{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 200,
  "global_step": 668,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0029940119760479044,
      "grad_norm": 2.360150470792367,
      "learning_rate": 9.999944704978835e-06,
      "loss": 0.6092,
      "step": 1
    },
    {
      "epoch": 0.005988023952095809,
      "grad_norm": 2.4295311904157155,
      "learning_rate": 9.999778821138357e-06,
      "loss": 0.575,
      "step": 2
    },
    {
      "epoch": 0.008982035928143712,
      "grad_norm": 1.6045552308559259,
      "learning_rate": 9.999502352147583e-06,
      "loss": 0.5736,
      "step": 3
    },
    {
      "epoch": 0.011976047904191617,
      "grad_norm": 1.7003243086131508,
      "learning_rate": 9.999115304121459e-06,
      "loss": 0.6014,
      "step": 4
    },
    {
      "epoch": 0.014970059880239521,
      "grad_norm": 1.499989194091945,
      "learning_rate": 9.998617685620715e-06,
      "loss": 0.4662,
      "step": 5
    },
    {
      "epoch": 0.017964071856287425,
      "grad_norm": 1.61955016345233,
      "learning_rate": 9.998009507651683e-06,
      "loss": 0.5257,
      "step": 6
    },
    {
      "epoch": 0.020958083832335328,
      "grad_norm": 1.465934804743867,
      "learning_rate": 9.997290783666048e-06,
      "loss": 0.481,
      "step": 7
    },
    {
      "epoch": 0.023952095808383235,
      "grad_norm": 1.4260547177810452,
      "learning_rate": 9.996461529560553e-06,
      "loss": 0.4611,
      "step": 8
    },
    {
      "epoch": 0.02694610778443114,
      "grad_norm": 1.594334635936141,
      "learning_rate": 9.995521763676645e-06,
      "loss": 0.5282,
      "step": 9
    },
    {
      "epoch": 0.029940119760479042,
      "grad_norm": 1.5412598472431942,
      "learning_rate": 9.994471506800078e-06,
      "loss": 0.4832,
      "step": 10
    },
    {
      "epoch": 0.03293413173652695,
      "grad_norm": 1.3382556026543801,
      "learning_rate": 9.993310782160439e-06,
      "loss": 0.5119,
      "step": 11
    },
    {
      "epoch": 0.03592814371257485,
      "grad_norm": 1.3602716806719026,
      "learning_rate": 9.992039615430648e-06,
      "loss": 0.4866,
      "step": 12
    },
    {
      "epoch": 0.038922155688622756,
      "grad_norm": 1.3257085493157688,
      "learning_rate": 9.99065803472638e-06,
      "loss": 0.456,
      "step": 13
    },
    {
      "epoch": 0.041916167664670656,
      "grad_norm": 1.240022327929267,
      "learning_rate": 9.989166070605447e-06,
      "loss": 0.4411,
      "step": 14
    },
    {
      "epoch": 0.04491017964071856,
      "grad_norm": 1.1452990663610407,
      "learning_rate": 9.98756375606713e-06,
      "loss": 0.4269,
      "step": 15
    },
    {
      "epoch": 0.04790419161676647,
      "grad_norm": 1.3019072667574294,
      "learning_rate": 9.985851126551428e-06,
      "loss": 0.4653,
      "step": 16
    },
    {
      "epoch": 0.05089820359281437,
      "grad_norm": 0.9875023401666374,
      "learning_rate": 9.9840282199383e-06,
      "loss": 0.3829,
      "step": 17
    },
    {
      "epoch": 0.05389221556886228,
      "grad_norm": 1.0529387527844354,
      "learning_rate": 9.982095076546806e-06,
      "loss": 0.4399,
      "step": 18
    },
    {
      "epoch": 0.05688622754491018,
      "grad_norm": 1.179874914954893,
      "learning_rate": 9.980051739134235e-06,
      "loss": 0.4457,
      "step": 19
    },
    {
      "epoch": 0.059880239520958084,
      "grad_norm": 1.215394016628416,
      "learning_rate": 9.977898252895133e-06,
      "loss": 0.46,
      "step": 20
    },
    {
      "epoch": 0.06287425149700598,
      "grad_norm": 1.100471009595285,
      "learning_rate": 9.975634665460333e-06,
      "loss": 0.4327,
      "step": 21
    },
    {
      "epoch": 0.0658682634730539,
      "grad_norm": 1.1308041938018731,
      "learning_rate": 9.973261026895878e-06,
      "loss": 0.4032,
      "step": 22
    },
    {
      "epoch": 0.0688622754491018,
      "grad_norm": 1.0857215593281804,
      "learning_rate": 9.970777389701927e-06,
      "loss": 0.4898,
      "step": 23
    },
    {
      "epoch": 0.0718562874251497,
      "grad_norm": 1.0677097396510304,
      "learning_rate": 9.968183808811586e-06,
      "loss": 0.3944,
      "step": 24
    },
    {
      "epoch": 0.0748502994011976,
      "grad_norm": 1.2863204713458667,
      "learning_rate": 9.965480341589702e-06,
      "loss": 0.4591,
      "step": 25
    },
    {
      "epoch": 0.07784431137724551,
      "grad_norm": 1.099240850527482,
      "learning_rate": 9.962667047831585e-06,
      "loss": 0.4611,
      "step": 26
    },
    {
      "epoch": 0.08083832335329341,
      "grad_norm": 1.1178720677514475,
      "learning_rate": 9.95974398976169e-06,
      "loss": 0.4008,
      "step": 27
    },
    {
      "epoch": 0.08383233532934131,
      "grad_norm": 1.127386283354638,
      "learning_rate": 9.95671123203224e-06,
      "loss": 0.4507,
      "step": 28
    },
    {
      "epoch": 0.08682634730538923,
      "grad_norm": 1.2066334083956984,
      "learning_rate": 9.953568841721796e-06,
      "loss": 0.4805,
      "step": 29
    },
    {
      "epoch": 0.08982035928143713,
      "grad_norm": 0.9775903615301047,
      "learning_rate": 9.950316888333775e-06,
      "loss": 0.3911,
      "step": 30
    },
    {
      "epoch": 0.09281437125748503,
      "grad_norm": 1.0222727254646968,
      "learning_rate": 9.946955443794908e-06,
      "loss": 0.4688,
      "step": 31
    },
    {
      "epoch": 0.09580838323353294,
      "grad_norm": 1.2017413132687296,
      "learning_rate": 9.943484582453653e-06,
      "loss": 0.4383,
      "step": 32
    },
    {
      "epoch": 0.09880239520958084,
      "grad_norm": 0.9625560512846535,
      "learning_rate": 9.939904381078553e-06,
      "loss": 0.3857,
      "step": 33
    },
    {
      "epoch": 0.10179640718562874,
      "grad_norm": 0.9265832067065958,
      "learning_rate": 9.93621491885653e-06,
      "loss": 0.3652,
      "step": 34
    },
    {
      "epoch": 0.10479041916167664,
      "grad_norm": 1.09656335883221,
      "learning_rate": 9.932416277391144e-06,
      "loss": 0.4103,
      "step": 35
    },
    {
      "epoch": 0.10778443113772455,
      "grad_norm": 1.091981004351075,
      "learning_rate": 9.928508540700775e-06,
      "loss": 0.4308,
      "step": 36
    },
    {
      "epoch": 0.11077844311377245,
      "grad_norm": 1.1471983189479533,
      "learning_rate": 9.924491795216777e-06,
      "loss": 0.4722,
      "step": 37
    },
    {
      "epoch": 0.11377245508982035,
      "grad_norm": 1.1015423992585116,
      "learning_rate": 9.920366129781564e-06,
      "loss": 0.4317,
      "step": 38
    },
    {
      "epoch": 0.11676646706586827,
      "grad_norm": 0.9662315165782208,
      "learning_rate": 9.916131635646635e-06,
      "loss": 0.3857,
      "step": 39
    },
    {
      "epoch": 0.11976047904191617,
      "grad_norm": 1.1042042362974234,
      "learning_rate": 9.91178840647057e-06,
      "loss": 0.4587,
      "step": 40
    },
    {
      "epoch": 0.12275449101796407,
      "grad_norm": 1.0967203910091963,
      "learning_rate": 9.907336538316946e-06,
      "loss": 0.4116,
      "step": 41
    },
    {
      "epoch": 0.12574850299401197,
      "grad_norm": 0.9924909654132533,
      "learning_rate": 9.902776129652223e-06,
      "loss": 0.3572,
      "step": 42
    },
    {
      "epoch": 0.12874251497005987,
      "grad_norm": 1.0528124902395053,
      "learning_rate": 9.898107281343557e-06,
      "loss": 0.3685,
      "step": 43
    },
    {
      "epoch": 0.1317365269461078,
      "grad_norm": 0.9511215627795053,
      "learning_rate": 9.893330096656576e-06,
      "loss": 0.3436,
      "step": 44
    },
    {
      "epoch": 0.1347305389221557,
      "grad_norm": 1.0836924506952512,
      "learning_rate": 9.888444681253087e-06,
      "loss": 0.4279,
      "step": 45
    },
    {
      "epoch": 0.1377245508982036,
      "grad_norm": 1.1151679775891317,
      "learning_rate": 9.883451143188753e-06,
      "loss": 0.4221,
      "step": 46
    },
    {
      "epoch": 0.1407185628742515,
      "grad_norm": 1.064674167680838,
      "learning_rate": 9.878349592910694e-06,
      "loss": 0.3527,
      "step": 47
    },
    {
      "epoch": 0.1437125748502994,
      "grad_norm": 1.1036459061321013,
      "learning_rate": 9.873140143255035e-06,
      "loss": 0.4467,
      "step": 48
    },
    {
      "epoch": 0.1467065868263473,
      "grad_norm": 1.0400888612361692,
      "learning_rate": 9.867822909444435e-06,
      "loss": 0.4393,
      "step": 49
    },
    {
      "epoch": 0.1497005988023952,
      "grad_norm": 1.0251238188251106,
      "learning_rate": 9.862398009085511e-06,
      "loss": 0.4082,
      "step": 50
    },
    {
      "epoch": 0.15269461077844312,
      "grad_norm": 1.0567380954404586,
      "learning_rate": 9.856865562166256e-06,
      "loss": 0.3869,
      "step": 51
    },
    {
      "epoch": 0.15568862275449102,
      "grad_norm": 1.0102720103461709,
      "learning_rate": 9.851225691053382e-06,
      "loss": 0.3592,
      "step": 52
    },
    {
      "epoch": 0.15868263473053892,
      "grad_norm": 1.1127063801621502,
      "learning_rate": 9.8454785204896e-06,
      "loss": 0.4344,
      "step": 53
    },
    {
      "epoch": 0.16167664670658682,
      "grad_norm": 0.9065152442964788,
      "learning_rate": 9.83962417759088e-06,
      "loss": 0.3664,
      "step": 54
    },
    {
      "epoch": 0.16467065868263472,
      "grad_norm": 1.072288748663969,
      "learning_rate": 9.833662791843628e-06,
      "loss": 0.3651,
      "step": 55
    },
    {
      "epoch": 0.16766467065868262,
      "grad_norm": 1.0107526179429653,
      "learning_rate": 9.827594495101824e-06,
      "loss": 0.3898,
      "step": 56
    },
    {
      "epoch": 0.17065868263473055,
      "grad_norm": 1.0293793758687437,
      "learning_rate": 9.821419421584108e-06,
      "loss": 0.4018,
      "step": 57
    },
    {
      "epoch": 0.17365269461077845,
      "grad_norm": 1.040641029363235,
      "learning_rate": 9.815137707870806e-06,
      "loss": 0.4427,
      "step": 58
    },
    {
      "epoch": 0.17664670658682635,
      "grad_norm": 1.1346149341238732,
      "learning_rate": 9.808749492900917e-06,
      "loss": 0.3956,
      "step": 59
    },
    {
      "epoch": 0.17964071856287425,
      "grad_norm": 1.002697448139183,
      "learning_rate": 9.802254917969033e-06,
      "loss": 0.4166,
      "step": 60
    },
    {
      "epoch": 0.18263473053892215,
      "grad_norm": 1.166908537675008,
      "learning_rate": 9.795654126722218e-06,
      "loss": 0.3991,
      "step": 61
    },
    {
      "epoch": 0.18562874251497005,
      "grad_norm": 0.9918959030412066,
      "learning_rate": 9.788947265156828e-06,
      "loss": 0.3852,
      "step": 62
    },
    {
      "epoch": 0.18862275449101795,
      "grad_norm": 1.0324508520275155,
      "learning_rate": 9.782134481615282e-06,
      "loss": 0.3498,
      "step": 63
    },
    {
      "epoch": 0.19161676646706588,
      "grad_norm": 1.0257932465807185,
      "learning_rate": 9.775215926782788e-06,
      "loss": 0.4258,
      "step": 64
    },
    {
      "epoch": 0.19461077844311378,
      "grad_norm": 1.0782444651324654,
      "learning_rate": 9.768191753683997e-06,
      "loss": 0.3829,
      "step": 65
    },
    {
      "epoch": 0.19760479041916168,
      "grad_norm": 1.0912461346785043,
      "learning_rate": 9.761062117679632e-06,
      "loss": 0.3816,
      "step": 66
    },
    {
      "epoch": 0.20059880239520958,
      "grad_norm": 1.0614318310008004,
      "learning_rate": 9.75382717646304e-06,
      "loss": 0.3616,
      "step": 67
    },
    {
      "epoch": 0.20359281437125748,
      "grad_norm": 1.1501837516722428,
      "learning_rate": 9.746487090056712e-06,
      "loss": 0.4522,
      "step": 68
    },
    {
      "epoch": 0.20658682634730538,
      "grad_norm": 0.9006942564473596,
      "learning_rate": 9.739042020808746e-06,
      "loss": 0.3474,
      "step": 69
    },
    {
      "epoch": 0.20958083832335328,
      "grad_norm": 1.019042335184148,
      "learning_rate": 9.73149213338924e-06,
      "loss": 0.4297,
      "step": 70
    },
    {
      "epoch": 0.2125748502994012,
      "grad_norm": 1.0261132022875048,
      "learning_rate": 9.72383759478667e-06,
      "loss": 0.433,
      "step": 71
    },
    {
      "epoch": 0.2155688622754491,
      "grad_norm": 1.0279884605503198,
      "learning_rate": 9.71607857430419e-06,
      "loss": 0.3331,
      "step": 72
    },
    {
      "epoch": 0.218562874251497,
      "grad_norm": 1.10261368241984,
      "learning_rate": 9.708215243555875e-06,
      "loss": 0.4255,
      "step": 73
    },
    {
      "epoch": 0.2215568862275449,
      "grad_norm": 1.1938856577221162,
      "learning_rate": 9.700247776462944e-06,
      "loss": 0.3986,
      "step": 74
    },
    {
      "epoch": 0.2245508982035928,
      "grad_norm": 1.1282208342635738,
      "learning_rate": 9.6921763492499e-06,
      "loss": 0.391,
      "step": 75
    },
    {
      "epoch": 0.2275449101796407,
      "grad_norm": 0.9900141058207153,
      "learning_rate": 9.68400114044064e-06,
      "loss": 0.3177,
      "step": 76
    },
    {
      "epoch": 0.23053892215568864,
      "grad_norm": 1.0789984345178527,
      "learning_rate": 9.6757223308545e-06,
      "loss": 0.4004,
      "step": 77
    },
    {
      "epoch": 0.23353293413173654,
      "grad_norm": 0.9894907044520515,
      "learning_rate": 9.667340103602263e-06,
      "loss": 0.4484,
      "step": 78
    },
    {
      "epoch": 0.23652694610778444,
      "grad_norm": 0.9965832888156311,
      "learning_rate": 9.658854644082099e-06,
      "loss": 0.3594,
      "step": 79
    },
    {
      "epoch": 0.23952095808383234,
      "grad_norm": 1.2512456460659447,
      "learning_rate": 9.650266139975474e-06,
      "loss": 0.4393,
      "step": 80
    },
    {
      "epoch": 0.24251497005988024,
      "grad_norm": 1.1185407406813124,
      "learning_rate": 9.641574781242999e-06,
      "loss": 0.384,
      "step": 81
    },
    {
      "epoch": 0.24550898203592814,
      "grad_norm": 1.079792166966554,
      "learning_rate": 9.632780760120217e-06,
      "loss": 0.3987,
      "step": 82
    },
    {
      "epoch": 0.24850299401197604,
      "grad_norm": 1.0335442030139939,
      "learning_rate": 9.62388427111336e-06,
      "loss": 0.3622,
      "step": 83
    },
    {
      "epoch": 0.25149700598802394,
      "grad_norm": 0.9758738026602853,
      "learning_rate": 9.614885510995047e-06,
      "loss": 0.3328,
