{
  "best_global_step": 10000,
  "best_metric": 0.8275137305036369,
  "best_model_checkpoint": "./results/run-0/checkpoint-10000",
  "epoch": 7.0,
  "eval_steps": 500,
  "global_step": 14000,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.05,
      "grad_norm": 2.039576768875122,
      "learning_rate": 1.447537100215313e-06,
      "loss": 0.0,
      "step": 100
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.460484504699707,
      "learning_rate": 2.9096957873014878e-06,
      "loss": 0.0821,
      "step": 200
    },
    {
      "epoch": 0.15,
      "grad_norm": 26.84711456298828,
      "learning_rate": 4.371854474387663e-06,
      "loss": 0.0519,
      "step": 300
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6646966934204102,
      "learning_rate": 5.834013161473838e-06,
      "loss": 0.0,
      "step": 400
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8385223150253296,
      "learning_rate": 7.296171848560012e-06,
      "loss": 0.0,
      "step": 500
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9234638810157776,
      "learning_rate": 8.758330535646188e-06,
      "loss": 0.0,
      "step": 600
    },
    {
      "epoch": 0.35,
      "grad_norm": 38.24382400512695,
      "learning_rate": 9.9933989317167e-06,
      "loss": 0.0,
      "step": 700
    },
    {
      "epoch": 0.4,
      "grad_norm": 42.83884811401367,
      "learning_rate": 9.94162234536513e-06,
      "loss": 0.0,
      "step": 800
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8020399808883667,
      "learning_rate": 9.889845759013559e-06,
      "loss": 0.0,
      "step": 900
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.86091148853302,
      "learning_rate": 9.838069172661988e-06,
      "loss": 0.0,
      "step": 1000
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.16439934074878693,
      "learning_rate": 9.786292586310415e-06,
      "loss": 0.0,
      "step": 1100
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.752115488052368,
      "learning_rate": 9.734515999958846e-06,
      "loss": 0.0,
      "step": 1200
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5777802467346191,
      "learning_rate": 9.682739413607275e-06,
      "loss": 0.0,
      "step": 1300
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.708035945892334,
      "learning_rate": 9.630962827255703e-06,
      "loss": 0.0598,
      "step": 1400
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3160113990306854,
      "learning_rate": 9.579186240904132e-06,
      "loss": 0.0441,
      "step": 1500
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2273201942443848,
      "learning_rate": 9.527409654552563e-06,
      "loss": 0.0,
      "step": 1600
    },
    {
      "epoch": 0.85,
      "grad_norm": 29.215551376342773,
      "learning_rate": 9.47563306820099e-06,
      "loss": 0.0,
      "step": 1700
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.18806450068950653,
      "learning_rate": 9.42385648184942e-06,
      "loss": 0.0,
      "step": 1800
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.41723617911338806,
      "learning_rate": 9.372079895497849e-06,
      "loss": 0.0183,
      "step": 1900
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.9550623893737793,
      "learning_rate": 9.32030330914628e-06,
      "loss": 0.0,
      "step": 2000
    },
    {
      "epoch": 1.0,
      "eval_economic_inequality_accuracy": 0.8047858942065491,
      "eval_economic_inequality_f1": 0.7536914695213144,
      "eval_economic_policy_benefits_accuracy": 0.8282828282828283,
      "eval_economic_policy_benefits_f1": 0.7567090529234769,
      "eval_ethnic_boundaries_accuracy": 0.9526143790849673,
      "eval_ethnic_boundaries_f1": 0.9344984933861545,
      "eval_language_policy_accuracy": 0.7058823529411765,
      "eval_language_policy_f1": 0.6071428571428571,
      "eval_loss": NaN,
      "eval_mother_tongue_education_accuracy": 0.8888888888888888,
      "eval_mother_tongue_education_f1": 0.837037037037037,
      "eval_overall_accuracy": 0.8392311075242039,
      "eval_overall_f1": 0.782435168791527,
      "eval_religion_ethnic_identity_accuracy": 0.8549323017408124,
      "eval_religion_ethnic_identity_f1": 0.8055321027383222,
      "eval_runtime": 4.4874,
      "eval_samples_per_second": 891.378,
      "eval_steps_per_second": 55.711,
