{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 51504,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.029123951537744643,
      "grad_norm": 6.774685382843018,
      "learning_rate": 4.951557160608885e-05,
      "loss": 0.8291,
      "step": 500
    },
    {
      "epoch": 0.058247903075489285,
      "grad_norm": 6.7878875732421875,
      "learning_rate": 4.9030172413793105e-05,
      "loss": 0.6161,
      "step": 1000
    },
    {
      "epoch": 0.08737185461323392,
      "grad_norm": 4.789400577545166,
      "learning_rate": 4.854477322149736e-05,
      "loss": 0.5785,
      "step": 1500
    },
    {
      "epoch": 0.11649580615097857,
      "grad_norm": 8.038426399230957,
      "learning_rate": 4.805937402920162e-05,
      "loss": 0.546,
      "step": 2000
    },
    {
      "epoch": 0.1456197576887232,
      "grad_norm": 5.627825736999512,
      "learning_rate": 4.757397483690587e-05,
      "loss": 0.5363,
      "step": 2500
    },
    {
      "epoch": 0.17474370922646784,
      "grad_norm": 4.209435939788818,
      "learning_rate": 4.7088575644610125e-05,
      "loss": 0.5225,
      "step": 3000
    },
    {
      "epoch": 0.20386766076421248,
      "grad_norm": 6.513331890106201,
      "learning_rate": 4.660317645231438e-05,
      "loss": 0.5104,
      "step": 3500
    },
    {
      "epoch": 0.23299161230195714,
      "grad_norm": 8.145315170288086,
      "learning_rate": 4.611777726001864e-05,
      "loss": 0.4894,
      "step": 4000
    },
    {
      "epoch": 0.26211556383970175,
      "grad_norm": 5.720129489898682,
      "learning_rate": 4.56323780677229e-05,
      "loss": 0.483,
      "step": 4500
    },
    {
      "epoch": 0.2912395153774464,
      "grad_norm": 4.3693718910217285,
      "learning_rate": 4.514697887542715e-05,
      "loss": 0.4747,
      "step": 5000
    },
    {
      "epoch": 0.3203634669151911,
      "grad_norm": 5.498645305633545,
      "learning_rate": 4.466157968313141e-05,
      "loss": 0.4686,
      "step": 5500
    },
    {
      "epoch": 0.3494874184529357,
      "grad_norm": 4.044460296630859,
      "learning_rate": 4.417618049083567e-05,
      "loss": 0.4679,
      "step": 6000
    },
    {
      "epoch": 0.37861136999068035,
      "grad_norm": 6.850383758544922,
      "learning_rate": 4.369078129853993e-05,
      "loss": 0.4632,
      "step": 6500
    },
    {
      "epoch": 0.40773532152842495,
      "grad_norm": 6.392009258270264,
      "learning_rate": 4.320538210624418e-05,
      "loss": 0.4521,
      "step": 7000
    },
    {
      "epoch": 0.4368592730661696,
      "grad_norm": 4.33333683013916,
      "learning_rate": 4.271998291394843e-05,
      "loss": 0.447,
      "step": 7500
    },
    {
      "epoch": 0.4659832246039143,
      "grad_norm": 3.611548662185669,
      "learning_rate": 4.223458372165269e-05,
      "loss": 0.4465,
      "step": 8000
    },
    {
      "epoch": 0.4951071761416589,
      "grad_norm": 5.756487846374512,
      "learning_rate": 4.1749184529356947e-05,
      "loss": 0.4354,
      "step": 8500
    },
    {
      "epoch": 0.5242311276794035,
      "grad_norm": 5.3887786865234375,
      "learning_rate": 4.12637853370612e-05,
      "loss": 0.4395,
      "step": 9000
    },
    {
      "epoch": 0.5533550792171482,
      "grad_norm": 4.947877883911133,
      "learning_rate": 4.0778386144765457e-05,
      "loss": 0.4431,
      "step": 9500
    },
    {
      "epoch": 0.5824790307548928,
      "grad_norm": 6.60713529586792,
      "learning_rate": 4.0292986952469715e-05,
      "loss": 0.4355,
      "step": 10000
    },
    {
      "epoch": 0.6116029822926374,
      "grad_norm": 4.627679347991943,
