.2138,
      "step": 2100
    },
    {
      "epoch": 0.9803921568627451,
      "grad_norm": 1.3125,
      "learning_rate": 0.00019576755891100435,
      "loss": 0.2043,
      "step": 2150
    },
    {
      "epoch": 1.0031919744642044,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019565316861130177,
      "loss": 0.1855,
      "step": 2200
    },
    {
      "epoch": 1.0259917920656634,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00019553877831159918,
      "loss": 0.2099,
      "step": 2250
    },
    {
      "epoch": 1.0487916096671226,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001954243880118966,
      "loss": 0.2032,
      "step": 2300
    },
    {
      "epoch": 1.0715914272685818,
      "grad_norm": 0.91015625,
      "learning_rate": 0.000195309997712194,
      "loss": 0.1753,
      "step": 2350
    },
    {
      "epoch": 1.094391244870041,
      "grad_norm": 1.359375,
      "learning_rate": 0.00019519560741249143,
      "loss": 0.1997,
      "step": 2400
    },
    {
      "epoch": 1.1171910624715002,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00019508121711278885,
      "loss": 0.2101,
      "step": 2450
    },
    {
      "epoch": 1.1399908800729595,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00019496682681308626,
      "loss": 0.1734,
      "step": 2500
    },
    {
      "epoch": 1.1627906976744187,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019485243651338368,
      "loss": 0.1716,
      "step": 2550
    },
    {
      "epoch": 1.1855905152758779,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001947380462136811,
      "loss": 0.1588,
      "step": 2600
    },
    {
      "epoch": 1.2083903328773369,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001946236559139785,
      "loss": 0.166,
      "step": 2650
    },
    {
      "epoch": 1.231190150478796,
      "grad_norm": 1.296875,
      "learning_rate": 0.00019450926561427593,
      "loss": 0.1514,
      "step": 2700
    },
    {
      "epoch": 1.2539899680802553,
      "grad_norm": 1.3125,
      "learning_rate": 0.00019439487531457334,
      "loss": 0.2103,
      "step": 2750
    },
    {
      "epoch": 1.2767897856817145,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00019428048501487076,
      "loss": 0.185,
      "step": 2800
    },
    {
      "epoch": 1.2995896032831737,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019416609471516815,
      "loss": 0.1498,
      "step": 2850
    },
    {
      "epoch": 1.322389420884633,
      "grad_norm": 1.28125,
      "learning_rate": 0.0001940517044154656,
      "loss": 0.1706,
      "step": 2900
    },
    {
      "epoch": 1.3451892384860922,
      "grad_norm": 1.515625,
      "learning_rate": 0.00019393731411576298,
      "loss": 0.1579,
      "step": 2950
    },
    {
      "epoch": 1.3679890560875512,
      "grad_norm": 1.28125,
      "learning_rate": 0.0001938229238160604,
      "loss": 0.1777,
      "step": 3000
    },
    {
      "epoch": 1.3907888736890106,
      "grad_norm": 0.734375,
      "learning_rate": 0.00019370853351635784,
      "loss": 0.168,
      "step": 3050
    },
    {
      "epoch": 1.4135886912904696,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019359414321665523,
      "loss": 0.1325,
      "step": 3100
    },
    {
      "epoch": 1.4363885088919288,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00019347975291695267,
      "loss": 0.1483,
      "step": 3150
    },
    {
      "epoch": 1.459188326493388,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019336536261725006,
      "loss": 0.1424,
      "step": 3200
    },
    {
      "epoch": 1.4819881440948472,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019325097231754748,
      "loss": 0.1352,
      "step": 3250
    },
    {
      "epoch": 1.5047879616963065,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0001931365820178449,
      "loss": 0.1583,
      "step": 3300
    },
    {
      "epoch": 1.5275877792977655,
      "grad_norm": 1.40625,
      "learning_rate": 0.0001930221917181423,
      "loss": 0.1496,
      "step": 3350
    },
    {
      "epoch": 1.550387596899225,
      "grad_norm": 1.578125,
      "learning_rate": 0.00019290780141843972,
      "loss": 0.1551,
      "step": 3400
    },
    {
      "epoch": 1.573187414500684,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019279341111873714,
      "loss": 0.1275,
      "step": 3450
    },
    {
      "epoch": 1.5959872321021433,
      "grad_norm": 1.25,
      "learning_rate": 0.00019267902081903456,
      "loss": 0.1685,
      "step": 3500
    },
    {
      "epoch": 1.6187870497036023,