      "step": 2000
    },
    {
      "epoch": 1.05,
      "grad_norm": 2.762986183166504,
      "learning_rate": 9.268526722794707e-06,
      "loss": 0.0,
      "step": 2100
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.5487369298934937,
      "learning_rate": 9.216750136443136e-06,
      "loss": 0.0,
      "step": 2200
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.76751184463501,
      "learning_rate": 9.164973550091565e-06,
      "loss": 0.0102,
      "step": 2300
    },
    {
      "epoch": 1.2,
      "grad_norm": 2.681756019592285,
      "learning_rate": 9.113196963739995e-06,
      "loss": 0.0,
      "step": 2400
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.24124185740947723,
      "learning_rate": 9.061420377388424e-06,
      "loss": 0.0,
      "step": 2500
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.943154215812683,
      "learning_rate": 9.009643791036853e-06,
      "loss": 0.0,
      "step": 2600
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.28053101897239685,
      "learning_rate": 8.957867204685282e-06,
      "loss": 0.0347,
      "step": 2700
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7952568531036377,
      "learning_rate": 8.906090618333711e-06,
      "loss": 0.4436,
      "step": 2800
    },
    {
      "epoch": 1.45,
      "grad_norm": 3.859933614730835,
      "learning_rate": 8.85431403198214e-06,
      "loss": 0.0,
      "step": 2900
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.094464898109436,
      "learning_rate": 8.80253744563057e-06,
      "loss": 0.0,
      "step": 3000
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.5670285820960999,
      "learning_rate": 8.750760859278999e-06,
      "loss": 0.0,
      "step": 3100
    },
    {
      "epoch": 1.6,
      "grad_norm": 2.8412418365478516,
      "learning_rate": 8.698984272927428e-06,
      "loss": 0.0,
      "step": 3200
    },
    {
      "epoch": 1.65,
      "grad_norm": 65.82475280761719,
      "learning_rate": 8.647207686575855e-06,
      "loss": 0.0,
      "step": 3300
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.976615905761719,
      "learning_rate": 8.595431100224286e-06,
      "loss": 0.0048,
      "step": 3400
    },
    {
      "epoch": 1.75,
      "grad_norm": 126.71548461914062,
      "learning_rate": 8.543654513872716e-06,
      "loss": 0.0,
      "step": 3500
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.595015823841095,
      "learning_rate": 8.491877927521143e-06,
      "loss": 0.0033,
      "step": 3600
    },
    {
      "epoch": 1.85,
      "grad_norm": 3.98215651512146,
      "learning_rate": 8.440101341169572e-06,
      "loss": 0.1482,
      "step": 3700
    },
    {
      "epoch": 1.9,
      "grad_norm": 3.0417540073394775,
      "learning_rate": 8.388324754818003e-06,
      "loss": 0.0031,
      "step": 3800
    },
    {
      "epoch": 1.95,
      "grad_norm": 27.85649299621582,
      "learning_rate": 8.33654816846643e-06,
      "loss": 0.0589,
      "step": 3900
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.3094232082366943,
      "learning_rate": 8.28477158211486e-06,
      "loss": 0.0,
      "step": 4000
    },
    {
      "epoch": 2.0,
      "eval_economic_inequality_accuracy": 0.8236775818639799,
      "eval_economic_inequality_f1": 0.8161521311449569,
      "eval_economic_policy_benefits_accuracy": 0.835016835016835,
      "eval_economic_policy_benefits_f1": 0.81117677081242,
      "eval_ethnic_boundaries_accuracy": 0.9477124183006536,
      "eval_ethnic_boundaries_f1": 0.9421262334088456,
      "eval_language_policy_accuracy": 0.7058823529411765,
      "eval_language_policy_f1": 0.6071428571428571,
      "eval_loss": NaN,
      "eval_mother_tongue_education_accuracy": 0.7777777777777778,
      "eval_mother_tongue_education_f1": 0.7407407407407408,
      "eval_overall_accuracy": 0.8277126438976526,
      "eval_overall_f1": 0.7960173130694123,
      "eval_religion_ethnic_identity_accuracy": 0.8762088974854932,
      "eval_religion_ethnic_identity_f1": 0.8587651451666537,
      "eval_runtime": 4.4654,
      "eval_samples_per_second": 895.775,
      "eval_steps_per_second": 55.986,
      "step": 4000
    },
    {
      "epoch": 2.05,
      "grad_norm": 8.547677040100098,
      "learning_rate": 8.232994995763289e-06,
      "loss": 0.0,
      "step": 4100
    },
    {