      "learning_rate": 3.980758776017397e-05,
      "loss": 0.4317,
      "step": 10500
    },
    {
      "epoch": 0.6407269338303822,
      "grad_norm": 4.922262668609619,
      "learning_rate": 3.9322188567878225e-05,
      "loss": 0.4247,
      "step": 11000
    },
    {
      "epoch": 0.6698508853681268,
      "grad_norm": 3.914152145385742,
      "learning_rate": 3.883678937558248e-05,
      "loss": 0.4334,
      "step": 11500
    },
    {
      "epoch": 0.6989748369058714,
      "grad_norm": 4.864943027496338,
      "learning_rate": 3.835139018328674e-05,
      "loss": 0.4185,
      "step": 12000
    },
    {
      "epoch": 0.7280987884436161,
      "grad_norm": 10.3380708694458,
      "learning_rate": 3.786599099099099e-05,
      "loss": 0.4223,
      "step": 12500
    },
    {
      "epoch": 0.7572227399813607,
      "grad_norm": 4.065073490142822,
      "learning_rate": 3.7380591798695245e-05,
      "loss": 0.4086,
      "step": 13000
    },
    {
      "epoch": 0.7863466915191053,
      "grad_norm": 4.190847873687744,
      "learning_rate": 3.68951926063995e-05,
      "loss": 0.4156,
      "step": 13500
    },
    {
      "epoch": 0.8154706430568499,
      "grad_norm": 5.804997444152832,
      "learning_rate": 3.640979341410376e-05,
      "loss": 0.4052,
      "step": 14000
    },
    {
      "epoch": 0.8445945945945946,
      "grad_norm": 7.701807975769043,
      "learning_rate": 3.592439422180802e-05,
      "loss": 0.4133,
      "step": 14500
    },
    {
      "epoch": 0.8737185461323392,
      "grad_norm": 17.949703216552734,
      "learning_rate": 3.543899502951227e-05,
      "loss": 0.4076,
      "step": 15000
    },
    {
      "epoch": 0.9028424976700838,
      "grad_norm": 5.222975254058838,
      "learning_rate": 3.495359583721653e-05,
      "loss": 0.4201,
      "step": 15500
    },
    {
      "epoch": 0.9319664492078286,
      "grad_norm": 4.896594047546387,
      "learning_rate": 3.446819664492079e-05,
      "loss": 0.4035,
      "step": 16000
    },
    {
      "epoch": 0.9610904007455732,
      "grad_norm": 5.599170207977295,
      "learning_rate": 3.398279745262504e-05,
      "loss": 0.4065,
      "step": 16500
    },
    {
      "epoch": 0.9902143522833178,
      "grad_norm": 3.4558660984039307,
      "learning_rate": 3.34973982603293e-05,
      "loss": 0.413,
      "step": 17000
    },
    {
      "epoch": 1.0193383038210624,
      "grad_norm": 6.404013156890869,
      "learning_rate": 3.301199906803355e-05,
      "loss": 0.3758,
      "step": 17500
    },
    {
      "epoch": 1.048462255358807,
      "grad_norm": 5.1876420974731445,
      "learning_rate": 3.252659987573781e-05,
      "loss": 0.3567,
      "step": 18000
    },
    {
      "epoch": 1.0775862068965518,
      "grad_norm": 3.246553421020508,
      "learning_rate": 3.204120068344206e-05,
      "loss": 0.362,
      "step": 18500
    },
    {
      "epoch": 1.1067101584342964,
      "grad_norm": 5.665416240692139,
      "learning_rate": 3.155580149114632e-05,
      "loss": 0.3639,
      "step": 19000
    },
    {
      "epoch": 1.135834109972041,
      "grad_norm": 4.938392162322998,
      "learning_rate": 3.1070402298850576e-05,
      "loss": 0.3649,
      "step": 19500
    },
    {
      "epoch": 1.1649580615097856,
      "grad_norm": 4.021746635437012,
      "learning_rate": 3.0585003106554835e-05,
      "loss": 0.3688,
      "step": 20000
    },
    {
      "epoch": 1.1940820130475303,
      "grad_norm": 5.639608860015869,
      "learning_rate": 3.009960391425909e-05,
      "loss": 0.357,
      "step": 20500
    },
    {
      "epoch": 1.2232059645852749,
      "grad_norm": 5.954344272613525,