      "step": 84
    },
    {
      "epoch": 0.25449101796407186,
      "grad_norm": 1.0689976945861128,
      "learning_rate": 9.605784678799934e-06,
      "loss": 0.3807,
      "step": 85
    },
    {
      "epoch": 0.25748502994011974,
      "grad_norm": 1.03144717267387,
      "learning_rate": 9.596581975820304e-06,
      "loss": 0.3869,
      "step": 86
    },
    {
      "epoch": 0.26047904191616766,
      "grad_norm": 1.2177802527359998,
      "learning_rate": 9.587277605601617e-06,
      "loss": 0.4208,
      "step": 87
    },
    {
      "epoch": 0.2634730538922156,
      "grad_norm": 1.1785454020154376,
      "learning_rate": 9.577871773938013e-06,
      "loss": 0.4236,
      "step": 88
    },
    {
      "epoch": 0.26646706586826346,
      "grad_norm": 1.1125909821479287,
      "learning_rate": 9.568364688867757e-06,
      "loss": 0.4873,
      "step": 89
    },
    {
      "epoch": 0.2694610778443114,
      "grad_norm": 1.041716945636208,
      "learning_rate": 9.558756560668637e-06,
      "loss": 0.3689,
      "step": 90
    },
    {
      "epoch": 0.27245508982035926,
      "grad_norm": 1.1972249911203883,
      "learning_rate": 9.549047601853313e-06,
      "loss": 0.3904,
      "step": 91
    },
    {
      "epoch": 0.2754491017964072,
      "grad_norm": 1.2145517606375182,
      "learning_rate": 9.539238027164618e-06,
      "loss": 0.4117,
      "step": 92
    },
    {
      "epoch": 0.27844311377245506,
      "grad_norm": 1.0847366638843148,
      "learning_rate": 9.52932805357081e-06,
      "loss": 0.3886,
      "step": 93
    },
    {
      "epoch": 0.281437125748503,
      "grad_norm": 1.107223706682208,
      "learning_rate": 9.519317900260769e-06,
      "loss": 0.4234,
      "step": 94
    },
    {
      "epoch": 0.2844311377245509,
      "grad_norm": 1.0512423029389995,
      "learning_rate": 9.509207788639148e-06,
      "loss": 0.4119,
      "step": 95
    },
    {
      "epoch": 0.2874251497005988,
      "grad_norm": 1.0125109372778616,
      "learning_rate": 9.498997942321484e-06,
      "loss": 0.4155,
      "step": 96
    },
    {
      "epoch": 0.2904191616766467,
      "grad_norm": 0.9459988882190549,
      "learning_rate": 9.488688587129243e-06,
      "loss": 0.4266,
      "step": 97
    },
    {
      "epoch": 0.2934131736526946,
      "grad_norm": 1.159832613070711,
      "learning_rate": 9.47827995108483e-06,
      "loss": 0.3686,
      "step": 98
    },
    {
      "epoch": 0.2964071856287425,
      "grad_norm": 1.2126346435851747,
      "learning_rate": 9.467772264406545e-06,
      "loss": 0.3967,
      "step": 99
    },
    {
      "epoch": 0.2994011976047904,
      "grad_norm": 0.9662788722957085,
      "learning_rate": 9.457165759503492e-06,
      "loss": 0.3529,
      "step": 100
    },
    {
      "epoch": 0.3023952095808383,
      "grad_norm": 1.0465067846012959,
      "learning_rate": 9.446460670970436e-06,
      "loss": 0.3625,
      "step": 101
    },
    {
      "epoch": 0.30538922155688625,
      "grad_norm": 1.0771426905673824,
      "learning_rate": 9.435657235582616e-06,
      "loss": 0.321,
      "step": 102
    },
    {
      "epoch": 0.3083832335329341,
      "grad_norm": 1.170291037733682,
      "learning_rate": 9.424755692290507e-06,
      "loss": 0.396,
      "step": 103
    },
    {
      "epoch": 0.31137724550898205,
      "grad_norm": 1.1608390028269628,
      "learning_rate": 9.413756282214538e-06,
      "loss": 0.4272,
      "step": 104
    },
    {
      "epoch": 0.3143712574850299,
      "grad_norm": 0.9854661428873333,
      "learning_rate": 9.402659248639749e-06,
      "loss": 0.3792,
      "step": 105
    },
    {
      "epoch": 0.31736526946107785,
      "grad_norm": 1.0205302189095444,
      "learning_rate": 9.391464837010428e-06,
      "loss": 0.376,
      "step": 106
    },
    {
      "epoch": 0.3203592814371258,
      "grad_norm": 1.1245385463246433,
      "learning_rate": 9.380173294924661e-06,
      "loss": 0.4283,
      "step": 107
    },
    {
      "epoch": 0.32335329341317365,
      "grad_norm": 1.063144047888206,
      "learning_rate": 9.368784872128877e-06,
      "loss": 0.3879,
      "step": 108
    },
    {
      "epoch": 0.3263473053892216,
      "grad_norm": 0.9722099661439674,
      "learning_rate": 9.357299820512305e-06,
      "loss": 0.3712,
      "step": 109
    },
    {
      "epoch": 0.32934131736526945,
      "grad_norm": 1.0475391247605021,
      "learning_rate": 9.345718394101412e-06,
      "loss": 0.3707,
      "step": 110
    },
    {
      "epoch": 0.3323353293413174,
      "grad_norm": 1.058149177559626,
      "learning_rate": 9.334040849054288e-06,
      "loss": 0.4115,
      "step": 111
    },
    {
      "epoch": 0.33532934131736525,
      "grad_norm": 1.088390795361788,
      "learning_rate": 9.322267443654974e-06,
      "loss": 0.3874,
      "step": 112
    },
    {
      "epoch": 0.3383233532934132,
      "grad_norm": 1.1013443338361675,
      "learning_rate": 9.310398438307747e-06,
      "loss": 0.4323,
      "step": 113
    },
    {
      "epoch": 0.3413173652694611,
      "grad_norm": 0.9944967696397207,
      "learning_rate": 9.29843409553137e-06,
      "loss": 0.3571,
      "step": 114
    },
    {
      "epoch": 0.344311377245509,
      "grad_norm": 0.9723803949692649,
      "learning_rate": 9.286374679953278e-06,
      "loss": 0.3619,
      "step": 115
    },
    {
      "epoch": 0.3473053892215569,
      "grad_norm": 1.0263431397562555,
      "learning_rate": 9.274220458303727e-06,
      "loss": 0.3407,
      "step": 116
    },
    {
      "epoch": 0.3502994011976048,
      "grad_norm": 1.0463627357456768,
      "learning_rate": 9.261971699409893e-06,
      "loss": 0.4016,
      "step": 117
    },
    {
      "epoch": 0.3532934131736527,
      "grad_norm": 1.07587077688648,
      "learning_rate": 9.249628674189928e-06,
      "loss": 0.3761,
      "step": 118
    },
    {
      "epoch": 0.3562874251497006,
      "grad_norm": 1.0847776478670657,
      "learning_rate": 9.237191655646972e-06,
      "loss": 0.4102,
      "step": 119
    },
    {
      "epoch": 0.3592814371257485,
      "grad_norm": 1.11812369249786,
      "learning_rate": 9.224660918863104e-06,
      "loss": 0.4422,
      "step": 120
    },
    {
      "epoch": 0.36227544910179643,
      "grad_norm": 1.135985769802511,
      "learning_rate": 9.212036740993265e-06,
      "loss": 0.391,
      "step": 121
    },
    {
      "epoch": 0.3652694610778443,
      "grad_norm": 1.080200371957256,
      "learning_rate": 9.199319401259132e-06,
      "loss": 0.3878,
      "step": 122
    },
    {
      "epoch": 0.36826347305389223,
      "grad_norm": 1.099637838806261,
      "learning_rate": 9.186509180942928e-06,
      "loss": 0.3652,
      "step": 123
    },
    {
      "epoch": 0.3712574850299401,
      "grad_norm": 1.0554720701766402,
      "learning_rate": 9.173606363381218e-06,
      "loss": 0.4284,
      "step": 124
    },
    {
      "epoch": 0.37425149700598803,
      "grad_norm": 1.0455932007891395,
      "learning_rate": 9.16061123395863e-06,
      "loss": 0.3833,
      "step": 125
    },
    {
      "epoch": 0.3772455089820359,
      "grad_norm": 1.1376490909593353,
      "learning_rate": 9.147524080101543e-06,
      "loss": 0.3876,
      "step": 126
    },
    {
      "epoch": 0.38023952095808383,
      "grad_norm": 1.188836716557053,
      "learning_rate": 9.134345191271742e-06,
      "loss": 0.4461,
      "step": 127
    },
    {
      "epoch": 0.38323353293413176,
      "grad_norm": 1.1491556565938952,
      "learning_rate": 9.121074858959997e-06,
      "loss": 0.4208,
      "step": 128
    },
    {
      "epoch": 0.38622754491017963,
      "grad_norm": 0.9829441169608918,
      "learning_rate": 9.107713376679634e-06,
      "loss": 0.3933,
      "step": 129
    },
    {
      "epoch": 0.38922155688622756,
      "grad_norm": 1.1745728924206915,
      "learning_rate": 9.094261039960028e-06,
      "loss": 0.4012,
      "step": 130
    },
    {
      "epoch": 0.39221556886227543,
      "grad_norm": 1.18181153163394,
      "learning_rate": 9.08071814634008e-06,
      "loss": 0.4552,
      "step": 131
    },
    {
      "epoch": 0.39520958083832336,
      "grad_norm": 1.1433168380702565,
      "learning_rate": 9.067084995361623e-06,
      "loss": 0.395,
      "step": 132
    },
    {
      "epoch": 0.39820359281437123,
      "grad_norm": 1.166913861368381,
      "learning_rate": 9.053361888562807e-06,
      "loss": 0.4411,
      "step": 133
    },
    {
      "epoch": 0.40119760479041916,
      "grad_norm": 1.1022003304573305,
      "learning_rate": 9.039549129471423e-06,
      "loss": 0.3753,
      "step": 134
    },
    {
      "epoch": 0.4041916167664671,
      "grad_norm": 0.9195981793191216,
      "learning_rate": 9.025647023598196e-06,
      "loss": 0.353,
      "step": 135
    },
    {
      "epoch": 0.40718562874251496,
      "grad_norm": 1.039297953987639,
      "learning_rate": 9.011655878430018e-06,
      "loss": 0.3611,
      "step": 136
    },
    {
      "epoch": 0.4101796407185629,
      "grad_norm": 1.161126504900604,
      "learning_rate": 8.99757600342316e-06,
      "loss": 0.423,
      "step": 137
    },
    {
      "epoch": 0.41317365269461076,
      "grad_norm": 1.1493232149734816,
      "learning_rate": 8.983407709996415e-06,
      "loss": 0.4031,
      "step": 138
    },
    {
      "epoch": 0.4161676646706587,
      "grad_norm": 1.1232381761248593,
      "learning_rate": 8.969151311524215e-06,
      "loss": 0.3778,
      "step": 139
    },
    {
      "epoch": 0.41916167664670656,
      "grad_norm": 1.1797682325377319,
      "learning_rate": 8.954807123329703e-06,
      "loss": 0.4066,
      "step": 140
    },
    {
      "epoch": 0.4221556886227545,
      "grad_norm": 0.9898700775146696,
      "learning_rate": 8.940375462677758e-06,
      "loss": 0.3396,
      "step": 141
    },
    {
      "epoch": 0.4251497005988024,
      "grad_norm": 1.0989992916462306,
      "learning_rate": 8.92585664876797e-06,
      "loss": 0.3787,
      "step": 142
    },
    {
      "epoch": 0.4281437125748503,
      "grad_norm": 0.9751743477236701,
      "learning_rate": 8.911251002727588e-06,
      "loss": 0.3697,
      "step": 143
    },
    {
      "epoch": 0.4311377245508982,
      "grad_norm": 1.0020331833699752,
      "learning_rate": 8.896558847604414e-06,
      "loss": 0.3543,
      "step": 144
    },
    {
      "epoch": 0.4341317365269461,
      "grad_norm": 1.0479535082629463,
      "learning_rate": 8.881780508359661e-06,
      "loss": 0.4605,
      "step": 145
    },
    {
      "epoch": 0.437125748502994,
      "grad_norm": 1.321950771951831,
      "learning_rate": 8.86691631186076e-06,
      "loss": 0.392,
      "step": 146
    },
    {
      "epoch": 0.44011976047904194,
      "grad_norm": 0.8988900523275937,
      "learning_rate": 8.851966586874138e-06,
      "loss": 0.3388,
      "step": 147
    },
    {
      "epoch": 0.4431137724550898,
      "grad_norm": 1.0463909572882797,
      "learning_rate": 8.836931664057935e-06,
      "loss": 0.3617,
      "step": 148
    },
    {
      "epoch": 0.44610778443113774,
      "grad_norm": 1.1289518767389093,
      "learning_rate": 8.821811875954705e-06,
      "loss": 0.388,
      "step": 149
    },
    {
      "epoch": 0.4491017964071856,
      "grad_norm": 1.0553303120139725,
      "learning_rate": 8.806607556984045e-06,
      "loss": 0.3405,
      "step": 150
    },
    {
      "epoch": 0.45209580838323354,
      "grad_norm": 1.0337259853792555,
      "learning_rate": 8.791319043435213e-06,
      "loss": 0.3749,
      "step": 151
    },
    {
      "epoch": 0.4550898203592814,
      "grad_norm": 0.9697649279113124,
      "learning_rate": 8.775946673459682e-06,
      "loss": 0.3803,
      "step": 152
    },
    {
      "epoch": 0.45808383233532934,
      "grad_norm": 1.0554720933913004,
      "learning_rate": 8.76049078706366e-06,
      "loss": 0.3547,
      "step": 153
    },
    {
      "epoch": 0.46107784431137727,
      "grad_norm": 1.1105934710364926,
      "learning_rate": 8.744951726100572e-06,
      "loss": 0.3338,
      "step": 154
    },
    {
      "epoch": 0.46407185628742514,
      "grad_norm": 1.2520450241935366,
      "learning_rate": 8.729329834263503e-06,
      "loss": 0.3677,
      "step": 155
    },
    {
      "epoch": 0.46706586826347307,
      "grad_norm": 1.1146784316043523,
      "learning_rate": 8.713625457077585e-06,
      "loss": 0.3994,
      "step": 156
    },
    {
      "epoch": 0.47005988023952094,
      "grad_norm": 1.0874986898542098,
      "learning_rate": 8.697838941892371e-06,
      "loss": 0.354,
      "step": 157
    },
    {
      "epoch": 0.47305389221556887,
      "grad_norm": 1.0973091543852385,
      "learning_rate": 8.681970637874131e-06,
      "loss": 0.4154,
      "step": 158
    },
    {
      "epoch": 0.47604790419161674,
      "grad_norm": 1.1625464475176286,
      "learning_rate": 8.666020895998154e-06,
      "loss": 0.3891,
      "step": 159
    },
    {
      "epoch": 0.47904191616766467,
      "grad_norm": 1.161010302826186,
      "learning_rate": 8.64999006904096e-06,
      "loss": 0.3961,
      "step": 160
    },
    {
      "epoch": 0.4820359281437126,
      "grad_norm": 1.1337346146886311,
      "learning_rate": 8.63387851157252e-06,
      "loss": 0.3942,
      "step": 161
    },
    {
      "epoch": 0.48502994011976047,
      "grad_norm": 1.0391529706550826,
      "learning_rate": 8.617686579948396e-06,
      "loss": 0.3394,
      "step": 162
    },
    {
      "epoch": 0.4880239520958084,
      "grad_norm": 1.1236983918674,
      "learning_rate": 8.60141463230187e-06,
      "loss": 0.3576,
      "step": 163
    },
    {
      "epoch": 0.49101796407185627,
      "grad_norm": 1.1148066475922545,
      "learning_rate": 8.585063028536015e-06,
      "loss": 0.4442,
      "step": 164
    },
    {
      "epoch": 0.4940119760479042,
      "grad_norm": 1.0243204434346274,
      "learning_rate": 8.568632130315747e-06,
      "loss": 0.3854,
      "step": 165
    },
    {
      "epoch": 0.49700598802395207,
      "grad_norm": 1.059220388171805,
      "learning_rate": 8.552122301059807e-06,
      "loss": 0.3738,
      "step": 166
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0387730035667786,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.36,
      "step": 167
    },
    {
      "epoch": 0.5029940119760479,
      "grad_norm": 1.0439677605881634,
      "learning_rate": 8.518867311836808e-06,
      "loss": 0.4093,
      "step": 168
    },
    {
      "epoch": 0.5059880239520959,