      "grad_norm": 1.171875,
      "learning_rate": 0.00019256463051933197,
      "loss": 0.1271,
      "step": 3550
    },
    {
      "epoch": 1.6415868673050615,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001924502402196294,
      "loss": 0.1344,
      "step": 3600
    },
    {
      "epoch": 1.6643866849065208,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001923358499199268,
      "loss": 0.1516,
      "step": 3650
    },
    {
      "epoch": 1.68718650250798,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00019222145962022422,
      "loss": 0.1379,
      "step": 3700
    },
    {
      "epoch": 1.7099863201094392,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00019210706932052164,
      "loss": 0.1396,
      "step": 3750
    },
    {
      "epoch": 1.7327861377108982,
      "grad_norm": 1.25,
      "learning_rate": 0.00019199267902081902,
      "loss": 0.1544,
      "step": 3800
    },
    {
      "epoch": 1.7555859553123576,
      "grad_norm": 1.390625,
      "learning_rate": 0.00019187828872111647,
      "loss": 0.1639,
      "step": 3850
    },
    {
      "epoch": 1.7783857729138166,
      "grad_norm": 1.734375,
      "learning_rate": 0.00019176389842141386,
      "loss": 0.1473,
      "step": 3900
    },
    {
      "epoch": 1.8011855905152758,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001916495081217113,
      "loss": 0.1606,
      "step": 3950
    },
    {
      "epoch": 1.823985408116735,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00019153511782200872,
      "loss": 0.155,
      "step": 4000
    },
    {
      "epoch": 1.8467852257181943,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001914207275223061,
      "loss": 0.1222,
      "step": 4050
    },
    {
      "epoch": 1.8695850433196535,
      "grad_norm": 1.828125,
      "learning_rate": 0.00019130633722260355,
      "loss": 0.1476,
      "step": 4100
    },
    {
      "epoch": 1.8923848609211125,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00019119194692290094,
      "loss": 0.1358,
      "step": 4150
    },
    {
      "epoch": 1.915184678522572,
      "grad_norm": 1.59375,
      "learning_rate": 0.00019107755662319838,
      "loss": 0.1439,
      "step": 4200
    },
    {
      "epoch": 1.937984496124031,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00019096316632349577,
      "loss": 0.1699,
      "step": 4250
    },
    {
      "epoch": 1.9607843137254903,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019084877602379319,
      "loss": 0.1364,
      "step": 4300
    },
    {
      "epoch": 1.9835841313269493,
      "grad_norm": 1.328125,
      "learning_rate": 0.0001907343857240906,
      "loss": 0.1242,
      "step": 4350
    },
    {
      "epoch": 2.0063839489284088,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00019061999542438802,
      "loss": 0.1146,
      "step": 4400
    },
    {
      "epoch": 2.0291837665298678,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00019050560512468543,
      "loss": 0.1327,
      "step": 4450
    },
    {
      "epoch": 2.0519835841313268,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019039121482498285,
      "loss": 0.1278,
      "step": 4500
    },
    {
      "epoch": 2.074783401732786,
      "grad_norm": 1.484375,
      "learning_rate": 0.00019027682452528027,
      "loss": 0.1069,
      "step": 4550
    },
    {
      "epoch": 2.097583219334245,
      "grad_norm": 1.3125,
      "learning_rate": 0.00019016243422557768,
      "loss": 0.1351,
      "step": 4600
    },
    {
      "epoch": 2.1203830369357046,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0001900480439258751,
      "loss": 0.1372,
      "step": 4650
    },
    {
      "epoch": 2.1431828545371636,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0001899336536261725,
      "loss": 0.1093,
      "step": 4700
    },
    {
      "epoch": 2.165982672138623,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00018981926332646993,
      "loss": 0.1,
      "step": 4750
    },
    {
      "epoch": 2.188782489740082,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018970487302676735,
      "loss": 0.1016,
      "step": 4800
    },
    {
      "epoch": 2.2115823073415415,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00018959048272706473,
      "loss": 0.0976,
      "step": 4850
    },
    {
      "epoch": 2.2343821249430005,
      "grad_norm": 1.7734375,
      "learning_rate": 0.00018947609242736218,
      "loss": 0.1132,
      "step": 4900
    },
    {
      "epoch": 2.2571819425444595,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018936170212765957,
      "loss": 0.1174,