      "epoch": 2.1,
      "grad_norm": 117.06266784667969,
      "learning_rate": 8.181218409411718e-06,
      "loss": 0.0,
      "step": 4200
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.19274334609508514,
      "learning_rate": 8.129441823060147e-06,
      "loss": 0.0,
      "step": 4300
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.0360257625579834,
      "learning_rate": 8.077665236708576e-06,
      "loss": 0.0,
      "step": 4400
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.08476172387599945,
      "learning_rate": 8.025888650357006e-06,
      "loss": 0.0,
      "step": 4500
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.08015169203281403,
      "learning_rate": 7.974112064005435e-06,
      "loss": 0.0,
      "step": 4600
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.22844159603118896,
      "learning_rate": 7.922335477653864e-06,
      "loss": 0.0006,
      "step": 4700
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.08015509694814682,
      "learning_rate": 7.870558891302293e-06,
      "loss": 0.0006,
      "step": 4800
    },
    {
      "epoch": 2.45,
      "grad_norm": 5.426948547363281,
      "learning_rate": 7.818782304950722e-06,
      "loss": 0.0,
      "step": 4900
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.022730231285095215,
      "learning_rate": 7.767005718599151e-06,
      "loss": 0.0,
      "step": 5000
    },
    {
      "epoch": 2.55,
      "grad_norm": 5.662137508392334,
      "learning_rate": 7.71522913224758e-06,
      "loss": 0.0,
      "step": 5100
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.05326724052429199,
      "learning_rate": 7.66345254589601e-06,
      "loss": 0.0,
      "step": 5200
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.7219042778015137,
      "learning_rate": 7.611675959544438e-06,
      "loss": 0.0003,
      "step": 5300
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.045983791351318,
      "learning_rate": 7.559899373192868e-06,
      "loss": 0.0,
      "step": 5400
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.17610976099967957,
      "learning_rate": 7.5081227868412964e-06,
      "loss": 0.0,
      "step": 5500
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.03921931982040405,
      "learning_rate": 7.456346200489726e-06,
      "loss": 0.0,
      "step": 5600
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.3493025302886963,
      "learning_rate": 7.404569614138155e-06,
      "loss": 0.0,
      "step": 5700
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.17839455604553223,
      "learning_rate": 7.352793027786584e-06,
      "loss": 0.0,
      "step": 5800
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.01533100288361311,
      "learning_rate": 7.301016441435013e-06,
      "loss": 0.0,
      "step": 5900
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.6168579459190369,
      "learning_rate": 7.249239855083442e-06,
      "loss": 0.0,
      "step": 6000
    },
    {
      "epoch": 3.0,
      "eval_economic_inequality_accuracy": 0.818639798488665,
      "eval_economic_inequality_f1": 0.8128041981528128,
      "eval_economic_policy_benefits_accuracy": 0.8686868686868687,
      "eval_economic_policy_benefits_f1": 0.849679269639744,
      "eval_ethnic_boundaries_accuracy": 0.9379084967320261,
      "eval_ethnic_boundaries_f1": 0.9377319200237987,
      "eval_language_policy_accuracy": 0.6470588235294118,
      "eval_language_policy_f1": 0.5673202614379085,
      "eval_loss": NaN,
      "eval_mother_tongue_education_accuracy": 0.7777777777777778,
      "eval_mother_tongue_education_f1": 0.7407407407407408,
      "eval_overall_accuracy": 0.8181454231515234,
      "eval_overall_f1": 0.794560338168735,
      "eval_religion_ethnic_identity_accuracy": 0.8588007736943907,
      "eval_religion_ethnic_identity_f1": 0.8590856390174055,
      "eval_runtime": 4.4722,
      "eval_samples_per_second": 894.406,
      "eval_steps_per_second": 55.9,
      "step": 6000
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.06797665357589722,
      "learning_rate": 7.197463268731871e-06,
      "loss": 0.0,
      "step": 6100
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.05142182484269142,
      "learning_rate": 7.145686682380301e-06,
      "loss": 0.0002,
      "step": 6200