      "grad_norm": 1.1135308466754046,
      "learning_rate": 8.502122887403882e-06,
      "loss": 0.4136,
      "step": 169
    },
    {
      "epoch": 0.5089820359281437,
      "grad_norm": 1.2076685795281583,
      "learning_rate": 8.485301002987285e-06,
      "loss": 0.3306,
      "step": 170
    },
    {
      "epoch": 0.5119760479041916,
      "grad_norm": 1.0431456753366977,
      "learning_rate": 8.468402030653598e-06,
      "loss": 0.4271,
      "step": 171
    },
    {
      "epoch": 0.5149700598802395,
      "grad_norm": 1.004312710195015,
      "learning_rate": 8.451426344174433e-06,
      "loss": 0.3612,
      "step": 172
    },
    {
      "epoch": 0.5179640718562875,
      "grad_norm": 1.1763053752804515,
      "learning_rate": 8.434374319018165e-06,
      "loss": 0.4386,
      "step": 173
    },
    {
      "epoch": 0.5209580838323353,
      "grad_norm": 1.002861520774924,
      "learning_rate": 8.417246332341638e-06,
      "loss": 0.3607,
      "step": 174
    },
    {
      "epoch": 0.5239520958083832,
      "grad_norm": 1.0163281602743386,
      "learning_rate": 8.4000427629818e-06,
      "loss": 0.3742,
      "step": 175
    },
    {
      "epoch": 0.5269461077844312,
      "grad_norm": 1.1136294757855683,
      "learning_rate": 8.382763991447344e-06,
      "loss": 0.41,
      "step": 176
    },
    {
      "epoch": 0.5299401197604791,
      "grad_norm": 1.1390366644182248,
      "learning_rate": 8.365410399910287e-06,
      "loss": 0.3872,
      "step": 177
    },
    {
      "epoch": 0.5329341317365269,
      "grad_norm": 1.2909433668063384,
      "learning_rate": 8.347982372197515e-06,
      "loss": 0.3782,
      "step": 178
    },
    {
      "epoch": 0.5359281437125748,
      "grad_norm": 1.0094599265789481,
      "learning_rate": 8.33048029378229e-06,
      "loss": 0.3918,
      "step": 179
    },
    {
      "epoch": 0.5389221556886228,
      "grad_norm": 0.9603157918644131,
      "learning_rate": 8.312904551775731e-06,
      "loss": 0.3454,
      "step": 180
    },
    {
      "epoch": 0.5419161676646707,
      "grad_norm": 0.977672929873693,
      "learning_rate": 8.295255534918249e-06,
      "loss": 0.3825,
      "step": 181
    },
    {
      "epoch": 0.5449101796407185,
      "grad_norm": 1.213979340299757,
      "learning_rate": 8.277533633570948e-06,
      "loss": 0.4422,
      "step": 182
    },
    {
      "epoch": 0.5479041916167665,
      "grad_norm": 1.0327111193464191,
      "learning_rate": 8.25973923970699e-06,
      "loss": 0.4067,
      "step": 183
    },
    {
      "epoch": 0.5508982035928144,
      "grad_norm": 0.9574992889413487,
      "learning_rate": 8.241872746902934e-06,
      "loss": 0.4474,
      "step": 184
    },
    {
      "epoch": 0.5538922155688623,
      "grad_norm": 1.0271449481355721,
      "learning_rate": 8.223934550330015e-06,
      "loss": 0.3573,
      "step": 185
    },
    {
      "epoch": 0.5568862275449101,
      "grad_norm": 1.085871604199334,
      "learning_rate": 8.20592504674542e-06,
      "loss": 0.4165,
      "step": 186
    },
    {
      "epoch": 0.5598802395209581,
      "grad_norm": 0.9177887450776159,
      "learning_rate": 8.187844634483495e-06,
      "loss": 0.4189,
      "step": 187
    },
    {
      "epoch": 0.562874251497006,
      "grad_norm": 1.145875691459872,
      "learning_rate": 8.16969371344696e-06,
      "loss": 0.4779,
      "step": 188
    },
    {
      "epoch": 0.5658682634730539,
      "grad_norm": 0.9058380786968147,
      "learning_rate": 8.151472685098037e-06,
      "loss": 0.387,
      "step": 189
    },
    {
      "epoch": 0.5688622754491018,
      "grad_norm": 0.9918785343604253,
      "learning_rate": 8.13318195244958e-06,
      "loss": 0.3642,
      "step": 190
    },
    {
      "epoch": 0.5718562874251497,
      "grad_norm": 1.0094499454429962,
      "learning_rate": 8.114821920056177e-06,
      "loss": 0.387,
      "step": 191
    },
    {
      "epoch": 0.5748502994011976,
      "grad_norm": 1.0117656185636714,
      "learning_rate": 8.096392994005177e-06,
      "loss": 0.4095,
      "step": 192
    },
    {
      "epoch": 0.5778443113772455,
      "grad_norm": 0.8813749104246654,
      "learning_rate": 8.077895581907719e-06,
      "loss": 0.3745,
      "step": 193
    },
    {
      "epoch": 0.5808383233532934,
      "grad_norm": 0.8787597823406559,
      "learning_rate": 8.059330092889724e-06,
      "loss": 0.3363,
      "step": 194
    },
    {
      "epoch": 0.5838323353293413,
      "grad_norm": 0.9920644099714983,
      "learning_rate": 8.040696937582833e-06,
      "loss": 0.3818,
      "step": 195
    },
    {
      "epoch": 0.5868263473053892,
      "grad_norm": 1.0660433857696556,
      "learning_rate": 8.021996528115335e-06,
      "loss": 0.3495,
      "step": 196
    },
    {
      "epoch": 0.5898203592814372,
      "grad_norm": 1.0694059285104096,
      "learning_rate": 8.003229278103044e-06,
      "loss": 0.3617,
      "step": 197
    },
    {
      "epoch": 0.592814371257485,
      "grad_norm": 1.06330338919478,
      "learning_rate": 7.984395602640153e-06,
      "loss": 0.3655,
      "step": 198
    },
    {
      "epoch": 0.5958083832335329,
      "grad_norm": 1.1525243383043442,
      "learning_rate": 7.96549591829006e-06,
      "loss": 0.379,
      "step": 199
    },
    {
      "epoch": 0.5988023952095808,
      "grad_norm": 0.9230482284229541,
      "learning_rate": 7.946530643076138e-06,
      "loss": 0.3274,
      "step": 200
    },
    {
      "epoch": 0.5988023952095808,
      "eval_loss": 0.37389740347862244,
      "eval_runtime": 2.0945,
      "eval_samples_per_second": 25.782,
      "eval_steps_per_second": 6.684,
      "step": 200
    },
    {
      "epoch": 0.6017964071856288,
      "grad_norm": 1.0019766539666788,
      "learning_rate": 7.927500196472506e-06,
      "loss": 0.3823,
      "step": 201
    },
    {
      "epoch": 0.6047904191616766,
      "grad_norm": 1.1123295404323308,
      "learning_rate": 7.908404999394747e-06,
      "loss": 0.3286,
      "step": 202
    },
    {
      "epoch": 0.6077844311377245,
      "grad_norm": 1.0548653358670876,
      "learning_rate": 7.889245474190588e-06,
      "loss": 0.3759,
      "step": 203
    },
    {
      "epoch": 0.6107784431137725,
      "grad_norm": 1.2433092859069619,
      "learning_rate": 7.870022044630569e-06,
      "loss": 0.4113,
      "step": 204
    },
    {
      "epoch": 0.6137724550898204,
      "grad_norm": 0.990791990832856,
      "learning_rate": 7.85073513589867e-06,
      "loss": 0.3817,
      "step": 205
    },
    {
      "epoch": 0.6167664670658682,
      "grad_norm": 1.2594009600174254,
      "learning_rate": 7.831385174582901e-06,
      "loss": 0.3861,
      "step": 206
    },
    {
      "epoch": 0.6197604790419161,
      "grad_norm": 1.0197727813815136,
      "learning_rate": 7.81197258866587e-06,
      "loss": 0.3858,
      "step": 207
    },
    {
      "epoch": 0.6227544910179641,
      "grad_norm": 1.029024178798685,
      "learning_rate": 7.792497807515317e-06,
      "loss": 0.3405,
      "step": 208
    },
    {
      "epoch": 0.625748502994012,
      "grad_norm": 1.0843471502861493,
      "learning_rate": 7.772961261874615e-06,
      "loss": 0.3311,
      "step": 209
    },
    {
      "epoch": 0.6287425149700598,
      "grad_norm": 1.0932823908799638,
      "learning_rate": 7.75336338385325e-06,
      "loss": 0.3674,
      "step": 210
    },
    {
      "epoch": 0.6317365269461078,
      "grad_norm": 1.0982688835218217,
      "learning_rate": 7.733704606917248e-06,
      "loss": 0.399,
      "step": 211
    },
    {
      "epoch": 0.6347305389221557,
      "grad_norm": 1.0360696347675225,
      "learning_rate": 7.713985365879607e-06,
      "loss": 0.343,
      "step": 212
    },
    {
      "epoch": 0.6377245508982036,
      "grad_norm": 1.0069925186594795,
      "learning_rate": 7.694206096890667e-06,
      "loss": 0.3541,
      "step": 213
    },
    {
      "epoch": 0.6407185628742516,
      "grad_norm": 0.9610077306255195,
      "learning_rate": 7.674367237428467e-06,
      "loss": 0.3386,
      "step": 214
    },
    {
      "epoch": 0.6437125748502994,
      "grad_norm": 0.9560418587007402,
      "learning_rate": 7.654469226289068e-06,
      "loss": 0.3263,
      "step": 215
    },
    {
      "epoch": 0.6467065868263473,
      "grad_norm": 1.0272151882071272,
      "learning_rate": 7.63451250357685e-06,
      "loss": 0.3906,
      "step": 216
    },
    {
      "epoch": 0.6497005988023952,
      "grad_norm": 1.0602665340805821,
      "learning_rate": 7.614497510694774e-06,
      "loss": 0.3981,
      "step": 217
    },
    {
      "epoch": 0.6526946107784432,
      "grad_norm": 1.091109550941202,
      "learning_rate": 7.5944246903346204e-06,
      "loss": 0.3585,
      "step": 218
    },
    {
      "epoch": 0.655688622754491,
      "grad_norm": 0.973104649648916,
      "learning_rate": 7.574294486467204e-06,
      "loss": 0.3523,
      "step": 219
    },
    {
      "epoch": 0.6586826347305389,
      "grad_norm": 1.0388023904136494,
      "learning_rate": 7.55410734433254e-06,
      "loss": 0.3984,
      "step": 220
    },
    {
      "epoch": 0.6616766467065869,
      "grad_norm": 1.1132398868564881,
      "learning_rate": 7.533863710430011e-06,
      "loss": 0.4175,
      "step": 221
    },
    {
      "epoch": 0.6646706586826348,
      "grad_norm": 1.0247318498739135,
      "learning_rate": 7.513564032508484e-06,
      "loss": 0.349,
      "step": 222
    },
    {
      "epoch": 0.6676646706586826,
      "grad_norm": 1.162703731242435,
      "learning_rate": 7.493208759556406e-06,
      "loss": 0.4039,
      "step": 223
    },
    {
      "epoch": 0.6706586826347305,
      "grad_norm": 1.151319734294355,
      "learning_rate": 7.472798341791877e-06,
      "loss": 0.4364,
      "step": 224
    },
    {
      "epoch": 0.6736526946107785,
      "grad_norm": 1.018414882097199,
      "learning_rate": 7.452333230652688e-06,
      "loss": 0.3628,
      "step": 225
    },
    {
      "epoch": 0.6766467065868264,
      "grad_norm": 0.9734290818533377,
      "learning_rate": 7.431813878786343e-06,
      "loss": 0.3525,
      "step": 226
    },
    {
      "epoch": 0.6796407185628742,
      "grad_norm": 1.0981823419124637,
      "learning_rate": 7.4112407400400395e-06,
      "loss": 0.3984,
      "step": 227
    },
    {
      "epoch": 0.6826347305389222,
      "grad_norm": 1.0082580347081591,
      "learning_rate": 7.390614269450633e-06,
      "loss": 0.3661,
      "step": 228
    },
    {
      "epoch": 0.6856287425149701,
      "grad_norm": 0.9094922182052086,
      "learning_rate": 7.369934923234577e-06,
      "loss": 0.3461,
      "step": 229
    },
    {
      "epoch": 0.688622754491018,
      "grad_norm": 1.1091280652518132,
      "learning_rate": 7.349203158777826e-06,
      "loss": 0.4246,
      "step": 230
    },
    {
      "epoch": 0.6916167664670658,
      "grad_norm": 1.1048056691020671,
      "learning_rate": 7.32841943462572e-06,
      "loss": 0.3855,
      "step": 231
    },
    {
      "epoch": 0.6946107784431138,
      "grad_norm": 1.0980287910054944,
      "learning_rate": 7.3075842104728445e-06,
      "loss": 0.3826,
      "step": 232
    },
    {
      "epoch": 0.6976047904191617,
      "grad_norm": 1.0677026866192707,
      "learning_rate": 7.286697947152868e-06,
      "loss": 0.3972,
      "step": 233
    },
    {
      "epoch": 0.7005988023952096,
      "grad_norm": 1.0422936388009736,
      "learning_rate": 7.265761106628338e-06,
      "loss": 0.3628,
      "step": 234
    },
    {
      "epoch": 0.7035928143712575,
      "grad_norm": 0.9387475861215742,
      "learning_rate": 7.244774151980466e-06,
      "loss": 0.3556,
      "step": 235
    },
    {
      "epoch": 0.7065868263473054,
      "grad_norm": 1.2605182184599664,
      "learning_rate": 7.223737547398898e-06,
      "loss": 0.3689,
      "step": 236
    },
    {
      "epoch": 0.7095808383233533,
      "grad_norm": 1.06340308742501,
      "learning_rate": 7.20265175817143e-06,
      "loss": 0.3811,
      "step": 237
    },
    {
      "epoch": 0.7125748502994012,
      "grad_norm": 1.1884369289045194,
      "learning_rate": 7.181517250673729e-06,
      "loss": 0.4115,
      "step": 238
    },
    {
      "epoch": 0.7155688622754491,
      "grad_norm": 1.0604750281111033,
      "learning_rate": 7.1603344923590065e-06,
      "loss": 0.368,
      "step": 239
    },
    {
      "epoch": 0.718562874251497,
      "grad_norm": 0.9884120920758936,
      "learning_rate": 7.139103951747694e-06,
      "loss": 0.3964,
      "step": 240
    },
    {
      "epoch": 0.7215568862275449,
      "grad_norm": 0.9882901892475843,
      "learning_rate": 7.1178260984170675e-06,
      "loss": 0.3402,
      "step": 241
    },
    {
      "epoch": 0.7245508982035929,
      "grad_norm": 1.0159073940426744,
      "learning_rate": 7.0965014029908654e-06,
      "loss": 0.3387,
      "step": 242
    },
    {
      "epoch": 0.7275449101796407,
      "grad_norm": 0.9732761192755145,
      "learning_rate": 7.075130337128883e-06,
      "loss": 0.3756,
      "step": 243
    },
    {
      "epoch": 0.7305389221556886,
      "grad_norm": 1.1181467661716453,
      "learning_rate": 7.053713373516538e-06,
      "loss": 0.3976,
      "step": 244
    },
    {
      "epoch": 0.7335329341317365,
      "grad_norm": 0.9914415495598553,
      "learning_rate": 7.03225098585441e-06,
      "loss": 0.406,
      "step": 245
    },
    {
      "epoch": 0.7365269461077845,
      "grad_norm": 0.9349380637899489,
      "learning_rate": 7.0107436488477694e-06,
      "loss": 0.3858,
      "step": 246
    },
    {
      "epoch": 0.7395209580838323,
      "grad_norm": 1.0671641205988223,
      "learning_rate": 6.989191838196083e-06,
      "loss": 0.4026,
      "step": 247
    },
    {
      "epoch": 0.7425149700598802,
      "grad_norm": 1.0353222194746492,
      "learning_rate": 6.9675960305824785e-06,
      "loss": 0.3536,
      "step": 248
    },
    {
      "epoch": 0.7455089820359282,
      "grad_norm": 1.1807206074512484,
      "learning_rate": 6.945956703663212e-06,
      "loss": 0.3996,
      "step": 249
    },
    {
      "epoch": 0.7485029940119761,
      "grad_norm": 0.9688487164109857,
      "learning_rate": 6.9242743360570985e-06,
      "loss": 0.3628,
      "step": 250
    },
    {
      "epoch": 0.7514970059880239,
      "grad_norm": 1.198625172577203,
      "learning_rate": 6.9025494073349284e-06,
      "loss": 0.4126,
      "step": 251
    },
    {
      "epoch": 0.7544910179640718,
      "grad_norm": 0.9178690511255925,