      "step": 4950
    },
    {
      "epoch": 2.279981760145919,
      "grad_norm": 1.59375,
      "learning_rate": 0.000189247311827957,
      "loss": 0.1191,
      "step": 5000
    },
    {
      "epoch": 2.302781577747378,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00018913292152825443,
      "loss": 0.0933,
      "step": 5050
    },
    {
      "epoch": 2.3255813953488373,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00018901853122855181,
      "loss": 0.1065,
      "step": 5100
    },
    {
      "epoch": 2.3483812129502963,
      "grad_norm": 0.8125,
      "learning_rate": 0.00018890414092884926,
      "loss": 0.0993,
      "step": 5150
    },
    {
      "epoch": 2.3711810305517558,
      "grad_norm": 1.265625,
      "learning_rate": 0.00018878975062914665,
      "loss": 0.1179,
      "step": 5200
    },
    {
      "epoch": 2.3939808481532148,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00018867536032944406,
      "loss": 0.0982,
      "step": 5250
    },
    {
      "epoch": 2.4167806657546738,
      "grad_norm": 1.203125,
      "learning_rate": 0.00018856097002974148,
      "loss": 0.0831,
      "step": 5300
    },
    {
      "epoch": 2.439580483356133,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001884465797300389,
      "loss": 0.0943,
      "step": 5350
    },
    {
      "epoch": 2.462380300957592,
      "grad_norm": 1.484375,
      "learning_rate": 0.00018833218943033634,
      "loss": 0.088,
      "step": 5400
    },
    {
      "epoch": 2.4851801185590516,
      "grad_norm": 1.59375,
      "learning_rate": 0.00018821779913063373,
      "loss": 0.0952,
      "step": 5450
    },
    {
      "epoch": 2.5079799361605106,
      "grad_norm": 1.7421875,
      "learning_rate": 0.00018810340883093114,
      "loss": 0.0952,
      "step": 5500
    },
    {
      "epoch": 2.53077975376197,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00018798901853122856,
      "loss": 0.0901,
      "step": 5550
    },
    {
      "epoch": 2.553579571363429,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00018787462823152598,
      "loss": 0.0989,
      "step": 5600
    },
    {
      "epoch": 2.576379388964888,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001877602379318234,
      "loss": 0.0794,
      "step": 5650
    },
    {
      "epoch": 2.5991792065663475,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001876458476321208,
      "loss": 0.1076,
      "step": 5700
    },
    {
      "epoch": 2.621979024167807,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00018753145733241822,
      "loss": 0.0745,
      "step": 5750
    },
    {
      "epoch": 2.644778841769266,
      "grad_norm": 0.6875,
      "learning_rate": 0.00018741706703271564,
      "loss": 0.0938,
      "step": 5800
    },
    {
      "epoch": 2.667578659370725,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00018730267673301306,
      "loss": 0.0888,
      "step": 5850
    },
    {
      "epoch": 2.6903784769721844,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00018718828643331044,
      "loss": 0.0836,
      "step": 5900
    },
    {
      "epoch": 2.7131782945736433,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001870738961336079,
      "loss": 0.0821,
      "step": 5950
    },
    {
      "epoch": 2.7359781121751023,
      "grad_norm": 0.875,
      "learning_rate": 0.0001869595058339053,
      "loss": 0.1043,
      "step": 6000
    },
    {
      "epoch": 2.7587779297765618,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00018684511553420272,
      "loss": 0.1005,
      "step": 6050
    },
    {
      "epoch": 2.781577747378021,
      "grad_norm": 2.3125,
      "learning_rate": 0.00018673072523450014,
      "loss": 0.1028,
      "step": 6100
    },
    {
      "epoch": 2.80437756497948,
      "grad_norm": 1.9453125,
      "learning_rate": 0.00018661633493479752,
      "loss": 0.1046,
      "step": 6150
    },
    {
      "epoch": 2.827177382580939,
      "grad_norm": 0.875,
      "learning_rate": 0.00018650194463509497,
      "loss": 0.0964,
      "step": 6200
    },
    {
      "epoch": 2.8499772001823986,
      "grad_norm": 1.421875,
      "learning_rate": 0.00018638755433539236,
      "loss": 0.0775,
      "step": 6250
    },
    {
      "epoch": 2.8727770177838576,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00018627316403568977,
      "loss": 0.0963,
      "step": 6300
    },
    {
      "epoch": 2.895576835385317,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001861587737359872,
      "loss": 0.0903,
      "step": 6350
    },
    {