      "learning_rate": 2.9614204721963345e-05,
      "loss": 0.3614,
      "step": 21000
    },
    {
      "epoch": 1.2523299161230197,
      "grad_norm": 4.498464107513428,
      "learning_rate": 2.9128805529667603e-05,
      "loss": 0.3592,
      "step": 21500
    },
    {
      "epoch": 1.281453867660764,
      "grad_norm": 7.547128200531006,
      "learning_rate": 2.8643406337371858e-05,
      "loss": 0.3597,
      "step": 22000
    },
    {
      "epoch": 1.310577819198509,
      "grad_norm": 6.078393459320068,
      "learning_rate": 2.815800714507611e-05,
      "loss": 0.3643,
      "step": 22500
    },
    {
      "epoch": 1.3397017707362535,
      "grad_norm": 5.5143303871154785,
      "learning_rate": 2.7672607952780365e-05,
      "loss": 0.359,
      "step": 23000
    },
    {
      "epoch": 1.3688257222739981,
      "grad_norm": 5.585659027099609,
      "learning_rate": 2.7187208760484623e-05,
      "loss": 0.3647,
      "step": 23500
    },
    {
      "epoch": 1.3979496738117427,
      "grad_norm": 6.159289836883545,
      "learning_rate": 2.6701809568188878e-05,
      "loss": 0.3525,
      "step": 24000
    },
    {
      "epoch": 1.4270736253494873,
      "grad_norm": 4.410373687744141,
      "learning_rate": 2.6216410375893136e-05,
      "loss": 0.3518,
      "step": 24500
    },
    {
      "epoch": 1.4561975768872322,
      "grad_norm": 4.7880754470825195,
      "learning_rate": 2.573101118359739e-05,
      "loss": 0.3568,
      "step": 25000
    },
    {
      "epoch": 1.4853215284249768,
      "grad_norm": 4.985711097717285,
      "learning_rate": 2.524561199130165e-05,
      "loss": 0.3566,
      "step": 25500
    },
    {
      "epoch": 1.5144454799627214,
      "grad_norm": 7.0648088455200195,
      "learning_rate": 2.47602127990059e-05,
      "loss": 0.3634,
      "step": 26000
    },
    {
      "epoch": 1.543569431500466,
      "grad_norm": 5.251628875732422,
      "learning_rate": 2.427481360671016e-05,
      "loss": 0.3526,
      "step": 26500
    },
    {
      "epoch": 1.5726933830382106,
      "grad_norm": 8.0952787399292,
      "learning_rate": 2.3789414414414415e-05,
      "loss": 0.3489,
      "step": 27000
    },
    {
      "epoch": 1.6018173345759554,
      "grad_norm": 3.6345725059509277,
      "learning_rate": 2.3304015222118673e-05,
      "loss": 0.3591,
      "step": 27500
    },
    {
      "epoch": 1.6309412861136998,
      "grad_norm": 6.777653217315674,
      "learning_rate": 2.2818616029822928e-05,
      "loss": 0.3566,
      "step": 28000
    },
    {
      "epoch": 1.6600652376514446,
      "grad_norm": 5.334507465362549,
      "learning_rate": 2.2333216837527183e-05,
      "loss": 0.3506,
      "step": 28500
    },
    {
      "epoch": 1.689189189189189,
      "grad_norm": 5.431693077087402,
      "learning_rate": 2.1847817645231438e-05,
      "loss": 0.358,
      "step": 29000
    },
    {
      "epoch": 1.7183131407269339,
      "grad_norm": 8.189932823181152,
      "learning_rate": 2.1362418452935696e-05,
      "loss": 0.3535,
      "step": 29500
    },
    {
      "epoch": 1.7474370922646785,
      "grad_norm": 6.549941062927246,
      "learning_rate": 2.087701926063995e-05,
      "loss": 0.3502,
      "step": 30000
    },
    {
      "epoch": 1.776561043802423,
      "grad_norm": 5.226093292236328,
      "learning_rate": 2.039162006834421e-05,
      "loss": 0.3509,
      "step": 30500
    },
    {
      "epoch": 1.805684995340168,
      "grad_norm": 4.530932426452637,
      "learning_rate": 1.990622087604846e-05,
      "loss": 0.3439,
      "step": 31000
    },
    {
      "epoch": 1.8348089468779123,