      "learning_rate": 6.880782398008862e-06,
      "loss": 0.3751,
      "step": 252
    },
    {
      "epoch": 0.7574850299401198,
      "grad_norm": 1.1026984639475659,
      "learning_rate": 6.858973789521792e-06,
      "loss": 0.3417,
      "step": 253
    },
    {
      "epoch": 0.7604790419161677,
      "grad_norm": 1.0833627322649082,
      "learning_rate": 6.837124064236709e-06,
      "loss": 0.4053,
      "step": 254
    },
    {
      "epoch": 0.7634730538922155,
      "grad_norm": 0.9820957906659576,
      "learning_rate": 6.815233705426019e-06,
      "loss": 0.3697,
      "step": 255
    },
    {
      "epoch": 0.7664670658682635,
      "grad_norm": 1.0455648368913306,
      "learning_rate": 6.7933031972608644e-06,
      "loss": 0.401,
      "step": 256
    },
    {
      "epoch": 0.7694610778443114,
      "grad_norm": 0.9814496368114252,
      "learning_rate": 6.771333024800411e-06,
      "loss": 0.3802,
      "step": 257
    },
    {
      "epoch": 0.7724550898203593,
      "grad_norm": 1.1922223059369825,
      "learning_rate": 6.74932367398112e-06,
      "loss": 0.3754,
      "step": 258
    },
    {
      "epoch": 0.7754491017964071,
      "grad_norm": 1.011433354510362,
      "learning_rate": 6.727275631605996e-06,
      "loss": 0.3292,
      "step": 259
    },
    {
      "epoch": 0.7784431137724551,
      "grad_norm": 1.0736498595732458,
      "learning_rate": 6.70518938533383e-06,
      "loss": 0.3712,
      "step": 260
    },
    {
      "epoch": 0.781437125748503,
      "grad_norm": 1.1162157690090078,
      "learning_rate": 6.683065423668403e-06,
      "loss": 0.3894,
      "step": 261
    },
    {
      "epoch": 0.7844311377245509,
      "grad_norm": 1.134092488993485,
      "learning_rate": 6.660904235947687e-06,
      "loss": 0.3691,
      "step": 262
    },
    {
      "epoch": 0.7874251497005988,
      "grad_norm": 1.012100290801701,
      "learning_rate": 6.638706312333018e-06,
      "loss": 0.3894,
      "step": 263
    },
    {
      "epoch": 0.7904191616766467,
      "grad_norm": 1.0528739506756517,
      "learning_rate": 6.61647214379826e-06,
      "loss": 0.4274,
      "step": 264
    },
    {
      "epoch": 0.7934131736526946,
      "grad_norm": 1.1867599275325125,
      "learning_rate": 6.594202222118941e-06,
      "loss": 0.3937,
      "step": 265
    },
    {
      "epoch": 0.7964071856287425,
      "grad_norm": 1.0932639266145312,
      "learning_rate": 6.571897039861377e-06,
      "loss": 0.375,
      "step": 266
    },
    {
      "epoch": 0.7994011976047904,
      "grad_norm": 1.02130071424371,
      "learning_rate": 6.549557090371775e-06,
      "loss": 0.3882,
      "step": 267
    },
    {
      "epoch": 0.8023952095808383,
      "grad_norm": 0.9547529629061537,
      "learning_rate": 6.527182867765333e-06,
      "loss": 0.2852,
      "step": 268
    },
    {
      "epoch": 0.8053892215568862,
      "grad_norm": 1.0800225605936187,
      "learning_rate": 6.504774866915291e-06,
      "loss": 0.367,
      "step": 269
    },
    {
      "epoch": 0.8083832335329342,
      "grad_norm": 1.1361032394179078,
      "learning_rate": 6.482333583442002e-06,
      "loss": 0.3461,
      "step": 270
    },
    {
      "epoch": 0.811377245508982,
      "grad_norm": 0.9405938729804465,
      "learning_rate": 6.459859513701967e-06,
      "loss": 0.3741,
      "step": 271
    },
    {
      "epoch": 0.8143712574850299,
      "grad_norm": 1.2342436132113987,
      "learning_rate": 6.437353154776848e-06,
      "loss": 0.3792,
      "step": 272
    },
    {
      "epoch": 0.8173652694610778,
      "grad_norm": 1.0822093992034192,
      "learning_rate": 6.414815004462483e-06,
      "loss": 0.412,
      "step": 273
    },
    {
      "epoch": 0.8203592814371258,
      "grad_norm": 1.0665482591873432,
      "learning_rate": 6.3922455612578715e-06,
      "loss": 0.4108,
      "step": 274
    },
    {
      "epoch": 0.8233532934131736,
      "grad_norm": 1.0201014669636055,
      "learning_rate": 6.369645324354149e-06,
      "loss": 0.3203,
      "step": 275
    },
    {
      "epoch": 0.8263473053892215,
      "grad_norm": 0.9313182024828596,
      "learning_rate": 6.3470147936235485e-06,
      "loss": 0.3795,
      "step": 276
    },
    {
      "epoch": 0.8293413173652695,
      "grad_norm": 0.8738812181770729,
      "learning_rate": 6.3243544696083355e-06,
      "loss": 0.341,
      "step": 277
    },
    {
      "epoch": 0.8323353293413174,
      "grad_norm": 0.9646214219623503,
      "learning_rate": 6.301664853509755e-06,
      "loss": 0.3233,
      "step": 278
    },
    {
      "epoch": 0.8353293413173652,
      "grad_norm": 0.9453214555432041,
      "learning_rate": 6.278946447176924e-06,
      "loss": 0.3685,
      "step": 279
    },
    {
      "epoch": 0.8383233532934131,
      "grad_norm": 0.9155013605437612,
      "learning_rate": 6.256199753095745e-06,
      "loss": 0.339,
      "step": 280
    },
    {
      "epoch": 0.8413173652694611,
      "grad_norm": 0.9723301479000019,
      "learning_rate": 6.233425274377793e-06,
      "loss": 0.3649,
      "step": 281
    },
    {
      "epoch": 0.844311377245509,
      "grad_norm": 1.3034219661741946,
      "learning_rate": 6.21062351474918e-06,
      "loss": 0.414,
      "step": 282
    },
    {
      "epoch": 0.8473053892215568,
      "grad_norm": 1.0763031079194918,
      "learning_rate": 6.18779497853942e-06,
      "loss": 0.3913,
      "step": 283
    },
    {
      "epoch": 0.8502994011976048,
      "grad_norm": 1.060826963618418,
      "learning_rate": 6.164940170670266e-06,
      "loss": 0.402,
      "step": 284
    },
    {
      "epoch": 0.8532934131736527,
      "grad_norm": 0.8904854078118,
      "learning_rate": 6.142059596644557e-06,
      "loss": 0.3196,
      "step": 285
    },
    {
      "epoch": 0.8562874251497006,
      "grad_norm": 1.0443174454117718,
      "learning_rate": 6.11915376253502e-06,
      "loss": 0.349,
      "step": 286
    },
    {
      "epoch": 0.8592814371257484,
      "grad_norm": 1.0898306656222858,
      "learning_rate": 6.096223174973091e-06,
      "loss": 0.3824,
      "step": 287
    },
    {
      "epoch": 0.8622754491017964,
      "grad_norm": 0.9763584233527128,
      "learning_rate": 6.073268341137694e-06,
      "loss": 0.3187,
      "step": 288
    },
    {
      "epoch": 0.8652694610778443,
      "grad_norm": 1.1037005155142565,
      "learning_rate": 6.050289768744042e-06,
      "loss": 0.3035,
      "step": 289
    },
    {
      "epoch": 0.8682634730538922,
      "grad_norm": 1.1552949893574622,
      "learning_rate": 6.0272879660323936e-06,
      "loss": 0.4089,
      "step": 290
    },
    {
      "epoch": 0.8712574850299402,
      "grad_norm": 1.029617548036642,
      "learning_rate": 6.004263441756815e-06,
      "loss": 0.3732,
      "step": 291
    },
    {
      "epoch": 0.874251497005988,
      "grad_norm": 1.0623508649511983,
      "learning_rate": 5.98121670517393e-06,
      "loss": 0.3514,
      "step": 292
    },
    {
      "epoch": 0.8772455089820359,
      "grad_norm": 0.8949625466854254,
      "learning_rate": 5.958148266031654e-06,
      "loss": 0.3273,
      "step": 293
    },
    {
      "epoch": 0.8802395209580839,
      "grad_norm": 0.9891586201706486,
      "learning_rate": 5.935058634557917e-06,
      "loss": 0.3381,
      "step": 294
    },
    {
      "epoch": 0.8832335329341318,
      "grad_norm": 0.8993280891841109,
      "learning_rate": 5.911948321449384e-06,
      "loss": 0.3698,
      "step": 295
    },
    {
      "epoch": 0.8862275449101796,
      "grad_norm": 1.1092255445223,
      "learning_rate": 5.8888178378601565e-06,
      "loss": 0.3865,
      "step": 296
    },
    {
      "epoch": 0.8892215568862275,
      "grad_norm": 1.1517746657898535,
      "learning_rate": 5.865667695390468e-06,
      "loss": 0.3717,
      "step": 297
    },
    {
      "epoch": 0.8922155688622755,
      "grad_norm": 0.9819545381570025,
      "learning_rate": 5.842498406075363e-06,
      "loss": 0.3514,
      "step": 298
    },
    {
      "epoch": 0.8952095808383234,
      "grad_norm": 1.0437201087745172,
      "learning_rate": 5.819310482373381e-06,
      "loss": 0.3679,
      "step": 299
    },
    {
      "epoch": 0.8982035928143712,
      "grad_norm": 1.017297447258417,
      "learning_rate": 5.796104437155213e-06,
      "loss": 0.4133,
      "step": 300
    },
    {
      "epoch": 0.9011976047904192,
      "grad_norm": 1.0426986159591238,
      "learning_rate": 5.772880783692363e-06,
      "loss": 0.3442,
      "step": 301
    },
    {
      "epoch": 0.9041916167664671,
      "grad_norm": 1.133705616825967,
      "learning_rate": 5.749640035645798e-06,
      "loss": 0.3362,
      "step": 302
    },
    {
      "epoch": 0.907185628742515,
      "grad_norm": 1.0806943931447806,
      "learning_rate": 5.726382707054578e-06,
      "loss": 0.3667,
      "step": 303
    },
    {
      "epoch": 0.9101796407185628,
      "grad_norm": 1.0320318529985406,
      "learning_rate": 5.703109312324493e-06,
      "loss": 0.3614,
      "step": 304
    },
    {
      "epoch": 0.9131736526946108,
      "grad_norm": 1.1120370669266448,
      "learning_rate": 5.679820366216684e-06,
      "loss": 0.4372,
      "step": 305
    },
    {
      "epoch": 0.9161676646706587,
      "grad_norm": 0.960621733278799,
      "learning_rate": 5.656516383836263e-06,
      "loss": 0.3198,
      "step": 306
    },
    {
      "epoch": 0.9191616766467066,
      "grad_norm": 1.0171414143063844,
      "learning_rate": 5.6331978806209044e-06,
      "loss": 0.3841,
      "step": 307
    },
    {
      "epoch": 0.9221556886227545,
      "grad_norm": 1.0475670466548357,
      "learning_rate": 5.609865372329461e-06,
      "loss": 0.3479,
      "step": 308
    },
    {
      "epoch": 0.9251497005988024,
      "grad_norm": 0.9816134421329039,
      "learning_rate": 5.586519375030549e-06,
      "loss": 0.313,
      "step": 309
    },
    {
      "epoch": 0.9281437125748503,
      "grad_norm": 0.9226555440216212,
      "learning_rate": 5.5631604050911354e-06,
      "loss": 0.3405,
      "step": 310
    },
    {
      "epoch": 0.9311377245508982,
      "grad_norm": 1.1390317632022344,
      "learning_rate": 5.539788979165115e-06,
      "loss": 0.3195,
      "step": 311
    },
    {
      "epoch": 0.9341317365269461,
      "grad_norm": 0.9262738399132571,
      "learning_rate": 5.516405614181883e-06,
      "loss": 0.3085,
      "step": 312
    },
    {
      "epoch": 0.937125748502994,
      "grad_norm": 0.9017024658346414,
      "learning_rate": 5.4930108273349034e-06,
      "loss": 0.3478,
      "step": 313
    },
    {
      "epoch": 0.9401197604790419,
      "grad_norm": 1.009518131571134,
      "learning_rate": 5.4696051360702725e-06,
      "loss": 0.3277,
      "step": 314
    },
    {
      "epoch": 0.9431137724550899,
      "grad_norm": 1.1427767571052792,
      "learning_rate": 5.446189058075265e-06,
      "loss": 0.3867,
      "step": 315
    },
    {
      "epoch": 0.9461077844311377,
      "grad_norm": 1.036250685743893,
      "learning_rate": 5.4227631112668955e-06,
      "loss": 0.3945,
      "step": 316
    },
    {
      "epoch": 0.9491017964071856,
      "grad_norm": 1.0948913765154202,
      "learning_rate": 5.39932781378045e-06,
      "loss": 0.3893,
      "step": 317
    },
    {
      "epoch": 0.9520958083832335,
      "grad_norm": 0.8970637950602717,
      "learning_rate": 5.375883683958041e-06,
      "loss": 0.3497,
      "step": 318
    },
    {
      "epoch": 0.9550898203592815,
      "grad_norm": 1.0225586335941694,
      "learning_rate": 5.3524312403371255e-06,
      "loss": 0.3653,
      "step": 319
    },
    {
      "epoch": 0.9580838323353293,
      "grad_norm": 1.1349947187399525,
      "learning_rate": 5.328971001639054e-06,
      "loss": 0.3296,
      "step": 320
    },
    {
      "epoch": 0.9610778443113772,
      "grad_norm": 0.9630464347251071,
      "learning_rate": 5.3055034867575825e-06,
      "loss": 0.3334,
      "step": 321
    },
    {
      "epoch": 0.9640718562874252,
      "grad_norm": 1.1766917719584578,
      "learning_rate": 5.282029214747404e-06,
      "loss": 0.3914,
      "step": 322
    },
    {
      "epoch": 0.9670658682634731,
      "grad_norm": 1.0090703010382933,
      "learning_rate": 5.258548704812667e-06,
      "loss": 0.3745,
      "step": 323
    },
    {
      "epoch": 0.9700598802395209,
      "grad_norm": 1.0240983740197453,
      "learning_rate": 5.235062476295488e-06,
      "loss": 0.3577,
      "step": 324
    },
    {
      "epoch": 0.9730538922155688,
      "grad_norm": 1.054207020462211,
      "learning_rate": 5.211571048664469e-06,
      "loss": 0.3121,
      "step": 325
    },
    {
      "epoch": 0.9760479041916168,
      "grad_norm": 1.0418905854977334,
      "learning_rate": 5.188074941503203e-06,
      "loss": 0.3721,
      "step": 326
    },
    {
      "epoch": 0.9790419161676647,
      "grad_norm": 1.0377401459754019,
      "learning_rate": 5.164574674498788e-06,
      "loss": 0.3648,
      "step": 327
    },
    {
      "epoch": 0.9820359281437125,
      "grad_norm": 1.1323566998059742,
      "learning_rate": 5.141070767430331e-06,
      "loss": 0.3516,
      "step": 328
    },
    {
      "epoch": 0.9850299401197605,
      "grad_norm": 1.0206982859117215,
      "learning_rate": 5.117563740157444e-06,
      "loss": 0.3494,
      "step": 329
    },
    {
      "epoch": 0.9880239520958084,
      "grad_norm": 1.1305169158319033,
      "learning_rate": 5.094054112608758e-06,
      "loss": 0.4197,
      "step": 330
    },
    {
      "epoch": 0.9910179640718563,
      "grad_norm": 0.9270024273081134,
      "learning_rate": 5.070542404770413e-06,
      "loss": 0.3946,
      "step": 331
    },
    {
      "epoch": 0.9940119760479041,
      "grad_norm": 0.9487207215039185,
      "learning_rate": 5.047029136674563e-06,
      "loss": 0.3234,
      "step": 332
    },
    {
      "epoch": 0.9970059880239521,
      "grad_norm": 1.1679059724655723,
      "learning_rate": 5.023514828387868e-06,
      "loss": 0.4811,
      "step": 333
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.963466772905833,
      "learning_rate": 5e-06,
      "loss": 0.2822,
      "step": 334
    },
    {
      "epoch": 1.0029940119760479,
      "grad_norm": 1.1038163202178393,
      "learning_rate": 4.976485171612134e-06,
      "loss": 0.2402,
      "step": 335
    },
    {
      "epoch": 1.0059880239520957,
      "grad_norm": 0.9439308178641818,
      "learning_rate": 4.95297086332544e-06,
      "loss": 0.2578,