      "epoch": 2.918376652986776,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001860443834362846,
      "loss": 0.0914,
      "step": 6400
    },
    {
      "epoch": 2.9411764705882355,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00018592999313658205,
      "loss": 0.108,
      "step": 6450
    },
    {
      "epoch": 2.9639762881896945,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00018581560283687944,
      "loss": 0.0876,
      "step": 6500
    },
    {
      "epoch": 2.9867761057911535,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00018570121253717685,
      "loss": 0.0834,
      "step": 6550
    },
    {
      "epoch": 3.009575923392613,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00018558682223747427,
      "loss": 0.0786,
      "step": 6600
    },
    {
      "epoch": 3.032375740994072,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00018547243193777169,
      "loss": 0.0803,
      "step": 6650
    },
    {
      "epoch": 3.0551755585955314,
      "grad_norm": 1.59375,
      "learning_rate": 0.0001853580416380691,
      "loss": 0.0846,
      "step": 6700
    },
    {
      "epoch": 3.0779753761969904,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00018524365133836652,
      "loss": 0.0562,
      "step": 6750
    },
    {
      "epoch": 3.10077519379845,
      "grad_norm": 1.9765625,
      "learning_rate": 0.00018512926103866393,
      "loss": 0.0934,
      "step": 6800
    },
    {
      "epoch": 3.123575011399909,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00018501487073896135,
      "loss": 0.0813,
      "step": 6850
    },
    {
      "epoch": 3.146374829001368,
      "grad_norm": 1.171875,
      "learning_rate": 0.00018490048043925877,
      "loss": 0.0661,
      "step": 6900
    },
    {
      "epoch": 3.169174646602827,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00018478609013955615,
      "loss": 0.0589,
      "step": 6950
    },
    {
      "epoch": 3.191974464204286,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0001846716998398536,
      "loss": 0.0679,
      "step": 7000
    },
    {
      "epoch": 3.2147742818057456,
      "grad_norm": 1.515625,
      "learning_rate": 0.000184557309540151,
      "loss": 0.0564,
      "step": 7050
    },
    {
      "epoch": 3.2375740994072046,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0001844429192404484,
      "loss": 0.0763,
      "step": 7100
    },
    {
      "epoch": 3.260373917008664,
      "grad_norm": 1.953125,
      "learning_rate": 0.00018432852894074585,
      "loss": 0.0768,
      "step": 7150
    },
    {
      "epoch": 3.283173734610123,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00018421413864104323,
      "loss": 0.0686,
      "step": 7200
    },
    {
      "epoch": 3.305973552211582,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018409974834134068,
      "loss": 0.0638,
      "step": 7250
    },
    {
      "epoch": 3.3287733698130415,
      "grad_norm": 0.275390625,
      "learning_rate": 0.00018398535804163807,
      "loss": 0.0643,
      "step": 7300
    },
    {
      "epoch": 3.3515731874145005,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018387096774193548,
      "loss": 0.067,
      "step": 7350
    },
    {
      "epoch": 3.37437300501596,
      "grad_norm": 1.28125,
      "learning_rate": 0.00018375657744223293,
      "loss": 0.0756,
      "step": 7400
    },
    {
      "epoch": 3.397172822617419,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00018364218714253031,
      "loss": 0.0579,
      "step": 7450
    },
    {
      "epoch": 3.4199726402188784,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00018352779684282773,
      "loss": 0.0533,
      "step": 7500
    },
    {
      "epoch": 3.4427724578203374,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00018341340654312515,
      "loss": 0.0569,
      "step": 7550
    },
    {
      "epoch": 3.465572275421797,
      "grad_norm": 1.328125,
      "learning_rate": 0.00018329901624342256,
      "loss": 0.0593,
      "step": 7600
    },
    {
      "epoch": 3.488372093023256,
      "grad_norm": 1.859375,
      "learning_rate": 0.00018318462594371998,
      "loss": 0.0664,
      "step": 7650
    },
    {
      "epoch": 3.5111719106247152,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001830702356440174,
      "loss": 0.0593,
      "step": 7700
    },
    {
      "epoch": 3.5339717282261742,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001829558453443148,
      "loss": 0.063,
      "step": 7750
    },
    {
      "epoch": 3.556771545827633,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.00018284145504461223,
      "loss": 0.0553,
