{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 5000.0,
  "global_step": 25371,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.023649048125812935,
      "grad_norm": 0.5222728848457336,
      "learning_rate": 2.4875e-05,
      "loss": 0.2603,
      "step": 200
    },
    {
      "epoch": 0.04729809625162587,
      "grad_norm": 0.5934835076332092,
      "learning_rate": 4.9875000000000006e-05,
      "loss": 0.2578,
      "step": 400
    },
    {
      "epoch": 0.0709471443774388,
      "grad_norm": 0.5335094332695007,
      "learning_rate": 4.960153778382924e-05,
      "loss": 0.2499,
      "step": 600
    },
    {
      "epoch": 0.09459619250325174,
      "grad_norm": 0.5203099846839905,
      "learning_rate": 4.920107324496416e-05,
      "loss": 0.252,
      "step": 800
    },
    {
      "epoch": 0.11824524062906468,
      "grad_norm": 0.4816642701625824,
      "learning_rate": 4.8800608706099076e-05,
      "loss": 0.2519,
      "step": 1000
    },
    {
      "epoch": 0.1418942887548776,
      "grad_norm": 0.5423477292060852,
      "learning_rate": 4.8400144167233995e-05,
      "loss": 0.2533,
      "step": 1200
    },
    {
      "epoch": 0.16554333688069056,
      "grad_norm": 0.810859203338623,
      "learning_rate": 4.799967962836891e-05,
      "loss": 0.2502,
      "step": 1400
    },
    {
      "epoch": 0.18919238500650348,
      "grad_norm": 0.6290557384490967,
      "learning_rate": 4.759921508950383e-05,
      "loss": 0.2533,
      "step": 1600
    },
    {
      "epoch": 0.21284143313231643,
      "grad_norm": 0.6436994671821594,
      "learning_rate": 4.719875055063874e-05,
      "loss": 0.2526,
      "step": 1800
    },
    {
      "epoch": 0.23649048125812935,
      "grad_norm": 0.48108312487602234,
      "learning_rate": 4.679828601177366e-05,
      "loss": 0.2514,
      "step": 2000
    },
    {
      "epoch": 0.2601395293839423,
      "grad_norm": 0.5231059193611145,
      "learning_rate": 4.639782147290857e-05,
      "loss": 0.2469,
      "step": 2200
    },
    {
      "epoch": 0.2837885775097552,
      "grad_norm": 0.5582587718963623,
      "learning_rate": 4.599735693404349e-05,
      "loss": 0.2462,
      "step": 2400
    },
    {
      "epoch": 0.30743762563556815,
      "grad_norm": 0.5926572680473328,
      "learning_rate": 4.5596892395178406e-05,
      "loss": 0.2498,
      "step": 2600
    },
    {
      "epoch": 0.3310866737613811,
      "grad_norm": 0.586698591709137,
      "learning_rate": 4.5196427856313325e-05,
      "loss": 0.255,
      "step": 2800
    },
    {
      "epoch": 0.35473572188719404,
      "grad_norm": 0.7075465321540833,
      "learning_rate": 4.4795963317448245e-05,
      "loss": 0.256,
      "step": 3000
    },
    {
      "epoch": 0.37838477001300697,
      "grad_norm": 0.573063313961029,
      "learning_rate": 4.4395498778583164e-05,
      "loss": 0.2482,
      "step": 3200
    },
    {
      "epoch": 0.4020338181388199,
      "grad_norm": 0.5777062773704529,
      "learning_rate": 4.399503423971808e-05,
      "loss": 0.2481,
      "step": 3400
    },
    {
      "epoch": 0.42568286626463286,
      "grad_norm": 0.47794800996780396,
      "learning_rate": 4.3594569700853e-05,
      "loss": 0.2456,
      "step": 3600
    },
    {
      "epoch": 0.4493319143904458,
      "grad_norm": 0.5826495289802551,
      "learning_rate": 4.319410516198791e-05,
      "loss": 0.2441,
      "step": 3800
    },
    {
      "epoch": 0.4729809625162587,
      "grad_norm": 0.6569980978965759,
      "learning_rate": 4.279364062312283e-05,
      "loss": 0.2472,
      "step": 4000
    },
    {
      "epoch": 0.4966300106420717,
      "grad_norm": 0.6123746633529663,
      "learning_rate": 4.239317608425774e-05,
      "loss": 0.2476,
      "step": 4200
    },