      "step": 336
    },
    {
      "epoch": 1.0089820359281436,
      "grad_norm": 1.0459901921231076,
      "learning_rate": 4.9294575952295896e-06,
      "loss": 0.274,
      "step": 337
    },
    {
      "epoch": 1.0119760479041917,
      "grad_norm": 1.0752838501456172,
      "learning_rate": 4.905945887391242e-06,
      "loss": 0.3031,
      "step": 338
    },
    {
      "epoch": 1.0149700598802396,
      "grad_norm": 1.0670794122757394,
      "learning_rate": 4.882436259842556e-06,
      "loss": 0.2774,
      "step": 339
    },
    {
      "epoch": 1.0179640718562875,
      "grad_norm": 0.9674785854060131,
      "learning_rate": 4.858929232569671e-06,
      "loss": 0.2693,
      "step": 340
    },
    {
      "epoch": 1.0209580838323353,
      "grad_norm": 1.0418787504336953,
      "learning_rate": 4.835425325501214e-06,
      "loss": 0.2867,
      "step": 341
    },
    {
      "epoch": 1.0239520958083832,
      "grad_norm": 0.9273975514140991,
      "learning_rate": 4.811925058496799e-06,
      "loss": 0.2279,
      "step": 342
    },
    {
      "epoch": 1.026946107784431,
      "grad_norm": 1.1214794184235446,
      "learning_rate": 4.788428951335534e-06,
      "loss": 0.2909,
      "step": 343
    },
    {
      "epoch": 1.029940119760479,
      "grad_norm": 0.9056792166529178,
      "learning_rate": 4.7649375237045135e-06,
      "loss": 0.2628,
      "step": 344
    },
    {
      "epoch": 1.032934131736527,
      "grad_norm": 0.9942035520971275,
      "learning_rate": 4.741451295187333e-06,
      "loss": 0.2406,
      "step": 345
    },
    {
      "epoch": 1.035928143712575,
      "grad_norm": 0.968551910495657,
      "learning_rate": 4.717970785252596e-06,
      "loss": 0.3153,
      "step": 346
    },
    {
      "epoch": 1.0389221556886228,
      "grad_norm": 0.959066511538171,
      "learning_rate": 4.694496513242418e-06,
      "loss": 0.302,
      "step": 347
    },
    {
      "epoch": 1.0419161676646707,
      "grad_norm": 1.0651106382353495,
      "learning_rate": 4.671028998360947e-06,
      "loss": 0.2481,
      "step": 348
    },
    {
      "epoch": 1.0449101796407185,
      "grad_norm": 1.0645658172847647,
      "learning_rate": 4.647568759662876e-06,
      "loss": 0.3041,
      "step": 349
    },
    {
      "epoch": 1.0479041916167664,
      "grad_norm": 0.9641040212080574,
      "learning_rate": 4.624116316041962e-06,
      "loss": 0.2847,
      "step": 350
    },
    {
      "epoch": 1.0508982035928143,
      "grad_norm": 0.9705453522757115,
      "learning_rate": 4.600672186219551e-06,
      "loss": 0.2926,
      "step": 351
    },
    {
      "epoch": 1.0538922155688624,
      "grad_norm": 1.090750995748997,
      "learning_rate": 4.5772368887331044e-06,
      "loss": 0.329,
      "step": 352
    },
    {
      "epoch": 1.0568862275449102,
      "grad_norm": 0.9860942834457607,
      "learning_rate": 4.553810941924735e-06,
      "loss": 0.3182,
      "step": 353
    },
    {
      "epoch": 1.0598802395209581,
      "grad_norm": 0.9947516928313157,
      "learning_rate": 4.530394863929728e-06,
      "loss": 0.2805,
      "step": 354
    },
    {
      "epoch": 1.062874251497006,
      "grad_norm": 1.0152103368161638,
      "learning_rate": 4.506989172665097e-06,
      "loss": 0.2683,
      "step": 355
    },
    {
      "epoch": 1.0658682634730539,
      "grad_norm": 1.0671034757235769,
      "learning_rate": 4.483594385818119e-06,
      "loss": 0.2613,
      "step": 356
    },
    {
      "epoch": 1.0688622754491017,
      "grad_norm": 1.1101237480576047,
      "learning_rate": 4.460211020834887e-06,
      "loss": 0.2776,
      "step": 357
    },
    {
      "epoch": 1.0718562874251496,
      "grad_norm": 1.0926944083273815,
      "learning_rate": 4.436839594908866e-06,
      "loss": 0.3395,
      "step": 358
    },
    {
      "epoch": 1.0748502994011977,
      "grad_norm": 1.162819526757365,
      "learning_rate": 4.4134806249694514e-06,
      "loss": 0.3096,
      "step": 359
    },
    {
      "epoch": 1.0778443113772456,
      "grad_norm": 1.106813840488466,
      "learning_rate": 4.39013462767054e-06,
      "loss": 0.3285,
      "step": 360
    },
    {
      "epoch": 1.0808383233532934,
      "grad_norm": 0.9835710232116306,
      "learning_rate": 4.366802119379098e-06,
      "loss": 0.225,
      "step": 361
    },
    {
      "epoch": 1.0838323353293413,
      "grad_norm": 0.936604619916719,
      "learning_rate": 4.34348361616374e-06,
      "loss": 0.3305,
      "step": 362
    },
    {
      "epoch": 1.0868263473053892,
      "grad_norm": 1.186118149852767,
      "learning_rate": 4.3201796337833165e-06,
      "loss": 0.3048,
      "step": 363
    },
    {
      "epoch": 1.089820359281437,
      "grad_norm": 1.2077002584599066,
      "learning_rate": 4.29689068767551e-06,
      "loss": 0.333,
      "step": 364
    },
    {
      "epoch": 1.092814371257485,
      "grad_norm": 0.9936218749411708,
      "learning_rate": 4.273617292945425e-06,
      "loss": 0.2776,
      "step": 365
    },
    {
      "epoch": 1.095808383233533,
      "grad_norm": 0.9510483327426793,
      "learning_rate": 4.250359964354203e-06,
      "loss": 0.2886,
      "step": 366
    },
    {
      "epoch": 1.098802395209581,
      "grad_norm": 1.0933620437452058,
      "learning_rate": 4.227119216307637e-06,
      "loss": 0.2919,
      "step": 367
    },
    {
      "epoch": 1.1017964071856288,
      "grad_norm": 1.0497701233499976,
      "learning_rate": 4.203895562844789e-06,
      "loss": 0.3287,
      "step": 368
    },
    {
      "epoch": 1.1047904191616766,
      "grad_norm": 1.1246610130672805,
      "learning_rate": 4.18068951762662e-06,
      "loss": 0.3377,
      "step": 369
    },
    {
      "epoch": 1.1077844311377245,
      "grad_norm": 1.0272251389183615,
      "learning_rate": 4.157501593924638e-06,
      "loss": 0.2609,
      "step": 370
    },
    {
      "epoch": 1.1107784431137724,
      "grad_norm": 0.9591853088905555,
      "learning_rate": 4.134332304609533e-06,
      "loss": 0.3029,
      "step": 371
    },
    {
      "epoch": 1.1137724550898203,
      "grad_norm": 0.8559562239842542,
      "learning_rate": 4.111182162139844e-06,
      "loss": 0.2827,
      "step": 372
    },
    {
      "epoch": 1.1167664670658684,
      "grad_norm": 1.1089804658991524,
      "learning_rate": 4.088051678550617e-06,
      "loss": 0.2745,
      "step": 373
    },
    {
      "epoch": 1.1197604790419162,
      "grad_norm": 1.0692427931482376,
      "learning_rate": 4.064941365442084e-06,
      "loss": 0.2911,
      "step": 374
    },
    {
      "epoch": 1.122754491017964,
      "grad_norm": 1.2437802366455162,
      "learning_rate": 4.041851733968348e-06,
      "loss": 0.335,
      "step": 375
    },
    {
      "epoch": 1.125748502994012,
      "grad_norm": 1.0566923905800043,
      "learning_rate": 4.018783294826071e-06,
      "loss": 0.3087,
      "step": 376
    },
    {
      "epoch": 1.1287425149700598,
      "grad_norm": 1.123414045867036,
      "learning_rate": 3.995736558243186e-06,
      "loss": 0.3271,
      "step": 377
    },
    {
      "epoch": 1.1317365269461077,
      "grad_norm": 1.0045033372465337,
      "learning_rate": 3.972712033967608e-06,
      "loss": 0.2893,
      "step": 378
    },
    {
      "epoch": 1.1347305389221556,
      "grad_norm": 0.9903061130943368,
      "learning_rate": 3.949710231255961e-06,
      "loss": 0.3376,
      "step": 379
    },
    {
      "epoch": 1.1377245508982037,
      "grad_norm": 1.2510349069206157,
      "learning_rate": 3.926731658862307e-06,
      "loss": 0.2889,
      "step": 380
    },
    {
      "epoch": 1.1407185628742516,
      "grad_norm": 1.0038425254175143,
      "learning_rate": 3.903776825026912e-06,
      "loss": 0.3289,
      "step": 381
    },
    {
      "epoch": 1.1437125748502994,
      "grad_norm": 0.9938234890357625,
      "learning_rate": 3.8808462374649805e-06,
      "loss": 0.2998,
      "step": 382
    },
    {
      "epoch": 1.1467065868263473,
      "grad_norm": 1.0246750644660116,
      "learning_rate": 3.857940403355444e-06,
      "loss": 0.2662,
      "step": 383
    },
    {
      "epoch": 1.1497005988023952,
      "grad_norm": 1.039803508748742,
      "learning_rate": 3.8350598293297345e-06,
      "loss": 0.3201,
      "step": 384
    },
    {
      "epoch": 1.152694610778443,
      "grad_norm": 1.014959652124324,
      "learning_rate": 3.8122050214605822e-06,
      "loss": 0.2702,
      "step": 385
    },
    {
      "epoch": 1.1556886227544911,
      "grad_norm": 0.9975291572056869,
      "learning_rate": 3.7893764852508207e-06,
      "loss": 0.3035,
      "step": 386
    },
    {
      "epoch": 1.158682634730539,
      "grad_norm": 1.0199777385264022,
      "learning_rate": 3.766574725622208e-06,
      "loss": 0.2586,
      "step": 387
    },
    {
      "epoch": 1.1616766467065869,
      "grad_norm": 0.9759448313629795,
      "learning_rate": 3.7438002469042567e-06,
      "loss": 0.289,
      "step": 388
    },
    {
      "epoch": 1.1646706586826348,
      "grad_norm": 1.0765071106811648,
      "learning_rate": 3.721053552823078e-06,
      "loss": 0.2991,
      "step": 389
    },
    {
      "epoch": 1.1676646706586826,
      "grad_norm": 1.0202142143948365,
      "learning_rate": 3.698335146490246e-06,
      "loss": 0.2855,
      "step": 390
    },
    {
      "epoch": 1.1706586826347305,
      "grad_norm": 1.1582146793042545,
      "learning_rate": 3.675645530391665e-06,
      "loss": 0.2823,
      "step": 391
    },
    {
      "epoch": 1.1736526946107784,
      "grad_norm": 1.0397077710100364,
      "learning_rate": 3.652985206376455e-06,
      "loss": 0.2614,
      "step": 392
    },
    {
      "epoch": 1.1766467065868262,
      "grad_norm": 0.8459228285294651,
      "learning_rate": 3.630354675645853e-06,
      "loss": 0.2509,
      "step": 393
    },
    {
      "epoch": 1.1796407185628743,
      "grad_norm": 1.074927652267696,
      "learning_rate": 3.6077544387421293e-06,
      "loss": 0.3075,
      "step": 394
    },
    {
      "epoch": 1.1826347305389222,
      "grad_norm": 1.1066622312555507,
      "learning_rate": 3.5851849955375177e-06,
      "loss": 0.2545,
      "step": 395
    },
    {
      "epoch": 1.18562874251497,
      "grad_norm": 1.0380594135380292,
      "learning_rate": 3.5626468452231534e-06,
      "loss": 0.2664,
      "step": 396
    },
    {
      "epoch": 1.188622754491018,
      "grad_norm": 1.0350740479215617,
      "learning_rate": 3.540140486298035e-06,
      "loss": 0.2239,
      "step": 397
    },
    {
      "epoch": 1.1916167664670658,
      "grad_norm": 0.9671972197294936,
      "learning_rate": 3.517666416557999e-06,
      "loss": 0.323,
      "step": 398
    },
    {
      "epoch": 1.1946107784431137,
      "grad_norm": 1.0553578297585076,
      "learning_rate": 3.495225133084712e-06,
      "loss": 0.2942,
      "step": 399
    },
    {
      "epoch": 1.1976047904191618,
      "grad_norm": 0.9668856112058133,
      "learning_rate": 3.472817132234669e-06,
      "loss": 0.3097,
      "step": 400
    },
    {
      "epoch": 1.1976047904191618,
      "eval_loss": 0.3693498969078064,
      "eval_runtime": 2.0745,
      "eval_samples_per_second": 26.03,
      "eval_steps_per_second": 6.749,
      "step": 400
    },
    {
      "epoch": 1.2005988023952097,
      "grad_norm": 0.9818768718871981,
      "learning_rate": 3.4504429096282246e-06,
      "loss": 0.3246,
      "step": 401
    },
    {
      "epoch": 1.2035928143712575,
      "grad_norm": 1.056124662708042,
      "learning_rate": 3.428102960138625e-06,
      "loss": 0.3118,
      "step": 402
    },
    {
      "epoch": 1.2065868263473054,
      "grad_norm": 1.085430749955511,
      "learning_rate": 3.405797777881059e-06,
      "loss": 0.3247,
      "step": 403
    },
    {
      "epoch": 1.2095808383233533,
      "grad_norm": 1.109869803902745,
      "learning_rate": 3.3835278562017405e-06,
      "loss": 0.2679,
      "step": 404
    },
    {
      "epoch": 1.2125748502994012,
      "grad_norm": 0.9653457634585856,
      "learning_rate": 3.3612936876669834e-06,
      "loss": 0.2724,
      "step": 405
    },
    {
      "epoch": 1.215568862275449,
      "grad_norm": 0.9138963278075037,
      "learning_rate": 3.3390957640523147e-06,
      "loss": 0.2675,
      "step": 406
    },
    {
      "epoch": 1.218562874251497,
      "grad_norm": 0.9328413785625903,
      "learning_rate": 3.3169345763315986e-06,
      "loss": 0.2853,
      "step": 407
    },
    {
      "epoch": 1.221556886227545,
      "grad_norm": 1.0516384087001678,
      "learning_rate": 3.29481061466617e-06,
      "loss": 0.3054,
      "step": 408
    },
    {
      "epoch": 1.2245508982035929,
      "grad_norm": 1.1253453724856757,
      "learning_rate": 3.2727243683940045e-06,
      "loss": 0.2819,
      "step": 409
    },
    {
      "epoch": 1.2275449101796407,
      "grad_norm": 1.0571410903586207,
      "learning_rate": 3.2506763260188824e-06,
      "loss": 0.3098,
      "step": 410
    },
    {
      "epoch": 1.2305389221556886,
      "grad_norm": 0.9251864173640302,
      "learning_rate": 3.2286669751995905e-06,
      "loss": 0.2918,
      "step": 411
    },
    {
      "epoch": 1.2335329341317365,
      "grad_norm": 1.1343137223136206,
      "learning_rate": 3.2066968027391377e-06,
      "loss": 0.2913,
      "step": 412
    },
    {
      "epoch": 1.2365269461077844,
      "grad_norm": 0.9482192781318599,
      "learning_rate": 3.1847662945739833e-06,
      "loss": 0.2809,
      "step": 413
    },
    {
      "epoch": 1.2395209580838324,
      "grad_norm": 1.0489394179575684,
      "learning_rate": 3.1628759357632943e-06,
      "loss": 0.307,
      "step": 414
    },
    {
      "epoch": 1.2425149700598803,
      "grad_norm": 1.1995655945447297,
      "learning_rate": 3.1410262104782086e-06,
      "loss": 0.2633,
      "step": 415
    },
    {
      "epoch": 1.2455089820359282,
      "grad_norm": 1.0965588624520188,
      "learning_rate": 3.119217601991139e-06,
      "loss": 0.2885,
      "step": 416
    },
    {
      "epoch": 1.248502994011976,
      "grad_norm": 1.0978620772266037,
      "learning_rate": 3.0974505926650724e-06,
      "loss": 0.2449,
      "step": 417
    },
    {
      "epoch": 1.251497005988024,
      "grad_norm": 0.9566279299243072,
      "learning_rate": 3.0757256639429027e-06,
      "loss": 0.2785,
      "step": 418
    },
    {
      "epoch": 1.2544910179640718,
      "grad_norm": 1.080252236964104,
      "learning_rate": 3.0540432963367907e-06,
      "loss": 0.3217,