      "step": 7800
    },
    {
      "epoch": 3.5795713634290927,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00018272706474490964,
      "loss": 0.0513,
      "step": 7850
    },
    {
      "epoch": 3.6023711810305517,
      "grad_norm": 0.421875,
      "learning_rate": 0.00018261267444520706,
      "loss": 0.0678,
      "step": 7900
    },
    {
      "epoch": 3.625170998632011,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00018249828414550448,
      "loss": 0.0446,
      "step": 7950
    },
    {
      "epoch": 3.64797081623347,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001823838938458019,
      "loss": 0.0588,
      "step": 8000
    },
    {
      "epoch": 3.6707706338349295,
      "grad_norm": 0.2734375,
      "learning_rate": 0.0001822695035460993,
      "loss": 0.0564,
      "step": 8050
    },
    {
      "epoch": 3.6935704514363885,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00018215511324639672,
      "loss": 0.0526,
      "step": 8100
    },
    {
      "epoch": 3.7163702690378475,
      "grad_norm": 1.765625,
      "learning_rate": 0.0001820407229466941,
      "loss": 0.0609,
      "step": 8150
    },
    {
      "epoch": 3.739170086639307,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00018192633264699156,
      "loss": 0.0679,
      "step": 8200
    },
    {
      "epoch": 3.761969904240766,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00018181194234728894,
      "loss": 0.0638,
      "step": 8250
    },
    {
      "epoch": 3.7847697218422254,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001816975520475864,
      "loss": 0.0647,
      "step": 8300
    },
    {
      "epoch": 3.8075695394436844,
      "grad_norm": 0.84375,
      "learning_rate": 0.00018158316174788378,
      "loss": 0.0747,
      "step": 8350
    },
    {
      "epoch": 3.830369357045144,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001814687714481812,
      "loss": 0.0586,
      "step": 8400
    },
    {
      "epoch": 3.853169174646603,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00018135438114847864,
      "loss": 0.0544,
      "step": 8450
    },
    {
      "epoch": 3.875968992248062,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00018123999084877602,
      "loss": 0.0644,
      "step": 8500
    },
    {
      "epoch": 3.8987688098495212,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00018112560054907344,
      "loss": 0.0559,
      "step": 8550
    },
    {
      "epoch": 3.9215686274509802,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00018101121024937086,
      "loss": 0.0608,
      "step": 8600
    },
    {
      "epoch": 3.9443684450524397,
      "grad_norm": 1.578125,
      "learning_rate": 0.00018089681994966827,
      "loss": 0.0688,
      "step": 8650
    },
    {
      "epoch": 3.9671682626538987,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001807824296499657,
      "loss": 0.0614,
      "step": 8700
    },
    {
      "epoch": 3.989968080255358,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001806680393502631,
      "loss": 0.0515,
      "step": 8750
    },
    {
      "epoch": 4.0127678978568175,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018055364905056052,
      "loss": 0.0527,
      "step": 8800
    },
    {
      "epoch": 4.035567715458276,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00018043925875085794,
      "loss": 0.0567,
      "step": 8850
    },
    {
      "epoch": 4.0583675330597355,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018032486845115535,
      "loss": 0.0501,
      "step": 8900
    },
    {
      "epoch": 4.081167350661195,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018021047815145274,
      "loss": 0.0329,
      "step": 8950
    },
    {
      "epoch": 4.1039671682626535,
      "grad_norm": 1.453125,
      "learning_rate": 0.00018009608785175019,
      "loss": 0.0615,
      "step": 9000
    },
    {
      "epoch": 4.126766985864113,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0001799816975520476,
      "loss": 0.0471,
      "step": 9050
    },
    {
      "epoch": 4.149566803465572,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00017986730725234502,
      "loss": 0.0467,
      "step": 9100
    },
    {
      "epoch": 4.172366621067032,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00017975291695264243,
      "loss": 0.0381,
      "step": 9150
    },
    {
      "epoch": 4.19516643866849,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00017963852665293982,
      "loss": 0.0414,
      "step": 9200
    },
    {