    {
      "epoch": 0.5202790587678846,
      "grad_norm": 0.5969842076301575,
      "learning_rate": 4.199271154539266e-05,
      "loss": 0.2553,
      "step": 4400
    },
    {
      "epoch": 0.5439281068936975,
      "grad_norm": 0.480925589799881,
      "learning_rate": 4.1592247006527575e-05,
      "loss": 0.245,
      "step": 4600
    },
    {
      "epoch": 0.5675771550195104,
      "grad_norm": 0.4748068153858185,
      "learning_rate": 4.1191782467662494e-05,
      "loss": 0.248,
      "step": 4800
    },
    {
      "epoch": 0.5912262031453234,
      "grad_norm": 0.6042199730873108,
      "learning_rate": 4.079131792879741e-05,
      "loss": 0.2435,
      "step": 5000
    },
    {
      "epoch": 0.6148752512711363,
      "grad_norm": 0.724879264831543,
      "learning_rate": 4.039085338993233e-05,
      "loss": 0.2481,
      "step": 5200
    },
    {
      "epoch": 0.6385242993969493,
      "grad_norm": 0.6051663160324097,
      "learning_rate": 3.999038885106724e-05,
      "loss": 0.2437,
      "step": 5400
    },
    {
      "epoch": 0.6621733475227622,
      "grad_norm": 0.5058975219726562,
      "learning_rate": 3.958992431220216e-05,
      "loss": 0.2405,
      "step": 5600
    },
    {
      "epoch": 0.6858223956485752,
      "grad_norm": 0.7219721674919128,
      "learning_rate": 3.918945977333707e-05,
      "loss": 0.2491,
      "step": 5800
    },
    {
      "epoch": 0.7094714437743881,
      "grad_norm": 0.505952775478363,
      "learning_rate": 3.878899523447199e-05,
      "loss": 0.2479,
      "step": 6000
    },
    {
      "epoch": 0.733120491900201,
      "grad_norm": 0.6167413592338562,
      "learning_rate": 3.8388530695606904e-05,
      "loss": 0.2482,
      "step": 6200
    },
    {
      "epoch": 0.7567695400260139,
      "grad_norm": 0.6561195850372314,
      "learning_rate": 3.7988066156741824e-05,
      "loss": 0.2446,
      "step": 6400
    },
    {
      "epoch": 0.7804185881518269,
      "grad_norm": 0.5154906511306763,
      "learning_rate": 3.758760161787674e-05,
      "loss": 0.2419,
      "step": 6600
    },
    {
      "epoch": 0.8040676362776398,
      "grad_norm": 0.5314207077026367,
      "learning_rate": 3.7187137079011656e-05,
      "loss": 0.2486,
      "step": 6800
    },
    {
      "epoch": 0.8277166844034528,
      "grad_norm": 0.5757410526275635,
      "learning_rate": 3.678667254014657e-05,
      "loss": 0.2446,
      "step": 7000
    },
    {
      "epoch": 0.8513657325292657,
      "grad_norm": 0.6532585620880127,
      "learning_rate": 3.638620800128149e-05,
      "loss": 0.249,
      "step": 7200
    },
    {
      "epoch": 0.8750147806550787,
      "grad_norm": 0.5678747892379761,
      "learning_rate": 3.59857434624164e-05,
      "loss": 0.2468,
      "step": 7400
    },
    {
      "epoch": 0.8986638287808916,
      "grad_norm": 0.9313775300979614,
      "learning_rate": 3.558527892355132e-05,
      "loss": 0.246,
      "step": 7600
    },
    {
      "epoch": 0.9223128769067045,
      "grad_norm": 0.5327833294868469,
      "learning_rate": 3.518481438468624e-05,
      "loss": 0.2451,
      "step": 7800
    },
    {
      "epoch": 0.9459619250325174,
      "grad_norm": 0.6526719331741333,
      "learning_rate": 3.4784349845821154e-05,
      "loss": 0.2441,
      "step": 8000
    },
    {
      "epoch": 0.9696109731583303,
      "grad_norm": 0.49486011266708374,
      "learning_rate": 3.4383885306956073e-05,
      "loss": 0.2446,
      "step": 8200
    },
    {
      "epoch": 0.9932600212841434,
      "grad_norm": 0.614295482635498,
      "learning_rate": 3.3983420768090986e-05,
      "loss": 0.2376,
      "step": 8400
    },
    {
      "epoch": 1.0169090694099563,
      "grad_norm": 0.5915587544441223,
      "learning_rate": 3.3582956229225906e-05,