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.04868694022297859,
      "learning_rate": 7.09391009602873e-06,
      "loss": 0.0,
      "step": 6300
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.026197949424386024,
      "learning_rate": 7.042133509677158e-06,
      "loss": 0.0005,
      "step": 6400
    },
    {
      "epoch": 3.25,
      "grad_norm": 12.894441604614258,
      "learning_rate": 6.990356923325587e-06,
      "loss": 0.0,
      "step": 6500
    },
    {
      "epoch": 3.3,
      "grad_norm": 4.21354341506958,
      "learning_rate": 6.938580336974017e-06,
      "loss": 0.0,
      "step": 6600
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.0193119365721941,
      "learning_rate": 6.886803750622446e-06,
      "loss": 0.0,
      "step": 6700
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.07511546462774277,
      "learning_rate": 6.835027164270875e-06,
      "loss": 0.0,
      "step": 6800
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.9280033111572266,
      "learning_rate": 6.783250577919304e-06,
      "loss": 0.0003,
      "step": 6900
    },
    {
      "epoch": 3.5,
      "grad_norm": 108.68927001953125,
      "learning_rate": 6.731473991567734e-06,
      "loss": 0.0,
      "step": 7000
    },
    {
      "epoch": 3.55,
      "grad_norm": 4.475680828094482,
      "learning_rate": 6.679697405216162e-06,
      "loss": 0.0,
      "step": 7100
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.040891651064157486,
      "learning_rate": 6.6279208188645915e-06,
      "loss": 0.0,
      "step": 7200
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.032305650413036346,
      "learning_rate": 6.576144232513021e-06,
      "loss": 0.0,
      "step": 7300
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.027986843138933182,
      "learning_rate": 6.524367646161449e-06,
      "loss": 0.0002,
      "step": 7400
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.023173924535512924,
      "learning_rate": 6.472591059809879e-06,
      "loss": 0.0,
      "step": 7500
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.10551729053258896,
      "learning_rate": 6.420814473458308e-06,
      "loss": 0.0,
      "step": 7600
    },
    {
      "epoch": 3.85,
      "grad_norm": 4.631871223449707,
      "learning_rate": 6.3690378871067366e-06,
      "loss": 0.0,
      "step": 7700
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.09392645955085754,
      "learning_rate": 6.317261300755166e-06,
      "loss": 0.0,
      "step": 7800
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.32573947310447693,
      "learning_rate": 6.265484714403596e-06,
      "loss": 0.0,
      "step": 7900
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.012679875828325748,
      "learning_rate": 6.213708128052024e-06,
      "loss": 0.0,
      "step": 8000
    },
    {
      "epoch": 4.0,
      "eval_economic_inequality_accuracy": 0.809823677581864,
      "eval_economic_inequality_f1": 0.8053367188692558,
      "eval_economic_policy_benefits_accuracy": 0.8451178451178452,
      "eval_economic_policy_benefits_f1": 0.8262377279238239,
      "eval_ethnic_boundaries_accuracy": 0.9379084967320261,
      "eval_ethnic_boundaries_f1": 0.9368969810146281,
      "eval_language_policy_accuracy": 0.7058823529411765,
      "eval_language_policy_f1": 0.6071428571428571,
      "eval_loss": NaN,
      "eval_mother_tongue_education_accuracy": 0.8888888888888888,
      "eval_mother_tongue_education_f1": 0.837037037037037,
      "eval_overall_accuracy": 0.8436493204617509,
      "eval_overall_f1": 0.8135939571355589,
      "eval_religion_ethnic_identity_accuracy": 0.874274661508704,
      "eval_religion_ethnic_identity_f1": 0.8689124208257524,
      "eval_runtime": 4.4538,
      "eval_samples_per_second": 898.1,
      "eval_steps_per_second": 56.131,
      "step": 8000
    },
    {
      "epoch": 4.05,
      "grad_norm": 0.08179322630167007,
      "learning_rate": 6.161931541700453e-06,
      "loss": 0.0,
      "step": 8100
    },
    {
      "epoch": 4.1,
      "grad_norm": 1.8278212547302246,
      "learning_rate": 6.1101549553488824e-06,
      "loss": 0.0,
      "step": 8200
    },
    {
      "epoch": 4.15,
      "grad_norm": 0.9412699341773987,
      "learning_rate": 6.058378368997311e-06,
      "loss": 0.0,
      "step": 8300
    },
    {
      "epoch": 4.2,