      "step": 419
    },
    {
      "epoch": 1.2574850299401197,
      "grad_norm": 1.063971060965149,
      "learning_rate": 3.032403969417523e-06,
      "loss": 0.2812,
      "step": 420
    },
    {
      "epoch": 1.2604790419161676,
      "grad_norm": 0.9704172296003442,
      "learning_rate": 3.010808161803917e-06,
      "loss": 0.267,
      "step": 421
    },
    {
      "epoch": 1.2634730538922156,
      "grad_norm": 0.9939978410887771,
      "learning_rate": 2.9892563511522305e-06,
      "loss": 0.3196,
      "step": 422
    },
    {
      "epoch": 1.2664670658682635,
      "grad_norm": 1.0217305858381498,
      "learning_rate": 2.9677490141455915e-06,
      "loss": 0.2686,
      "step": 423
    },
    {
      "epoch": 1.2694610778443114,
      "grad_norm": 1.0077762369146446,
      "learning_rate": 2.946286626483463e-06,
      "loss": 0.321,
      "step": 424
    },
    {
      "epoch": 1.2724550898203593,
      "grad_norm": 1.2045671843348047,
      "learning_rate": 2.924869662871117e-06,
      "loss": 0.2911,
      "step": 425
    },
    {
      "epoch": 1.2754491017964071,
      "grad_norm": 1.0330676293641845,
      "learning_rate": 2.903498597009136e-06,
      "loss": 0.2989,
      "step": 426
    },
    {
      "epoch": 1.278443113772455,
      "grad_norm": 1.116401826098273,
      "learning_rate": 2.8821739015829338e-06,
      "loss": 0.2991,
      "step": 427
    },
    {
      "epoch": 1.281437125748503,
      "grad_norm": 1.0183731095003274,
      "learning_rate": 2.8608960482523058e-06,
      "loss": 0.285,
      "step": 428
    },
    {
      "epoch": 1.284431137724551,
      "grad_norm": 1.0146098027998318,
      "learning_rate": 2.839665507640992e-06,
      "loss": 0.2796,
      "step": 429
    },
    {
      "epoch": 1.2874251497005988,
      "grad_norm": 0.9389899747366528,
      "learning_rate": 2.818482749326272e-06,
      "loss": 0.3354,
      "step": 430
    },
    {
      "epoch": 1.2904191616766467,
      "grad_norm": 0.9334020290971811,
      "learning_rate": 2.797348241828569e-06,
      "loss": 0.2898,
      "step": 431
    },
    {
      "epoch": 1.2934131736526946,
      "grad_norm": 0.9367010903945576,
      "learning_rate": 2.776262452601104e-06,
      "loss": 0.3172,
      "step": 432
    },
    {
      "epoch": 1.2964071856287425,
      "grad_norm": 1.0293549689092885,
      "learning_rate": 2.7552258480195348e-06,
      "loss": 0.2794,
      "step": 433
    },
    {
      "epoch": 1.2994011976047903,
      "grad_norm": 0.9240368132541306,
      "learning_rate": 2.734238893371667e-06,
      "loss": 0.2701,
      "step": 434
    },
    {
      "epoch": 1.3023952095808382,
      "grad_norm": 0.9906999330027314,
      "learning_rate": 2.7133020528471322e-06,
      "loss": 0.2809,
      "step": 435
    },
    {
      "epoch": 1.3053892215568863,
      "grad_norm": 0.9744488517144564,
      "learning_rate": 2.6924157895271563e-06,
      "loss": 0.293,
      "step": 436
    },
    {
      "epoch": 1.3083832335329342,
      "grad_norm": 1.0033858656719574,
      "learning_rate": 2.671580565374282e-06,
      "loss": 0.3217,
      "step": 437
    },
    {
      "epoch": 1.311377245508982,
      "grad_norm": 1.011600307309713,
      "learning_rate": 2.6507968412221763e-06,
      "loss": 0.2422,
      "step": 438
    },
    {
      "epoch": 1.31437125748503,
      "grad_norm": 1.11189134066282,
      "learning_rate": 2.6300650767654234e-06,
      "loss": 0.2744,
      "step": 439
    },
    {
      "epoch": 1.3173652694610778,
      "grad_norm": 0.9469018716498592,
      "learning_rate": 2.6093857305493666e-06,
      "loss": 0.2858,
      "step": 440
    },
    {
      "epoch": 1.3203592814371259,
      "grad_norm": 0.9803118925621244,
      "learning_rate": 2.588759259959962e-06,
      "loss": 0.2624,
      "step": 441
    },
    {
      "epoch": 1.3233532934131738,
      "grad_norm": 1.0239619282343333,
      "learning_rate": 2.568186121213658e-06,
      "loss": 0.3087,
      "step": 442
    },
    {
      "epoch": 1.3263473053892216,
      "grad_norm": 1.0337294346895696,
      "learning_rate": 2.547666769347312e-06,
      "loss": 0.2979,
      "step": 443
    },
    {
      "epoch": 1.3293413173652695,
      "grad_norm": 0.9690903080271954,
      "learning_rate": 2.5272016582081236e-06,
      "loss": 0.353,
      "step": 444
    },
    {
      "epoch": 1.3323353293413174,
      "grad_norm": 1.0917776679498428,
      "learning_rate": 2.5067912404435952e-06,
      "loss": 0.3154,
      "step": 445
    },
    {
      "epoch": 1.3353293413173652,
      "grad_norm": 1.0341877452666952,
      "learning_rate": 2.486435967491516e-06,
      "loss": 0.2781,
      "step": 446
    },
    {
      "epoch": 1.3383233532934131,
      "grad_norm": 1.0088471355077344,
      "learning_rate": 2.4661362895699903e-06,
      "loss": 0.2732,
      "step": 447
    },
    {
      "epoch": 1.341317365269461,
      "grad_norm": 1.0429412895760934,
      "learning_rate": 2.445892655667462e-06,
      "loss": 0.3666,
      "step": 448
    },
    {
      "epoch": 1.3443113772455089,
      "grad_norm": 1.1443662715754788,
      "learning_rate": 2.425705513532798e-06,
      "loss": 0.3141,
      "step": 449
    },
    {
      "epoch": 1.347305389221557,
      "grad_norm": 0.999254604061352,
      "learning_rate": 2.4055753096653795e-06,
      "loss": 0.3234,
      "step": 450
    },
    {
      "epoch": 1.3502994011976048,
      "grad_norm": 0.9554262386411405,
      "learning_rate": 2.3855024893052286e-06,
      "loss": 0.2588,
      "step": 451
    },
    {
      "epoch": 1.3532934131736527,
      "grad_norm": 1.0092424482732858,
      "learning_rate": 2.365487496423152e-06,
      "loss": 0.2867,
      "step": 452
    },
    {
      "epoch": 1.3562874251497006,
      "grad_norm": 0.8824405698753732,
      "learning_rate": 2.3455307737109338e-06,
      "loss": 0.2418,
      "step": 453
    },
    {
      "epoch": 1.3592814371257484,
      "grad_norm": 0.9269365911693727,
      "learning_rate": 2.3256327625715345e-06,
      "loss": 0.2406,
      "step": 454
    },
    {
      "epoch": 1.3622754491017965,
      "grad_norm": 1.0683071236638744,
      "learning_rate": 2.3057939031093346e-06,
      "loss": 0.2765,
      "step": 455
    },
    {
      "epoch": 1.3652694610778444,
      "grad_norm": 0.9094068896182096,
      "learning_rate": 2.2860146341203936e-06,
      "loss": 0.2669,
      "step": 456
    },
    {
      "epoch": 1.3682634730538923,
      "grad_norm": 1.0430179524496415,
      "learning_rate": 2.2662953930827546e-06,
      "loss": 0.3522,
      "step": 457
    },
    {
      "epoch": 1.3712574850299402,
      "grad_norm": 1.0967254686382057,
      "learning_rate": 2.2466366161467528e-06,
      "loss": 0.2885,
      "step": 458
    },
    {
      "epoch": 1.374251497005988,
      "grad_norm": 1.113655664070299,
      "learning_rate": 2.227038738125385e-06,
      "loss": 0.2621,
      "step": 459
    },
    {
      "epoch": 1.377245508982036,
      "grad_norm": 1.1947268702376181,
      "learning_rate": 2.207502192484685e-06,
      "loss": 0.28,
      "step": 460
    },
    {
      "epoch": 1.3802395209580838,
      "grad_norm": 0.928199271967082,
      "learning_rate": 2.188027411334131e-06,
      "loss": 0.2767,
      "step": 461
    },
    {
      "epoch": 1.3832335329341316,
      "grad_norm": 1.0741630054105522,
      "learning_rate": 2.1686148254171012e-06,
      "loss": 0.2768,
      "step": 462
    },
    {
      "epoch": 1.3862275449101795,
      "grad_norm": 1.186068232666103,
      "learning_rate": 2.1492648641013305e-06,
      "loss": 0.2665,
      "step": 463
    },
    {
      "epoch": 1.3892215568862276,
      "grad_norm": 0.9210593951202511,
      "learning_rate": 2.1299779553694323e-06,
      "loss": 0.2659,
      "step": 464
    },
    {
      "epoch": 1.3922155688622755,
      "grad_norm": 1.1044867591546301,
      "learning_rate": 2.1107545258094135e-06,
      "loss": 0.2803,
      "step": 465
    },
    {
      "epoch": 1.3952095808383234,
      "grad_norm": 1.031573990487462,
      "learning_rate": 2.0915950006052555e-06,
      "loss": 0.233,
      "step": 466
    },
    {
      "epoch": 1.3982035928143712,
      "grad_norm": 0.9655158836794998,
      "learning_rate": 2.0724998035274947e-06,
      "loss": 0.2764,
      "step": 467
    },
    {
      "epoch": 1.401197604790419,
      "grad_norm": 1.0172003587953486,
      "learning_rate": 2.053469356923865e-06,
      "loss": 0.229,
      "step": 468
    },
    {
      "epoch": 1.4041916167664672,
      "grad_norm": 1.0067623372218106,
      "learning_rate": 2.0345040817099433e-06,
      "loss": 0.2907,
      "step": 469
    },
    {
      "epoch": 1.407185628742515,
      "grad_norm": 0.9480842454728966,
      "learning_rate": 2.0156043973598475e-06,
      "loss": 0.2587,
      "step": 470
    },
    {
      "epoch": 1.410179640718563,
      "grad_norm": 1.1560079443177578,
      "learning_rate": 1.996770721896957e-06,
      "loss": 0.2935,
      "step": 471
    },
    {
      "epoch": 1.4131736526946108,
      "grad_norm": 1.1309043619436994,
      "learning_rate": 1.9780034718846653e-06,
      "loss": 0.2769,
      "step": 472
    },
    {
      "epoch": 1.4161676646706587,
      "grad_norm": 0.9960471449258698,
      "learning_rate": 1.9593030624171683e-06,
      "loss": 0.3077,
      "step": 473
    },
    {
      "epoch": 1.4191616766467066,
      "grad_norm": 1.1157076766455758,
      "learning_rate": 1.9406699071102774e-06,
      "loss": 0.3164,
      "step": 474
    },
    {
      "epoch": 1.4221556886227544,
      "grad_norm": 0.9781452758812129,
      "learning_rate": 1.9221044180922833e-06,
      "loss": 0.313,
      "step": 475
    },
    {
      "epoch": 1.4251497005988023,
      "grad_norm": 1.0510296364291654,
      "learning_rate": 1.9036070059948253e-06,
      "loss": 0.2925,
      "step": 476
    },
    {
      "epoch": 1.4281437125748502,
      "grad_norm": 1.1121349316079836,
      "learning_rate": 1.885178079943823e-06,
      "loss": 0.3117,
      "step": 477
    },
    {
      "epoch": 1.4311377245508983,
      "grad_norm": 1.0708487444353973,
      "learning_rate": 1.866818047550419e-06,
      "loss": 0.3608,
      "step": 478
    },
    {
      "epoch": 1.4341317365269461,
      "grad_norm": 0.9472532758442945,
      "learning_rate": 1.8485273149019655e-06,
      "loss": 0.2726,
      "step": 479
    },
    {
      "epoch": 1.437125748502994,
      "grad_norm": 1.1274922482499445,
      "learning_rate": 1.8303062865530407e-06,
      "loss": 0.255,
      "step": 480
    },
    {
      "epoch": 1.4401197604790419,
      "grad_norm": 1.089568199658486,
      "learning_rate": 1.8121553655165058e-06,
      "loss": 0.2699,
      "step": 481
    },
    {
      "epoch": 1.4431137724550898,
      "grad_norm": 1.0546253995538688,
      "learning_rate": 1.7940749532545832e-06,
      "loss": 0.326,
      "step": 482
    },
    {
      "epoch": 1.4461077844311379,
      "grad_norm": 1.2264137743297507,
      "learning_rate": 1.7760654496699876e-06,
      "loss": 0.282,
      "step": 483
    },
    {
      "epoch": 1.4491017964071857,
      "grad_norm": 1.0060274414788548,
      "learning_rate": 1.7581272530970666e-06,
      "loss": 0.2983,
      "step": 484
    },
    {
      "epoch": 1.4520958083832336,
      "grad_norm": 0.9619347223999136,
      "learning_rate": 1.7402607602930106e-06,
      "loss": 0.28,
      "step": 485
    },
    {
      "epoch": 1.4550898203592815,
      "grad_norm": 1.1355226571102557,
      "learning_rate": 1.7224663664290537e-06,
      "loss": 0.3382,
      "step": 486
    },
    {
      "epoch": 1.4580838323353293,
      "grad_norm": 1.0086173130977534,
      "learning_rate": 1.7047444650817518e-06,
      "loss": 0.2872,
      "step": 487
    },
    {
      "epoch": 1.4610778443113772,
      "grad_norm": 1.1127648516229565,
      "learning_rate": 1.6870954482242707e-06,
      "loss": 0.2724,
      "step": 488
    },
    {
      "epoch": 1.464071856287425,
      "grad_norm": 0.9832310826065905,
      "learning_rate": 1.669519706217711e-06,
      "loss": 0.2437,
      "step": 489
    },
    {
      "epoch": 1.467065868263473,
      "grad_norm": 1.0622879657436979,
      "learning_rate": 1.652017627802487e-06,
      "loss": 0.3091,
      "step": 490
    },
    {
      "epoch": 1.4700598802395208,
      "grad_norm": 0.9386687663863547,
      "learning_rate": 1.6345896000897122e-06,
      "loss": 0.2635,
      "step": 491
    },
    {
      "epoch": 1.473053892215569,
      "grad_norm": 0.897937240725058,
      "learning_rate": 1.6172360085526567e-06,
      "loss": 0.2986,
      "step": 492
    },
    {
      "epoch": 1.4760479041916168,
      "grad_norm": 0.9020970971560415,
      "learning_rate": 1.5999572370182016e-06,
      "loss": 0.2997,
      "step": 493
    },
    {
      "epoch": 1.4790419161676647,
      "grad_norm": 1.0441938176698309,
      "learning_rate": 1.5827536676583643e-06,
      "loss": 0.2974,
      "step": 494
    },
    {
      "epoch": 1.4820359281437125,
      "grad_norm": 1.0375052996463812,
      "learning_rate": 1.5656256809818343e-06,
      "loss": 0.2815,
      "step": 495
    },
    {
      "epoch": 1.4850299401197604,
      "grad_norm": 1.1276079811444029,
      "learning_rate": 1.54857365582557e-06,
      "loss": 0.2529,
      "step": 496
    },
    {
      "epoch": 1.4880239520958085,
      "grad_norm": 1.0318039132846863,
      "learning_rate": 1.5315979693464039e-06,
      "loss": 0.2983,
      "step": 497
    },
    {
      "epoch": 1.4910179640718564,
      "grad_norm": 1.153390011507833,
      "learning_rate": 1.5146989970127158e-06,
      "loss": 0.2511,
      "step": 498
    },
    {
      "epoch": 1.4940119760479043,
      "grad_norm": 1.0532180938840872,
      "learning_rate": 1.4978771125961177e-06,
      "loss": 0.2952,
      "step": 499
    },
    {
      "epoch": 1.4970059880239521,
      "grad_norm": 0.9281788535505504,
      "learning_rate": 1.4811326881631937e-06,
      "loss": 0.2927,
      "step": 500
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9857787780501807,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.2885,
      "step": 501
    },
    {
      "epoch": 1.5029940119760479,
      "grad_norm": 1.0408667450292723,
      "learning_rate": 1.4478776989401949e-06,
      "loss": 0.3037,
      "step": 502
    },
    {
      "epoch": 1.5059880239520957,
      "grad_norm": 1.0816927299696908,
      "learning_rate": 1.4313678696842559e-06,
      "loss": 0.2804,
      "step": 503
    },
    {