      "loss": 0.2319,
      "step": 8600
    },
    {
      "epoch": 1.0405581175357692,
      "grad_norm": 0.3928125202655792,
      "learning_rate": 3.318249169036082e-05,
      "loss": 0.2275,
      "step": 8800
    },
    {
      "epoch": 1.0642071656615821,
      "grad_norm": 0.4684751629829407,
      "learning_rate": 3.278202715149574e-05,
      "loss": 0.2331,
      "step": 9000
    },
    {
      "epoch": 1.087856213787395,
      "grad_norm": 0.5276994109153748,
      "learning_rate": 3.238156261263065e-05,
      "loss": 0.2319,
      "step": 9200
    },
    {
      "epoch": 1.111505261913208,
      "grad_norm": 0.788284957408905,
      "learning_rate": 3.198109807376557e-05,
      "loss": 0.2312,
      "step": 9400
    },
    {
      "epoch": 1.135154310039021,
      "grad_norm": 0.5523064732551575,
      "learning_rate": 3.1580633534900484e-05,
      "loss": 0.2277,
      "step": 9600
    },
    {
      "epoch": 1.1588033581648338,
      "grad_norm": 0.5746716260910034,
      "learning_rate": 3.11801689960354e-05,
      "loss": 0.2302,
      "step": 9800
    },
    {
      "epoch": 1.1824524062906467,
      "grad_norm": 0.4701022207736969,
      "learning_rate": 3.0779704457170316e-05,
      "loss": 0.2317,
      "step": 10000
    },
    {
      "epoch": 1.2061014544164597,
      "grad_norm": 0.39494338631629944,
      "learning_rate": 3.0379239918305236e-05,
      "loss": 0.2328,
      "step": 10200
    },
    {
      "epoch": 1.2297505025422726,
      "grad_norm": 0.6249374151229858,
      "learning_rate": 2.997877537944015e-05,
      "loss": 0.2394,
      "step": 10400
    },
    {
      "epoch": 1.2533995506680857,
      "grad_norm": 0.5556264519691467,
      "learning_rate": 2.9578310840575068e-05,
      "loss": 0.2333,
      "step": 10600
    },
    {
      "epoch": 1.2770485987938986,
      "grad_norm": 0.6378809809684753,
      "learning_rate": 2.917784630170998e-05,
      "loss": 0.2325,
      "step": 10800
    },
    {
      "epoch": 1.3006976469197116,
      "grad_norm": 0.5375252962112427,
      "learning_rate": 2.87773817628449e-05,
      "loss": 0.2348,
      "step": 11000
    },
    {
      "epoch": 1.3243466950455245,
      "grad_norm": 0.5387800931930542,
      "learning_rate": 2.8376917223979817e-05,
      "loss": 0.2326,
      "step": 11200
    },
    {
      "epoch": 1.3479957431713374,
      "grad_norm": 0.6196502447128296,
      "learning_rate": 2.7976452685114733e-05,
      "loss": 0.2307,
      "step": 11400
    },
    {
      "epoch": 1.3716447912971503,
      "grad_norm": 0.6024973392486572,
      "learning_rate": 2.757598814624965e-05,
      "loss": 0.2311,
      "step": 11600
    },
    {
      "epoch": 1.3952938394229633,
      "grad_norm": 0.6320472359657288,
      "learning_rate": 2.717552360738457e-05,
      "loss": 0.2382,
      "step": 11800
    },
    {
      "epoch": 1.4189428875487762,
      "grad_norm": 0.5864059329032898,
      "learning_rate": 2.6775059068519482e-05,
      "loss": 0.23,
      "step": 12000
    },
    {
      "epoch": 1.442591935674589,
      "grad_norm": 0.46693137288093567,
      "learning_rate": 2.63745945296544e-05,
      "loss": 0.2326,
      "step": 12200
    },
    {
      "epoch": 1.466240983800402,
      "grad_norm": 0.7086625695228577,
      "learning_rate": 2.5974129990789314e-05,
      "loss": 0.2334,
      "step": 12400
    },
    {
      "epoch": 1.489890031926215,
      "grad_norm": 0.6105942130088806,
      "learning_rate": 2.5573665451924234e-05,
      "loss": 0.2296,
      "step": 12600
    },
    {
      "epoch": 1.5135390800520279,
      "grad_norm": 0.5020413398742676,
      "learning_rate": 2.5173200913059147e-05,
      "loss": 0.2336,
      "step": 12800
    },
    {
      "epoch": 1.5371881281778408,