      "epoch": 4.21796625626995,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00017952413635323727,
      "loss": 0.0412,
      "step": 9250
    },
    {
      "epoch": 4.240766073871409,
      "grad_norm": 0.96875,
      "learning_rate": 0.00017940974605353465,
      "loss": 0.0492,
      "step": 9300
    },
    {
      "epoch": 4.263565891472869,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00017929535575383207,
      "loss": 0.0526,
      "step": 9350
    },
    {
      "epoch": 4.286365709074327,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0001791809654541295,
      "loss": 0.0403,
      "step": 9400
    },
    {
      "epoch": 4.309165526675787,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001790665751544269,
      "loss": 0.0415,
      "step": 9450
    },
    {
      "epoch": 4.331965344277246,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00017895218485472435,
      "loss": 0.04,
      "step": 9500
    },
    {
      "epoch": 4.354765161878705,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00017883779455502173,
      "loss": 0.0461,
      "step": 9550
    },
    {
      "epoch": 4.377564979480164,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00017872340425531915,
      "loss": 0.0499,
      "step": 9600
    },
    {
      "epoch": 4.4003647970816235,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00017860901395561657,
      "loss": 0.038,
      "step": 9650
    },
    {
      "epoch": 4.423164614683083,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00017849462365591398,
      "loss": 0.0383,
      "step": 9700
    },
    {
      "epoch": 4.4459644322845415,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00017838023335621143,
      "loss": 0.0385,
      "step": 9750
    },
    {
      "epoch": 4.468764249886001,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00017826584305650881,
      "loss": 0.0361,
      "step": 9800
    },
    {
      "epoch": 4.49156406748746,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00017815145275680623,
      "loss": 0.0485,
      "step": 9850
    },
    {
      "epoch": 4.514363885088919,
      "grad_norm": 2.125,
      "learning_rate": 0.00017803706245710365,
      "loss": 0.0372,
      "step": 9900
    },
    {
      "epoch": 4.537163702690378,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00017792267215740106,
      "loss": 0.0433,
      "step": 9950
    },
    {
      "epoch": 4.559963520291838,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00017780828185769848,
      "loss": 0.0337,
      "step": 10000
    },
    {
      "epoch": 4.582763337893297,
      "grad_norm": 1.375,
      "learning_rate": 0.0001776938915579959,
      "loss": 0.0392,
      "step": 10050
    },
    {
      "epoch": 4.605563155494756,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001775795012582933,
      "loss": 0.0404,
      "step": 10100
    },
    {
      "epoch": 4.628362973096215,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00017746511095859073,
      "loss": 0.0312,
      "step": 10150
    },
    {
      "epoch": 4.651162790697675,
      "grad_norm": 0.609375,
      "learning_rate": 0.00017735072065888814,
      "loss": 0.0398,
      "step": 10200
    },
    {
      "epoch": 4.673962608299133,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00017723633035918553,
      "loss": 0.0386,
      "step": 10250
    },
    {
      "epoch": 4.696762425900593,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00017712194005948298,
      "loss": 0.0343,
      "step": 10300
    },
    {
      "epoch": 4.719562243502052,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001770075497597804,
      "loss": 0.0472,
      "step": 10350
    },
    {
      "epoch": 4.7423620611035116,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00017689315946007778,
      "loss": 0.042,
      "step": 10400
    },
    {
      "epoch": 4.76516187870497,
      "grad_norm": 1.78125,
      "learning_rate": 0.00017677876916037522,
      "loss": 0.0412,
      "step": 10450
    },
    {
      "epoch": 4.7879616963064295,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001766643788606726,
      "loss": 0.0432,
      "step": 10500
    },
    {
      "epoch": 4.810761513907889,
      "grad_norm": 0.2265625,
      "learning_rate": 0.00017654998856097006,
      "loss": 0.0525,
      "step": 10550
    },
    {
      "epoch": 4.8335613315093475,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00017643559826126744,
      "loss": 0.0395,
      "step": 10600
    },
    {
      "epoch": 4.856361149110807,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00017632120796156486,
      "loss": 0.0387,