      "epoch": 1.5089820359281436,
      "grad_norm": 1.0188952138113099,
      "learning_rate": 1.4149369714639856e-06,
      "loss": 0.2801,
      "step": 504
    },
    {
      "epoch": 1.5119760479041915,
      "grad_norm": 1.0326848434661613,
      "learning_rate": 1.3985853676981316e-06,
      "loss": 0.289,
      "step": 505
    },
    {
      "epoch": 1.5149700598802394,
      "grad_norm": 1.029640567846141,
      "learning_rate": 1.3823134200516043e-06,
      "loss": 0.2796,
      "step": 506
    },
    {
      "epoch": 1.5179640718562875,
      "grad_norm": 1.1331584019397833,
      "learning_rate": 1.366121488427481e-06,
      "loss": 0.2752,
      "step": 507
    },
    {
      "epoch": 1.5209580838323353,
      "grad_norm": 1.1161301605607417,
      "learning_rate": 1.3500099309590397e-06,
      "loss": 0.2967,
      "step": 508
    },
    {
      "epoch": 1.5239520958083832,
      "grad_norm": 1.0801243931631266,
      "learning_rate": 1.3339791040018479e-06,
      "loss": 0.3321,
      "step": 509
    },
    {
      "epoch": 1.5269461077844313,
      "grad_norm": 0.9588033298979124,
      "learning_rate": 1.3180293621258694e-06,
      "loss": 0.2847,
      "step": 510
    },
    {
      "epoch": 1.5299401197604792,
      "grad_norm": 1.0714461729400162,
      "learning_rate": 1.3021610581076316e-06,
      "loss": 0.2462,
      "step": 511
    },
    {
      "epoch": 1.532934131736527,
      "grad_norm": 1.1293384420310908,
      "learning_rate": 1.2863745429224145e-06,
      "loss": 0.285,
      "step": 512
    },
    {
      "epoch": 1.535928143712575,
      "grad_norm": 0.9787160982168177,
      "learning_rate": 1.270670165736499e-06,
      "loss": 0.2895,
      "step": 513
    },
    {
      "epoch": 1.5389221556886228,
      "grad_norm": 1.2138400284810715,
      "learning_rate": 1.2550482738994284e-06,
      "loss": 0.3481,
      "step": 514
    },
    {
      "epoch": 1.5419161676646707,
      "grad_norm": 1.0577605935604286,
      "learning_rate": 1.239509212936343e-06,
      "loss": 0.2876,
      "step": 515
    },
    {
      "epoch": 1.5449101796407185,
      "grad_norm": 1.0634837772496648,
      "learning_rate": 1.22405332654032e-06,
      "loss": 0.2729,
      "step": 516
    },
    {
      "epoch": 1.5479041916167664,
      "grad_norm": 0.9020595183913013,
      "learning_rate": 1.2086809565647877e-06,
      "loss": 0.2677,
      "step": 517
    },
    {
      "epoch": 1.5508982035928143,
      "grad_norm": 1.2959297717360532,
      "learning_rate": 1.1933924430159571e-06,
      "loss": 0.2792,
      "step": 518
    },
    {
      "epoch": 1.5538922155688621,
      "grad_norm": 1.0110483516177153,
      "learning_rate": 1.1781881240452958e-06,
      "loss": 0.2822,
      "step": 519
    },
    {
      "epoch": 1.55688622754491,
      "grad_norm": 1.0446390155714294,
      "learning_rate": 1.1630683359420653e-06,
      "loss": 0.2935,
      "step": 520
    },
    {
      "epoch": 1.5598802395209581,
      "grad_norm": 1.0690986124227901,
      "learning_rate": 1.1480334131258626e-06,
      "loss": 0.2289,
      "step": 521
    },
    {
      "epoch": 1.562874251497006,
      "grad_norm": 1.1430403868252486,
      "learning_rate": 1.1330836881392405e-06,
      "loss": 0.2718,
      "step": 522
    },
    {
      "epoch": 1.5658682634730539,
      "grad_norm": 1.0375537372951282,
      "learning_rate": 1.11821949164034e-06,
      "loss": 0.297,
      "step": 523
    },
    {
      "epoch": 1.568862275449102,
      "grad_norm": 1.176974027653935,
      "learning_rate": 1.103441152395588e-06,
      "loss": 0.3502,
      "step": 524
    },
    {
      "epoch": 1.5718562874251498,
      "grad_norm": 1.0696722027143732,
      "learning_rate": 1.088748997272414e-06,
      "loss": 0.2993,
      "step": 525
    },
    {
      "epoch": 1.5748502994011977,
      "grad_norm": 1.0346499714868336,
      "learning_rate": 1.0741433512320316e-06,
      "loss": 0.2329,
      "step": 526
    },
    {
      "epoch": 1.5778443113772456,
      "grad_norm": 1.1517848564229314,
      "learning_rate": 1.0596245373222424e-06,
      "loss": 0.2675,
      "step": 527
    },
    {
      "epoch": 1.5808383233532934,
      "grad_norm": 1.0284607566075672,
      "learning_rate": 1.045192876670298e-06,
      "loss": 0.3123,
      "step": 528
    },
    {
      "epoch": 1.5838323353293413,
      "grad_norm": 1.011579712301013,
      "learning_rate": 1.0308486884757868e-06,
      "loss": 0.3192,
      "step": 529
    },
    {
      "epoch": 1.5868263473053892,
      "grad_norm": 0.9498420839416503,
      "learning_rate": 1.0165922900035886e-06,
      "loss": 0.2904,
      "step": 530
    },
    {
      "epoch": 1.589820359281437,
      "grad_norm": 0.9815837486238403,
      "learning_rate": 1.0024239965768417e-06,
      "loss": 0.2808,
      "step": 531
    },
    {
      "epoch": 1.592814371257485,
      "grad_norm": 1.0522449608275906,
      "learning_rate": 9.883441215699824e-07,
      "loss": 0.2142,
      "step": 532
    },
    {
      "epoch": 1.5958083832335328,
      "grad_norm": 1.2241523212020757,
      "learning_rate": 9.74352976401805e-07,
      "loss": 0.3493,
      "step": 533
    },
    {
      "epoch": 1.5988023952095807,
      "grad_norm": 1.1753755980088134,
      "learning_rate": 9.604508705285765e-07,
      "loss": 0.2939,
      "step": 534
    },
    {
      "epoch": 1.6017964071856288,
      "grad_norm": 1.0826047408827566,
      "learning_rate": 9.466381114371942e-07,
      "loss": 0.2975,
      "step": 535
    },
    {
      "epoch": 1.6047904191616766,
      "grad_norm": 1.0317977756518482,
      "learning_rate": 9.329150046383773e-07,
      "loss": 0.2789,
      "step": 536
    },
    {
      "epoch": 1.6077844311377245,
      "grad_norm": 1.0904599792581928,
      "learning_rate": 9.192818536599213e-07,
      "loss": 0.2923,
      "step": 537
    },
    {
      "epoch": 1.6107784431137726,
      "grad_norm": 1.0215977387126998,
      "learning_rate": 9.057389600399719e-07,
      "loss": 0.2997,
      "step": 538
    },
    {
      "epoch": 1.6137724550898205,
      "grad_norm": 0.976235862219578,
      "learning_rate": 8.922866233203681e-07,
      "loss": 0.2695,
      "step": 539
    },
    {
      "epoch": 1.6167664670658684,
      "grad_norm": 0.8292764935570617,
      "learning_rate": 8.789251410400024e-07,
      "loss": 0.223,
      "step": 540
    },
    {
      "epoch": 1.6197604790419162,
      "grad_norm": 0.8910398864974575,
      "learning_rate": 8.65654808728259e-07,
      "loss": 0.2765,
      "step": 541
    },
    {
      "epoch": 1.622754491017964,
      "grad_norm": 1.2447041601300974,
      "learning_rate": 8.524759198984567e-07,
      "loss": 0.254,
      "step": 542
    },
    {
      "epoch": 1.625748502994012,
      "grad_norm": 1.0180965706579113,
      "learning_rate": 8.393887660413719e-07,
      "loss": 0.2857,
      "step": 543
    },
    {
      "epoch": 1.6287425149700598,
      "grad_norm": 1.255705210510179,
      "learning_rate": 8.263936366187825e-07,
      "loss": 0.3466,
      "step": 544
    },
    {
      "epoch": 1.6317365269461077,
      "grad_norm": 1.0150198301967028,
      "learning_rate": 8.134908190570723e-07,
      "loss": 0.3045,
      "step": 545
    },
    {
      "epoch": 1.6347305389221556,
      "grad_norm": 0.890330932939154,
      "learning_rate": 8.006805987408705e-07,
      "loss": 0.2097,
      "step": 546
    },
    {
      "epoch": 1.6377245508982035,
      "grad_norm": 1.0598977821495503,
      "learning_rate": 7.879632590067354e-07,
      "loss": 0.2945,
      "step": 547
    },
    {
      "epoch": 1.6407185628742516,
      "grad_norm": 1.1140121729115857,
      "learning_rate": 7.753390811368972e-07,
      "loss": 0.3282,
      "step": 548
    },
    {
      "epoch": 1.6437125748502994,
      "grad_norm": 1.0160729111471394,
      "learning_rate": 7.628083443530287e-07,
      "loss": 0.2808,
      "step": 549
    },
    {
      "epoch": 1.6467065868263473,
      "grad_norm": 1.1725275613210664,
      "learning_rate": 7.503713258100726e-07,
      "loss": 0.3203,
      "step": 550
    },
    {
      "epoch": 1.6497005988023952,
      "grad_norm": 0.9756846296384402,
      "learning_rate": 7.380283005901084e-07,
      "loss": 0.2892,
      "step": 551
    },
    {
      "epoch": 1.6526946107784433,
      "grad_norm": 0.9589031538582429,
      "learning_rate": 7.257795416962754e-07,
      "loss": 0.2606,
      "step": 552
    },
    {
      "epoch": 1.6556886227544911,
      "grad_norm": 1.0260127763890292,
      "learning_rate": 7.136253200467231e-07,
      "loss": 0.2848,
      "step": 553
    },
    {
      "epoch": 1.658682634730539,
      "grad_norm": 1.0638338266318648,
      "learning_rate": 7.015659044686307e-07,
      "loss": 0.2764,
      "step": 554
    },
    {
      "epoch": 1.6616766467065869,
      "grad_norm": 1.034756665824474,
      "learning_rate": 6.896015616922535e-07,
      "loss": 0.2862,
      "step": 555
    },
    {
      "epoch": 1.6646706586826348,
      "grad_norm": 0.9517593679522471,
      "learning_rate": 6.777325563450282e-07,
      "loss": 0.3227,
      "step": 556
    },
    {
      "epoch": 1.6676646706586826,
      "grad_norm": 1.0087476237195276,
      "learning_rate": 6.659591509457125e-07,
      "loss": 0.277,
      "step": 557
    },
    {
      "epoch": 1.6706586826347305,
      "grad_norm": 1.1419540564400483,
      "learning_rate": 6.542816058985896e-07,
      "loss": 0.2985,
      "step": 558
    },
    {
      "epoch": 1.6736526946107784,
      "grad_norm": 1.0948747684546611,
      "learning_rate": 6.427001794876974e-07,
      "loss": 0.2833,
      "step": 559
    },
    {
      "epoch": 1.6766467065868262,
      "grad_norm": 0.9360608941611567,
      "learning_rate": 6.312151278711237e-07,
      "loss": 0.3004,
      "step": 560
    },
    {
      "epoch": 1.6796407185628741,
      "grad_norm": 0.939868141128382,
      "learning_rate": 6.198267050753387e-07,
      "loss": 0.2807,
      "step": 561
    },
    {
      "epoch": 1.6826347305389222,
      "grad_norm": 1.0266935563707245,
      "learning_rate": 6.085351629895736e-07,
      "loss": 0.2759,
      "step": 562
    },
    {
      "epoch": 1.68562874251497,
      "grad_norm": 1.1380332770043855,
      "learning_rate": 5.973407513602514e-07,
      "loss": 0.2602,
      "step": 563
    },
    {
      "epoch": 1.688622754491018,
      "grad_norm": 1.0519438463503428,
      "learning_rate": 5.862437177854629e-07,
      "loss": 0.3035,
      "step": 564
    },
    {
      "epoch": 1.6916167664670658,
      "grad_norm": 1.2169465575295926,
      "learning_rate": 5.752443077094927e-07,
      "loss": 0.2527,
      "step": 565
    },
    {
      "epoch": 1.694610778443114,
      "grad_norm": 1.00460920180784,
      "learning_rate": 5.643427644173838e-07,
      "loss": 0.2883,
      "step": 566
    },
    {
      "epoch": 1.6976047904191618,
      "grad_norm": 0.9166236516976071,
      "learning_rate": 5.535393290295643e-07,
      "loss": 0.2648,
      "step": 567
    },
    {
      "epoch": 1.7005988023952097,
      "grad_norm": 1.0224218447371933,
      "learning_rate": 5.428342404965076e-07,
      "loss": 0.2532,
      "step": 568
    },
    {
      "epoch": 1.7035928143712575,
      "grad_norm": 1.0827482075251529,
      "learning_rate": 5.322277355934557e-07,
      "loss": 0.324,
      "step": 569
    },
    {
      "epoch": 1.7065868263473054,
      "grad_norm": 0.9714111449515691,
      "learning_rate": 5.217200489151714e-07,
      "loss": 0.285,
      "step": 570
    },
    {
      "epoch": 1.7095808383233533,
      "grad_norm": 0.9364115965660317,
      "learning_rate": 5.113114128707592e-07,
      "loss": 0.2353,
      "step": 571
    },
    {
      "epoch": 1.7125748502994012,
      "grad_norm": 1.0763449904071511,
      "learning_rate": 5.010020576785174e-07,
      "loss": 0.3051,
      "step": 572
    },
    {
      "epoch": 1.715568862275449,
      "grad_norm": 0.969300404999829,
      "learning_rate": 4.907922113608532e-07,
      "loss": 0.2383,
      "step": 573
    },
    {
      "epoch": 1.718562874251497,
      "grad_norm": 1.0519712300939217,
      "learning_rate": 4.806820997392325e-07,
      "loss": 0.2826,
      "step": 574
    },
    {
      "epoch": 1.7215568862275448,
      "grad_norm": 0.9558157603046407,
      "learning_rate": 4.7067194642919036e-07,
      "loss": 0.282,
      "step": 575
    },
    {
      "epoch": 1.7245508982035929,
      "grad_norm": 1.1085334878885595,
      "learning_rate": 4.607619728353818e-07,
      "loss": 0.2681,
      "step": 576
    },
    {
      "epoch": 1.7275449101796407,
      "grad_norm": 0.9267873461117612,
      "learning_rate": 4.50952398146689e-07,
      "loss": 0.2772,
      "step": 577
    },
    {
      "epoch": 1.7305389221556886,
      "grad_norm": 1.0012895885975834,
      "learning_rate": 4.4124343933136525e-07,
      "loss": 0.2745,
      "step": 578
    },
    {
      "epoch": 1.7335329341317365,
      "grad_norm": 0.9412302789794276,
      "learning_rate": 4.3163531113224466e-07,
      "loss": 0.2533,
      "step": 579
    },
    {
      "epoch": 1.7365269461077846,
      "grad_norm": 0.9914749596703888,
      "learning_rate": 4.221282260619891e-07,
      "loss": 0.2932,
      "step": 580
    },
    {
      "epoch": 1.7395209580838324,
      "grad_norm": 0.9273325569543667,
      "learning_rate": 4.127223943983849e-07,
      "loss": 0.2623,
      "step": 581
    },
    {
      "epoch": 1.7425149700598803,
      "grad_norm": 0.974899200106077,
      "learning_rate": 4.03418024179697e-07,
      "loss": 0.2599,
      "step": 582
    },
    {
      "epoch": 1.7455089820359282,
      "grad_norm": 0.9661779192860693,
      "learning_rate": 3.9421532120006544e-07,
      "loss": 0.2561,
      "step": 583
    },
    {
      "epoch": 1.748502994011976,
      "grad_norm": 0.9556330265930582,
      "learning_rate": 3.851144890049535e-07,
      "loss": 0.2704,
      "step": 584
    },
    {
      "epoch": 1.751497005988024,
      "grad_norm": 1.05867747298971,
      "learning_rate": 3.761157288866418e-07,
      "loss": 0.3349,
      "step": 585
    },
    {
      "epoch": 1.7544910179640718,
      "grad_norm": 1.0013296746121176,
      "learning_rate": 3.672192398797858e-07,
      "loss": 0.2754,
      "step": 586
    },
    {
      "epoch": 1.7574850299401197,
      "grad_norm": 1.0703705550045968,
      "learning_rate": 3.58425218757002e-07,
      "loss": 0.2727,
      "step": 587
    },
    {
      "epoch": 1.7604790419161676,