      "grad_norm": 0.5275157690048218,
      "learning_rate": 2.4772736374194066e-05,
      "loss": 0.235,
      "step": 13000
    },
    {
      "epoch": 1.5608371763036537,
      "grad_norm": 0.38569340109825134,
      "learning_rate": 2.4372271835328983e-05,
      "loss": 0.2285,
      "step": 13200
    },
    {
      "epoch": 1.5844862244294666,
      "grad_norm": 0.5194272994995117,
      "learning_rate": 2.39718072964639e-05,
      "loss": 0.2324,
      "step": 13400
    },
    {
      "epoch": 1.6081352725552795,
      "grad_norm": 0.5696848630905151,
      "learning_rate": 2.3571342757598815e-05,
      "loss": 0.2314,
      "step": 13600
    },
    {
      "epoch": 1.6317843206810925,
      "grad_norm": 0.41934022307395935,
      "learning_rate": 2.317087821873373e-05,
      "loss": 0.2322,
      "step": 13800
    },
    {
      "epoch": 1.6554333688069054,
      "grad_norm": 0.485202431678772,
      "learning_rate": 2.2770413679868647e-05,
      "loss": 0.2344,
      "step": 14000
    },
    {
      "epoch": 1.6790824169327183,
      "grad_norm": 0.45751115679740906,
      "learning_rate": 2.2369949141003567e-05,
      "loss": 0.2322,
      "step": 14200
    },
    {
      "epoch": 1.7027314650585312,
      "grad_norm": 0.5662369728088379,
      "learning_rate": 2.1969484602138483e-05,
      "loss": 0.2327,
      "step": 14400
    },
    {
      "epoch": 1.7263805131843444,
      "grad_norm": 0.6048633456230164,
      "learning_rate": 2.15690200632734e-05,
      "loss": 0.2309,
      "step": 14600
    },
    {
      "epoch": 1.7500295613101573,
      "grad_norm": 0.43209484219551086,
      "learning_rate": 2.1168555524408316e-05,
      "loss": 0.2286,
      "step": 14800
    },
    {
      "epoch": 1.7736786094359702,
      "grad_norm": 0.49313920736312866,
      "learning_rate": 2.0768090985543232e-05,
      "loss": 0.2286,
      "step": 15000
    },
    {
      "epoch": 1.7973276575617831,
      "grad_norm": 0.6878641247749329,
      "learning_rate": 2.0367626446678148e-05,
      "loss": 0.2333,
      "step": 15200
    },
    {
      "epoch": 1.820976705687596,
      "grad_norm": 0.4174480736255646,
      "learning_rate": 1.9967161907813064e-05,
      "loss": 0.2312,
      "step": 15400
    },
    {
      "epoch": 1.844625753813409,
      "grad_norm": 0.48459678888320923,
      "learning_rate": 1.956669736894798e-05,
      "loss": 0.2321,
      "step": 15600
    },
    {
      "epoch": 1.868274801939222,
      "grad_norm": 0.8106375932693481,
      "learning_rate": 1.9166232830082897e-05,
      "loss": 0.2316,
      "step": 15800
    },
    {
      "epoch": 1.8919238500650348,
      "grad_norm": 0.47767356038093567,
      "learning_rate": 1.8765768291217813e-05,
      "loss": 0.23,
      "step": 16000
    },
    {
      "epoch": 1.915572898190848,
      "grad_norm": 0.4701699912548065,
      "learning_rate": 1.836530375235273e-05,
      "loss": 0.2322,
      "step": 16200
    },
    {
      "epoch": 1.939221946316661,
      "grad_norm": 0.4411082863807678,
      "learning_rate": 1.7964839213487646e-05,
      "loss": 0.235,
      "step": 16400
    },
    {
      "epoch": 1.9628709944424738,
      "grad_norm": 0.7355507612228394,
      "learning_rate": 1.7564374674622565e-05,
      "loss": 0.2285,
      "step": 16600
    },
    {
      "epoch": 1.9865200425682867,
      "grad_norm": 0.47375932335853577,
      "learning_rate": 1.716391013575748e-05,
      "loss": 0.2323,
      "step": 16800
    },
    {
      "epoch": 2.0101690906940997,
      "grad_norm": 0.47945132851600647,
      "learning_rate": 1.6763445596892398e-05,
      "loss": 0.2276,
      "step": 17000
    },
    {
      "epoch": 2.0338181388199126,
      "grad_norm": 0.5424835085868835,
      "learning_rate": 1.6362981058027314e-05,