      "step": 10650
    },
    {
      "epoch": 4.879160966712266,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00017620681766186228,
      "loss": 0.041,
      "step": 10700
    },
    {
      "epoch": 4.901960784313726,
      "grad_norm": 1.640625,
      "learning_rate": 0.0001760924273621597,
      "loss": 0.038,
      "step": 10750
    },
    {
      "epoch": 4.924760601915184,
      "grad_norm": 1.3125,
      "learning_rate": 0.0001759780370624571,
      "loss": 0.0448,
      "step": 10800
    },
    {
      "epoch": 4.947560419516644,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00017586364676275452,
      "loss": 0.0417,
      "step": 10850
    },
    {
      "epoch": 4.970360237118103,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00017574925646305194,
      "loss": 0.0393,
      "step": 10900
    },
    {
      "epoch": 4.993160054719562,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00017563486616334936,
      "loss": 0.0353,
      "step": 10950
    },
    {
      "epoch": 5.015959872321021,
      "grad_norm": 0.421875,
      "learning_rate": 0.00017552047586364677,
      "loss": 0.0376,
      "step": 11000
    },
    {
      "epoch": 5.038759689922481,
      "grad_norm": 1.828125,
      "learning_rate": 0.0001754060855639442,
      "loss": 0.0388,
      "step": 11050
    },
    {
      "epoch": 5.06155950752394,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0001752916952642416,
      "loss": 0.0318,
      "step": 11100
    },
    {
      "epoch": 5.084359325125399,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00017517730496453902,
      "loss": 0.0251,
      "step": 11150
    },
    {
      "epoch": 5.107159142726858,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0001750629146648364,
      "loss": 0.0443,
      "step": 11200
    },
    {
      "epoch": 5.1299589603283176,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00017494852436513385,
      "loss": 0.0285,
      "step": 11250
    },
    {
      "epoch": 5.152758777929777,
      "grad_norm": 0.75,
      "learning_rate": 0.00017483413406543124,
      "loss": 0.0319,
      "step": 11300
    },
    {
      "epoch": 5.1755585955312355,
      "grad_norm": 0.375,
      "learning_rate": 0.00017471974376572869,
      "loss": 0.0292,
      "step": 11350
    },
    {
      "epoch": 5.198358413132695,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0001746053534660261,
      "loss": 0.0281,
      "step": 11400
    },
    {
      "epoch": 5.221158230734154,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0001744909631663235,
      "loss": 0.029,
      "step": 11450
    },
    {
      "epoch": 5.243958048335613,
      "grad_norm": 0.265625,
      "learning_rate": 0.00017437657286662093,
      "loss": 0.0382,
      "step": 11500
    },
    {
      "epoch": 5.266757865937072,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00017426218256691832,
      "loss": 0.0328,
      "step": 11550
    },
    {
      "epoch": 5.289557683538532,
      "grad_norm": 1.6015625,
      "learning_rate": 0.00017414779226721577,
      "loss": 0.03,
      "step": 11600
    },
    {
      "epoch": 5.312357501139991,
      "grad_norm": 1.109375,
      "learning_rate": 0.00017403340196751315,
      "loss": 0.0251,
      "step": 11650
    },
    {
      "epoch": 5.33515731874145,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00017391901166781057,
      "loss": 0.0267,
      "step": 11700
    },
    {
      "epoch": 5.357957136342909,
      "grad_norm": 0.6171875,
      "learning_rate": 0.000173804621368108,
      "loss": 0.0357,
      "step": 11750
    },
    {
      "epoch": 5.380756953944369,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0001736902310684054,
      "loss": 0.035,
      "step": 11800
    },
    {
      "epoch": 5.403556771545827,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00017357584076870282,
      "loss": 0.0235,
      "step": 11850
    },
    {
      "epoch": 5.426356589147287,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00017346145046900023,
      "loss": 0.0277,
      "step": 11900
    },
    {
      "epoch": 5.449156406748746,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00017334706016929765,
      "loss": 0.0294,
      "step": 11950
    },
    {
      "epoch": 5.471956224350206,
      "grad_norm": 0.859375,
      "learning_rate": 0.00017323266986959507,
      "loss": 0.0245,
      "step": 12000
    }
  ],
  "logging_steps": 50,
  "max_steps": 87720,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 40,
  "save_steps": 200,
  "total_flos": 0.0,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": null
}