      "grad_norm": 2.6704397201538086,
      "learning_rate": 1.942082168375272e-05,
      "loss": 0.3417,
      "step": 31500
    },
    {
      "epoch": 1.8639328984156571,
      "grad_norm": 4.931678295135498,
      "learning_rate": 1.8935422491456975e-05,
      "loss": 0.3489,
      "step": 32000
    },
    {
      "epoch": 1.8930568499534017,
      "grad_norm": 3.0061357021331787,
      "learning_rate": 1.8450023299161233e-05,
      "loss": 0.3478,
      "step": 32500
    },
    {
      "epoch": 1.9221808014911463,
      "grad_norm": 5.605920314788818,
      "learning_rate": 1.7964624106865488e-05,
      "loss": 0.3469,
      "step": 33000
    },
    {
      "epoch": 1.951304753028891,
      "grad_norm": 8.776355743408203,
      "learning_rate": 1.7479224914569743e-05,
      "loss": 0.3479,
      "step": 33500
    },
    {
      "epoch": 1.9804287045666356,
      "grad_norm": 6.094631671905518,
      "learning_rate": 1.6993825722273998e-05,
      "loss": 0.3465,
      "step": 34000
    },
    {
      "epoch": 2.0095526561043804,
      "grad_norm": 3.9695520401000977,
      "learning_rate": 1.6508426529978256e-05,
      "loss": 0.337,
      "step": 34500
    },
    {
      "epoch": 2.0386766076421248,
      "grad_norm": 9.629318237304688,
      "learning_rate": 1.602302733768251e-05,
      "loss": 0.2991,
      "step": 35000
    },
    {
      "epoch": 2.0678005591798696,
      "grad_norm": 5.794491767883301,
      "learning_rate": 1.5537628145386766e-05,
      "loss": 0.3049,
      "step": 35500
    },
    {
      "epoch": 2.096924510717614,
      "grad_norm": 3.9909379482269287,
      "learning_rate": 1.5052228953091021e-05,
      "loss": 0.307,
      "step": 36000
    },
    {
      "epoch": 2.126048462255359,
      "grad_norm": 5.097212314605713,
      "learning_rate": 1.4566829760795278e-05,
      "loss": 0.3108,
      "step": 36500
    },
    {
      "epoch": 2.1551724137931036,
      "grad_norm": 4.131860733032227,
      "learning_rate": 1.4081430568499535e-05,
      "loss": 0.308,
      "step": 37000
    },
    {
      "epoch": 2.184296365330848,
      "grad_norm": 5.7827372550964355,
      "learning_rate": 1.3596031376203791e-05,
      "loss": 0.3022,
      "step": 37500
    },
    {
      "epoch": 2.213420316868593,
      "grad_norm": 5.650452613830566,
      "learning_rate": 1.3110632183908048e-05,
      "loss": 0.3084,
      "step": 38000
    },
    {
      "epoch": 2.2425442684063372,
      "grad_norm": 7.3451151847839355,
      "learning_rate": 1.2625232991612301e-05,
      "loss": 0.2974,
      "step": 38500
    },
    {
      "epoch": 2.271668219944082,
      "grad_norm": 6.839465141296387,
      "learning_rate": 1.2139833799316558e-05,
      "loss": 0.2978,
      "step": 39000
    },
    {
      "epoch": 2.3007921714818265,
      "grad_norm": 3.4287917613983154,
      "learning_rate": 1.1654434607020815e-05,
      "loss": 0.3137,
      "step": 39500
    },
    {
      "epoch": 2.3299161230195713,
      "grad_norm": 7.438332557678223,
      "learning_rate": 1.1169035414725071e-05,
      "loss": 0.3032,
      "step": 40000
    },
    {
      "epoch": 2.359040074557316,
      "grad_norm": 4.485251426696777,
      "learning_rate": 1.0683636222429326e-05,
      "loss": 0.31,
      "step": 40500
    },
    {
      "epoch": 2.3881640260950605,
      "grad_norm": 4.613361835479736,
      "learning_rate": 1.0198237030133583e-05,
      "loss": 0.3076,
      "step": 41000
    },
    {
      "epoch": 2.4172879776328053,
      "grad_norm": 3.4650862216949463,
      "learning_rate": 9.712837837837838e-06,
      "loss": 0.2925,
      "step": 41500
    },
    {