      "loss": 0.2254,
      "step": 17200
    },
    {
      "epoch": 2.0574671869457255,
      "grad_norm": 0.7722477912902832,
      "learning_rate": 1.596251651916223e-05,
      "loss": 0.2238,
      "step": 17400
    },
    {
      "epoch": 2.0811162350715384,
      "grad_norm": 0.47599107027053833,
      "learning_rate": 1.5562051980297146e-05,
      "loss": 0.22,
      "step": 17600
    },
    {
      "epoch": 2.1047652831973513,
      "grad_norm": 0.640536367893219,
      "learning_rate": 1.5161587441432063e-05,
      "loss": 0.2201,
      "step": 17800
    },
    {
      "epoch": 2.1284143313231643,
      "grad_norm": 0.6262871026992798,
      "learning_rate": 1.4761122902566979e-05,
      "loss": 0.2229,
      "step": 18000
    },
    {
      "epoch": 2.152063379448977,
      "grad_norm": 0.39407041668891907,
      "learning_rate": 1.4360658363701895e-05,
      "loss": 0.2246,
      "step": 18200
    },
    {
      "epoch": 2.17571242757479,
      "grad_norm": 0.5400912761688232,
      "learning_rate": 1.3960193824836811e-05,
      "loss": 0.2179,
      "step": 18400
    },
    {
      "epoch": 2.199361475700603,
      "grad_norm": 0.4764401912689209,
      "learning_rate": 1.3559729285971729e-05,
      "loss": 0.2217,
      "step": 18600
    },
    {
      "epoch": 2.223010523826416,
      "grad_norm": 0.5163056254386902,
      "learning_rate": 1.3159264747106645e-05,
      "loss": 0.2224,
      "step": 18800
    },
    {
      "epoch": 2.246659571952229,
      "grad_norm": 0.4986878037452698,
      "learning_rate": 1.2758800208241562e-05,
      "loss": 0.2203,
      "step": 19000
    },
    {
      "epoch": 2.270308620078042,
      "grad_norm": 0.7525689005851746,
      "learning_rate": 1.2358335669376478e-05,
      "loss": 0.2241,
      "step": 19200
    },
    {
      "epoch": 2.2939576682038547,
      "grad_norm": 0.4118233323097229,
      "learning_rate": 1.1957871130511394e-05,
      "loss": 0.2205,
      "step": 19400
    },
    {
      "epoch": 2.3176067163296676,
      "grad_norm": 0.49172499775886536,
      "learning_rate": 1.155740659164631e-05,
      "loss": 0.2212,
      "step": 19600
    },
    {
      "epoch": 2.3412557644554806,
      "grad_norm": 0.5130913853645325,
      "learning_rate": 1.1156942052781228e-05,
      "loss": 0.226,
      "step": 19800
    },
    {
      "epoch": 2.3649048125812935,
      "grad_norm": 0.3800196349620819,
      "learning_rate": 1.0756477513916144e-05,
      "loss": 0.2288,
      "step": 20000
    },
    {
      "epoch": 2.3885538607071064,
      "grad_norm": 0.4435960054397583,
      "learning_rate": 1.035601297505106e-05,
      "loss": 0.2221,
      "step": 20200
    },
    {
      "epoch": 2.4122029088329193,
      "grad_norm": 0.6992290616035461,
      "learning_rate": 9.955548436185977e-06,
      "loss": 0.2185,
      "step": 20400
    },
    {
      "epoch": 2.4358519569587322,
      "grad_norm": 0.6684884428977966,
      "learning_rate": 9.555083897320893e-06,
      "loss": 0.2139,
      "step": 20600
    },
    {
      "epoch": 2.459501005084545,
      "grad_norm": 0.45130160450935364,
      "learning_rate": 9.15461935845581e-06,
      "loss": 0.2238,
      "step": 20800
    },
    {
      "epoch": 2.483150053210358,
      "grad_norm": 0.5487006902694702,
      "learning_rate": 8.754154819590726e-06,
      "loss": 0.2261,
      "step": 21000
    },
    {
      "epoch": 2.5067991013361715,
      "grad_norm": 0.508743166923523,
      "learning_rate": 8.353690280725642e-06,
      "loss": 0.2284,
      "step": 21200
    },
    {
      "epoch": 2.530448149461984,
      "grad_norm": 0.48811525106430054,
      "learning_rate": 7.953225741860558e-06,
      "loss": 0.2235,
      "step": 21400
    },
    {
      "epoch": 2.5540971975877973,