      "grad_norm": 1.0890885790400893,
      "learning_rate": 3.497338600245254e-07,
      "loss": 0.2705,
      "step": 588
    },
    {
      "epoch": 1.7634730538922154,
      "grad_norm": 0.9741753118003843,
      "learning_rate": 3.4114535591790233e-07,
      "loss": 0.2515,
      "step": 589
    },
    {
      "epoch": 1.7664670658682635,
      "grad_norm": 1.0944452295877847,
      "learning_rate": 3.326598963977395e-07,
      "loss": 0.2589,
      "step": 590
    },
    {
      "epoch": 1.7694610778443114,
      "grad_norm": 1.0275735990331376,
      "learning_rate": 3.242776691455013e-07,
      "loss": 0.2982,
      "step": 591
    },
    {
      "epoch": 1.7724550898203593,
      "grad_norm": 0.8420726703593355,
      "learning_rate": 3.159988595593616e-07,
      "loss": 0.2455,
      "step": 592
    },
    {
      "epoch": 1.7754491017964071,
      "grad_norm": 0.9593924117616909,
      "learning_rate": 3.078236507501015e-07,
      "loss": 0.2408,
      "step": 593
    },
    {
      "epoch": 1.7784431137724552,
      "grad_norm": 0.9947122425930526,
      "learning_rate": 2.9975222353705757e-07,
      "loss": 0.3006,
      "step": 594
    },
    {
      "epoch": 1.781437125748503,
      "grad_norm": 0.9962935158980532,
      "learning_rate": 2.917847564441256e-07,
      "loss": 0.3145,
      "step": 595
    },
    {
      "epoch": 1.784431137724551,
      "grad_norm": 1.0508672303220938,
      "learning_rate": 2.839214256958106e-07,
      "loss": 0.2811,
      "step": 596
    },
    {
      "epoch": 1.7874251497005988,
      "grad_norm": 1.0397839051667999,
      "learning_rate": 2.7616240521332884e-07,
      "loss": 0.2657,
      "step": 597
    },
    {
      "epoch": 1.7904191616766467,
      "grad_norm": 0.9300479476583533,
      "learning_rate": 2.6850786661076047e-07,
      "loss": 0.2821,
      "step": 598
    },
    {
      "epoch": 1.7934131736526946,
      "grad_norm": 0.9935404720291086,
      "learning_rate": 2.6095797919125533e-07,
      "loss": 0.2629,
      "step": 599
    },
    {
      "epoch": 1.7964071856287425,
      "grad_norm": 0.9704851413753994,
      "learning_rate": 2.5351290994328703e-07,
      "loss": 0.3017,
      "step": 600
    },
    {
      "epoch": 1.7964071856287425,
      "eval_loss": 0.36320582032203674,
      "eval_runtime": 2.076,
      "eval_samples_per_second": 26.011,
      "eval_steps_per_second": 6.744,
      "step": 600
    },
    {
      "epoch": 1.7994011976047903,
      "grad_norm": 0.923994696783988,
      "learning_rate": 2.4617282353696093e-07,
      "loss": 0.2525,
      "step": 601
    },
    {
      "epoch": 1.8023952095808382,
      "grad_norm": 1.1099311606300768,
      "learning_rate": 2.3893788232036807e-07,
      "loss": 0.3488,
      "step": 602
    },
    {
      "epoch": 1.805389221556886,
      "grad_norm": 0.9986226652634611,
      "learning_rate": 2.318082463160032e-07,
      "loss": 0.3128,
      "step": 603
    },
    {
      "epoch": 1.8083832335329342,
      "grad_norm": 0.8792823539479557,
      "learning_rate": 2.2478407321721295e-07,
      "loss": 0.266,
      "step": 604
    },
    {
      "epoch": 1.811377245508982,
      "grad_norm": 0.9246721910855287,
      "learning_rate": 2.1786551838471892e-07,
      "loss": 0.272,
      "step": 605
    },
    {
      "epoch": 1.81437125748503,
      "grad_norm": 1.1421491657235932,
      "learning_rate": 2.1105273484317402e-07,
      "loss": 0.2355,
      "step": 606
    },
    {
      "epoch": 1.8173652694610778,
      "grad_norm": 0.9715050475323485,
      "learning_rate": 2.043458732777831e-07,
      "loss": 0.2502,
      "step": 607
    },
    {
      "epoch": 1.8203592814371259,
      "grad_norm": 1.0994232131890003,
      "learning_rate": 1.9774508203096843e-07,
      "loss": 0.2886,
      "step": 608
    },
    {
      "epoch": 1.8233532934131738,
      "grad_norm": 0.93613158163084,
      "learning_rate": 1.9125050709908388e-07,
      "loss": 0.2457,
      "step": 609
    },
    {
      "epoch": 1.8263473053892216,
      "grad_norm": 1.0329293204932921,
      "learning_rate": 1.8486229212919482e-07,
      "loss": 0.2926,
      "step": 610
    },
    {
      "epoch": 1.8293413173652695,
      "grad_norm": 0.9341728455541087,
      "learning_rate": 1.7858057841589281e-07,
      "loss": 0.276,
      "step": 611
    },
    {
      "epoch": 1.8323353293413174,
      "grad_norm": 1.1485004134731334,
      "learning_rate": 1.7240550489817652e-07,
      "loss": 0.2992,
      "step": 612
    },
    {
      "epoch": 1.8353293413173652,
      "grad_norm": 1.182138384430407,
      "learning_rate": 1.66337208156373e-07,
      "loss": 0.2679,
      "step": 613
    },
    {
      "epoch": 1.8383233532934131,
      "grad_norm": 1.0385308885451339,
      "learning_rate": 1.6037582240912175e-07,
      "loss": 0.2922,
      "step": 614
    },
    {
      "epoch": 1.841317365269461,
      "grad_norm": 0.9874804114145909,
      "learning_rate": 1.5452147951040165e-07,
      "loss": 0.2849,
      "step": 615
    },
    {
      "epoch": 1.8443113772455089,
      "grad_norm": 1.1832916798888142,
      "learning_rate": 1.4877430894662037e-07,
      "loss": 0.3274,
      "step": 616
    },
    {
      "epoch": 1.8473053892215567,
      "grad_norm": 1.0373093637606001,
      "learning_rate": 1.4313443783374405e-07,
      "loss": 0.2607,
      "step": 617
    },
    {
      "epoch": 1.8502994011976048,
      "grad_norm": 0.9414788336196014,
      "learning_rate": 1.3760199091449045e-07,
      "loss": 0.2399,
      "step": 618
    },
    {
      "epoch": 1.8532934131736527,
      "grad_norm": 0.9212450702922258,
      "learning_rate": 1.3217709055556638e-07,
      "loss": 0.24,
      "step": 619
    },
    {
      "epoch": 1.8562874251497006,
      "grad_norm": 0.9843751539995954,
      "learning_rate": 1.268598567449647e-07,
      "loss": 0.2801,
      "step": 620
    },
    {
      "epoch": 1.8592814371257484,
      "grad_norm": 1.113476298013695,
      "learning_rate": 1.2165040708930763e-07,
      "loss": 0.2724,
      "step": 621
    },
    {
      "epoch": 1.8622754491017965,
      "grad_norm": 0.8848774254012737,
      "learning_rate": 1.1654885681124661e-07,
      "loss": 0.272,
      "step": 622
    },
    {
      "epoch": 1.8652694610778444,
      "grad_norm": 1.098906114034813,
      "learning_rate": 1.1155531874691372e-07,
      "loss": 0.3077,
      "step": 623
    },
    {
      "epoch": 1.8682634730538923,
      "grad_norm": 0.9979512468135505,
      "learning_rate": 1.0666990334342708e-07,
      "loss": 0.2507,
      "step": 624
    },
    {
      "epoch": 1.8712574850299402,
      "grad_norm": 0.981938985525399,
      "learning_rate": 1.0189271865644445e-07,
      "loss": 0.2806,
      "step": 625
    },
    {
      "epoch": 1.874251497005988,
      "grad_norm": 1.2887570934487664,
      "learning_rate": 9.722387034777847e-08,
      "loss": 0.3324,
      "step": 626
    },
    {
      "epoch": 1.877245508982036,
      "grad_norm": 1.06498109733751,
      "learning_rate": 9.266346168305518e-08,
      "loss": 0.3294,
      "step": 627
    },
    {
      "epoch": 1.8802395209580838,
      "grad_norm": 0.9243248560068268,
      "learning_rate": 8.821159352943142e-08,
      "loss": 0.271,
      "step": 628
    },
    {
      "epoch": 1.8832335329341316,
      "grad_norm": 1.0289876918720562,
      "learning_rate": 8.38683643533661e-08,
      "loss": 0.289,
      "step": 629
    },
    {
      "epoch": 1.8862275449101795,
      "grad_norm": 1.1057964102171312,
      "learning_rate": 7.963387021843683e-08,
      "loss": 0.2905,
      "step": 630
    },
    {
      "epoch": 1.8892215568862274,
      "grad_norm": 1.0479344818531047,
      "learning_rate": 7.550820478322285e-08,
      "loss": 0.3142,
      "step": 631
    },
    {
      "epoch": 1.8922155688622755,
      "grad_norm": 0.970293984457037,
      "learning_rate": 7.149145929922607e-08,
      "loss": 0.2432,
      "step": 632
    },
    {
      "epoch": 1.8952095808383234,
      "grad_norm": 1.0293625951669665,
      "learning_rate": 6.758372260885714e-08,
      "loss": 0.2966,
      "step": 633
    },
    {
      "epoch": 1.8982035928143712,
      "grad_norm": 1.1560159962780443,
      "learning_rate": 6.378508114346982e-08,
      "loss": 0.2782,
      "step": 634
    },
    {
      "epoch": 1.9011976047904193,
      "grad_norm": 1.1616630038529756,
      "learning_rate": 6.009561892144744e-08,
      "loss": 0.2624,
      "step": 635
    },
    {
      "epoch": 1.9041916167664672,
      "grad_norm": 1.0304623182758117,
      "learning_rate": 5.651541754634726e-08,
      "loss": 0.2541,
      "step": 636
    },
    {
      "epoch": 1.907185628742515,
      "grad_norm": 1.0692775431317154,
      "learning_rate": 5.304455620509297e-08,
      "loss": 0.2949,
      "step": 637
    },
    {
      "epoch": 1.910179640718563,
      "grad_norm": 1.0270535615556404,
      "learning_rate": 4.968311166622553e-08,
      "loss": 0.2803,
      "step": 638
    },
    {
      "epoch": 1.9131736526946108,
      "grad_norm": 1.2246168065187621,
      "learning_rate": 4.643115827820399e-08,
      "loss": 0.2532,
      "step": 639
    },
    {
      "epoch": 1.9161676646706587,
      "grad_norm": 1.0974822810039906,
      "learning_rate": 4.328876796776071e-08,
      "loss": 0.3181,
      "step": 640
    },
    {
      "epoch": 1.9191616766467066,
      "grad_norm": 1.098535203964859,
      "learning_rate": 4.0256010238310936e-08,
      "loss": 0.2667,
      "step": 641
    },
    {
      "epoch": 1.9221556886227544,
      "grad_norm": 0.9786972084257104,
      "learning_rate": 3.733295216841626e-08,
      "loss": 0.2984,
      "step": 642
    },
    {
      "epoch": 1.9251497005988023,
      "grad_norm": 1.0311453074080557,
      "learning_rate": 3.451965841029914e-08,
      "loss": 0.2787,
      "step": 643
    },
    {
      "epoch": 1.9281437125748502,
      "grad_norm": 0.9231418615117579,
      "learning_rate": 3.181619118841517e-08,
      "loss": 0.3238,
      "step": 644
    },
    {
      "epoch": 1.931137724550898,
      "grad_norm": 1.1750677511871903,
      "learning_rate": 2.9222610298074717e-08,
      "loss": 0.2647,
      "step": 645
    },
    {
      "epoch": 1.9341317365269461,
      "grad_norm": 1.0120512030231554,
      "learning_rate": 2.673897310412288e-08,
      "loss": 0.309,
      "step": 646
    },
    {
      "epoch": 1.937125748502994,
      "grad_norm": 0.8817355352707243,
      "learning_rate": 2.4365334539667717e-08,
      "loss": 0.2438,
      "step": 647
    },
    {
      "epoch": 1.9401197604790419,
      "grad_norm": 1.0519606006081759,
      "learning_rate": 2.210174710486679e-08,
      "loss": 0.2717,
      "step": 648
    },
    {
      "epoch": 1.94311377245509,
      "grad_norm": 0.9806412476513693,
      "learning_rate": 1.99482608657664e-08,
      "loss": 0.2738,
      "step": 649
    },
    {
      "epoch": 1.9461077844311379,
      "grad_norm": 1.0726832807865938,
      "learning_rate": 1.7904923453193056e-08,
      "loss": 0.3137,
      "step": 650
    },
    {
      "epoch": 1.9491017964071857,
      "grad_norm": 0.9962822284300865,
      "learning_rate": 1.5971780061701524e-08,
      "loss": 0.2683,
      "step": 651
    },
    {
      "epoch": 1.9520958083832336,
      "grad_norm": 1.0572087758886517,
      "learning_rate": 1.4148873448573408e-08,
      "loss": 0.2509,
      "step": 652
    },
    {
      "epoch": 1.9550898203592815,
      "grad_norm": 0.9414554056307058,
      "learning_rate": 1.2436243932872349e-08,
      "loss": 0.2931,
      "step": 653
    },
    {
      "epoch": 1.9580838323353293,
      "grad_norm": 1.0242323388942745,
      "learning_rate": 1.0833929394552523e-08,
      "loss": 0.2475,
      "step": 654
    },
    {
      "epoch": 1.9610778443113772,
      "grad_norm": 0.9469512572922653,
      "learning_rate": 9.341965273621522e-09,
      "loss": 0.2922,
      "step": 655
    },
    {
      "epoch": 1.964071856287425,
      "grad_norm": 1.0307882551096461,
      "learning_rate": 7.96038456935322e-09,
      "loss": 0.2969,
      "step": 656
    },
    {
      "epoch": 1.967065868263473,
      "grad_norm": 1.0254156089121207,
      "learning_rate": 6.6892178395611125e-09,
      "loss": 0.3197,
      "step": 657
    },
    {
      "epoch": 1.9700598802395208,
      "grad_norm": 1.1992555938837843,
      "learning_rate": 5.528493199922769e-09,
      "loss": 0.3176,
      "step": 658
    },
    {
      "epoch": 1.9730538922155687,
      "grad_norm": 1.0192367851852362,
      "learning_rate": 4.478236323355312e-09,
      "loss": 0.2554,
      "step": 659
    },
    {
      "epoch": 1.9760479041916168,
      "grad_norm": 1.1010374506572544,
      "learning_rate": 3.538470439448105e-09,
      "loss": 0.342,
      "step": 660
    },
    {
      "epoch": 1.9790419161676647,
      "grad_norm": 1.0043897021209338,
      "learning_rate": 2.709216333952602e-09,
      "loss": 0.2927,
      "step": 661
    },
    {
      "epoch": 1.9820359281437125,
      "grad_norm": 0.9301795131335764,
      "learning_rate": 1.9904923483171632e-09,
      "loss": 0.2454,
      "step": 662
    },
    {
      "epoch": 1.9850299401197606,
      "grad_norm": 1.0714797530134446,
      "learning_rate": 1.3823143792851545e-09,
      "loss": 0.334,
      "step": 663
    },
    {
      "epoch": 1.9880239520958085,
      "grad_norm": 1.1049420069328866,
      "learning_rate": 8.846958785418969e-10,
      "loss": 0.3257,
      "step": 664
    },
    {
      "epoch": 1.9910179640718564,
      "grad_norm": 1.0252696177799858,
      "learning_rate": 4.97647852417682e-10,
      "loss": 0.2529,
      "step": 665
    },
    {
      "epoch": 1.9940119760479043,
      "grad_norm": 0.998494728565151,
      "learning_rate": 2.2117886164407797e-10,
      "loss": 0.2675,
      "step": 666
    },
    {
      "epoch": 1.9970059880239521,
      "grad_norm": 0.8885624060187792,
      "learning_rate": 5.529502116519148e-11,
      "loss": 0.2458,
      "step": 667
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9847232801633843,
      "learning_rate": 0.0,
      "loss": 0.1923,
      "step": 668
    },
    {
      "epoch": 2.0,
      "step": 668,
      "total_flos": 21199392964608.0,
      "train_loss": 0.3376133635327845,
      "train_runtime": 1178.6141,
      "train_samples_per_second": 9.043,
      "train_steps_per_second": 0.567
    }
  ],
  "logging_steps": 1,
  "max_steps": 668,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 21199392964608.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}