      "epoch": 2.4464119291705497,
      "grad_norm": 2.9402899742126465,
      "learning_rate": 9.227438645542095e-06,
      "loss": 0.3019,
      "step": 42000
    },
    {
      "epoch": 2.4755358807082946,
      "grad_norm": 3.6428680419921875,
      "learning_rate": 8.742039453246351e-06,
      "loss": 0.3148,
      "step": 42500
    },
    {
      "epoch": 2.5046598322460394,
      "grad_norm": 4.835502624511719,
      "learning_rate": 8.256640260950606e-06,
      "loss": 0.2923,
      "step": 43000
    },
    {
      "epoch": 2.5337837837837838,
      "grad_norm": 6.484007835388184,
      "learning_rate": 7.771241068654863e-06,
      "loss": 0.2889,
      "step": 43500
    },
    {
      "epoch": 2.562907735321528,
      "grad_norm": 7.5975565910339355,
      "learning_rate": 7.285841876359117e-06,
      "loss": 0.2909,
      "step": 44000
    },
    {
      "epoch": 2.592031686859273,
      "grad_norm": 5.18363094329834,
      "learning_rate": 6.800442684063374e-06,
      "loss": 0.3019,
      "step": 44500
    },
    {
      "epoch": 2.621155638397018,
      "grad_norm": 9.518139839172363,
      "learning_rate": 6.31504349176763e-06,
      "loss": 0.3029,
      "step": 45000
    },
    {
      "epoch": 2.650279589934762,
      "grad_norm": 3.660771369934082,
      "learning_rate": 5.829644299471886e-06,
      "loss": 0.301,
      "step": 45500
    },
    {
      "epoch": 2.679403541472507,
      "grad_norm": 6.393496036529541,
      "learning_rate": 5.344245107176142e-06,
      "loss": 0.3006,
      "step": 46000
    },
    {
      "epoch": 2.7085274930102514,
      "grad_norm": 5.712442398071289,
      "learning_rate": 4.858845914880398e-06,
      "loss": 0.315,
      "step": 46500
    },
    {
      "epoch": 2.7376514445479962,
      "grad_norm": 7.901050090789795,
      "learning_rate": 4.373446722584654e-06,
      "loss": 0.3021,
      "step": 47000
    },
    {
      "epoch": 2.766775396085741,
      "grad_norm": 6.925070762634277,
      "learning_rate": 3.8880475302889095e-06,
      "loss": 0.2989,
      "step": 47500
    },
    {
      "epoch": 2.7958993476234855,
      "grad_norm": 2.2326674461364746,
      "learning_rate": 3.402648337993166e-06,
      "loss": 0.3024,
      "step": 48000
    },
    {
      "epoch": 2.8250232991612303,
      "grad_norm": 7.3733320236206055,
      "learning_rate": 2.917249145697422e-06,
      "loss": 0.3078,
      "step": 48500
    },
    {
      "epoch": 2.8541472506989747,
      "grad_norm": 5.280145645141602,
      "learning_rate": 2.431849953401678e-06,
      "loss": 0.2995,
      "step": 49000
    },
    {
      "epoch": 2.8832712022367195,
      "grad_norm": 5.809254169464111,
      "learning_rate": 1.9464507611059337e-06,
      "loss": 0.3008,
      "step": 49500
    },
    {
      "epoch": 2.9123951537744643,
      "grad_norm": 6.8635430335998535,
      "learning_rate": 1.4610515688101895e-06,
      "loss": 0.2995,
      "step": 50000
    },
    {
      "epoch": 2.9415191053122087,
      "grad_norm": 6.390039920806885,
      "learning_rate": 9.756523765144455e-07,
      "loss": 0.301,
      "step": 50500
    },
    {
      "epoch": 2.9706430568499536,
      "grad_norm": 7.5721964836120605,
      "learning_rate": 4.902531842187015e-07,
      "loss": 0.305,
      "step": 51000
    },
    {
      "epoch": 2.999767008387698,
      "grad_norm": 6.955125331878662,
      "learning_rate": 4.85399192295744e-09,
      "loss": 0.3002,
      "step": 51500
    }
  ],
  "logging_steps": 500,
  "max_steps": 51504,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.2121938098975232e+16,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}