      "grad_norm": 0.45879384875297546,
      "learning_rate": 7.552761202995474e-06,
      "loss": 0.2209,
      "step": 21600
    },
    {
      "epoch": 2.5777462457136098,
      "grad_norm": 0.5419099926948547,
      "learning_rate": 7.152296664130391e-06,
      "loss": 0.2239,
      "step": 21800
    },
    {
      "epoch": 2.601395293839423,
      "grad_norm": 0.5520743727684021,
      "learning_rate": 6.7518321252653075e-06,
      "loss": 0.2246,
      "step": 22000
    },
    {
      "epoch": 2.625044341965236,
      "grad_norm": 0.6102694272994995,
      "learning_rate": 6.351367586400224e-06,
      "loss": 0.2192,
      "step": 22200
    },
    {
      "epoch": 2.648693390091049,
      "grad_norm": 0.5299508571624756,
      "learning_rate": 5.950903047535141e-06,
      "loss": 0.2213,
      "step": 22400
    },
    {
      "epoch": 2.672342438216862,
      "grad_norm": 0.5461446642875671,
      "learning_rate": 5.550438508670058e-06,
      "loss": 0.2251,
      "step": 22600
    },
    {
      "epoch": 2.695991486342675,
      "grad_norm": 0.6796964406967163,
      "learning_rate": 5.149973969804974e-06,
      "loss": 0.2227,
      "step": 22800
    },
    {
      "epoch": 2.7196405344684877,
      "grad_norm": 0.5118963122367859,
      "learning_rate": 4.74950943093989e-06,
      "loss": 0.2201,
      "step": 23000
    },
    {
      "epoch": 2.7432895825943007,
      "grad_norm": 0.7696477174758911,
      "learning_rate": 4.349044892074807e-06,
      "loss": 0.2261,
      "step": 23200
    },
    {
      "epoch": 2.7669386307201136,
      "grad_norm": 0.5420796871185303,
      "learning_rate": 3.948580353209724e-06,
      "loss": 0.2195,
      "step": 23400
    },
    {
      "epoch": 2.7905876788459265,
      "grad_norm": 0.5679683089256287,
      "learning_rate": 3.5481158143446403e-06,
      "loss": 0.2233,
      "step": 23600
    },
    {
      "epoch": 2.8142367269717394,
      "grad_norm": 0.51070636510849,
      "learning_rate": 3.1476512754795565e-06,
      "loss": 0.2236,
      "step": 23800
    },
    {
      "epoch": 2.8378857750975524,
      "grad_norm": 0.4721522033214569,
      "learning_rate": 2.7471867366144727e-06,
      "loss": 0.2204,
      "step": 24000
    },
    {
      "epoch": 2.8615348232233653,
      "grad_norm": 0.4983059763908386,
      "learning_rate": 2.3467221977493894e-06,
      "loss": 0.227,
      "step": 24200
    },
    {
      "epoch": 2.885183871349178,
      "grad_norm": 0.49809587001800537,
      "learning_rate": 1.9462576588843056e-06,
      "loss": 0.2275,
      "step": 24400
    },
    {
      "epoch": 2.908832919474991,
      "grad_norm": 0.5711411833763123,
      "learning_rate": 1.5457931200192225e-06,
      "loss": 0.2248,
      "step": 24600
    },
    {
      "epoch": 2.932481967600804,
      "grad_norm": 0.4639867842197418,
      "learning_rate": 1.145328581154139e-06,
      "loss": 0.2218,
      "step": 24800
    },
    {
      "epoch": 2.956131015726617,
      "grad_norm": 0.46570587158203125,
      "learning_rate": 7.448640422890554e-07,
      "loss": 0.2274,
      "step": 25000
    },
    {
      "epoch": 2.97978006385243,
      "grad_norm": 0.4548007845878601,
      "learning_rate": 3.443995034239718e-07,
      "loss": 0.2257,
      "step": 25200
    },
    {
      "epoch": 3.0,
      "step": 25371,
      "total_flos": 2.101904731490089e+17,
      "train_loss": 0.23435649501197156,
      "train_runtime": 11527.9839,
      "train_samples_per_second": 35.212,
      "train_steps_per_second": 2.201
    }
  ],
  "logging_steps": 200,
  "max_steps": 25371,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 5000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.101904731490089e+17,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}