      "grad_norm": 0.3116857707500458,
      "learning_rate": 6.006601782645741e-06,
      "loss": 0.0,
      "step": 8400
    },
    {
      "epoch": 4.25,
      "grad_norm": 0.04464244097471237,
      "learning_rate": 5.95482519629417e-06,
      "loss": 0.0,
      "step": 8500
    },
    {
      "epoch": 4.3,
      "grad_norm": 0.0,
      "learning_rate": 5.903048609942598e-06,
      "loss": 0.0,
      "step": 8600
    },
    {
      "epoch": 4.35,
      "grad_norm": 0.011906923726201057,
      "learning_rate": 5.8512720235910274e-06,
      "loss": 0.0,
      "step": 8700
    },
    {
      "epoch": 4.4,
      "grad_norm": 0.022279787808656693,
      "learning_rate": 5.7994954372394575e-06,
      "loss": 0.0002,
      "step": 8800
    },
    {
      "epoch": 4.45,
      "grad_norm": 0.07576213777065277,
      "learning_rate": 5.747718850887886e-06,
      "loss": 0.0001,
      "step": 8900
    },
    {
      "epoch": 4.5,
      "grad_norm": 0.02016213722527027,
      "learning_rate": 5.695942264536315e-06,
      "loss": 0.0,
      "step": 9000
    },
    {
      "epoch": 4.55,
      "grad_norm": 0.07930700480937958,
      "learning_rate": 5.644165678184744e-06,
      "loss": 0.0,
      "step": 9100
    },
    {
      "epoch": 4.6,
      "grad_norm": 0.016734231263399124,
      "learning_rate": 5.592389091833174e-06,
      "loss": 0.0,
      "step": 9200
    },
    {
      "epoch": 4.65,
      "grad_norm": 0.2023557871580124,
      "learning_rate": 5.5406125054816025e-06,
      "loss": 0.0,
      "step": 9300
    },
    {
      "epoch": 4.7,
      "grad_norm": 0.03384646028280258,
      "learning_rate": 5.488835919130032e-06,
      "loss": 0.0,
      "step": 9400
    },
    {
      "epoch": 4.75,
      "grad_norm": 0.04355572536587715,
      "learning_rate": 5.437059332778461e-06,
      "loss": 0.0001,
      "step": 9500
    },
    {
      "epoch": 4.8,
      "grad_norm": 0.02533441036939621,
      "learning_rate": 5.385282746426889e-06,
      "loss": 0.0001,
      "step": 9600
    },
    {
      "epoch": 4.85,
      "grad_norm": 0.10049561411142349,
      "learning_rate": 5.333506160075319e-06,
      "loss": 0.0,
      "step": 9700
    },
    {
      "epoch": 4.9,
      "grad_norm": 0.018155105412006378,
      "learning_rate": 5.281729573723748e-06,
      "loss": 0.0,
      "step": 9800
    },
    {
      "epoch": 4.95,
      "grad_norm": 0.00853881984949112,
      "learning_rate": 5.229952987372177e-06,
      "loss": 0.0,
      "step": 9900
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.17767295241355896,
      "learning_rate": 5.178176401020606e-06,
      "loss": 0.0,
      "step": 10000
    },
    {
      "epoch": 5.0,
      "eval_economic_inequality_accuracy": 0.8261964735516373,
      "eval_economic_inequality_f1": 0.8121087395907841,
      "eval_economic_policy_benefits_accuracy": 0.8653198653198653,
      "eval_economic_policy_benefits_f1": 0.8471939977021603,
      "eval_ethnic_boundaries_accuracy": 0.9362745098039216,
      "eval_ethnic_boundaries_f1": 0.9365673542144131,
      "eval_language_policy_accuracy": 0.7352941176470589,
      "eval_language_policy_f1": 0.6572786690433748,
      "eval_loss": NaN,
      "eval_mother_tongue_education_accuracy": 0.8888888888888888,
      "eval_mother_tongue_education_f1": 0.837037037037037,
      "eval_overall_accuracy": 0.8553418707750738,
      "eval_overall_f1": 0.8275137305036369,
      "eval_religion_ethnic_identity_accuracy": 0.8800773694390716,
      "eval_religion_ethnic_identity_f1": 0.874896585434052,
      "eval_runtime": 4.4693,
      "eval_samples_per_second": 895.0,
      "eval_steps_per_second": 55.937,
      "step": 10000
    },
    {
      "epoch": 5.05,
      "grad_norm": 0.1333349049091339,
      "learning_rate": 5.126399814669036e-06,
      "loss": 0.0014,
      "step": 10100
    },
    {
      "epoch": 5.1,
      "grad_norm": 1.9743980169296265,
      "learning_rate": 5.074623228317464e-06,
      "loss": 0.0071,
      "step": 10200
    },
    {
      "epoch": 5.15,
      "grad_norm": 0.02137836627662182,
      "learning_rate": 5.022846641965893e-06,
      "loss": 0.0001,
      "step": 10300
    },
    {
      "epoch": 5.2,
      "grad_norm": 28.67204475402832,
      "learning_rate": 4.9710700556143225e-06,
      "loss": 0.0,
      "step": 10400
    },
    {
      "epoch": 5.25,
      "grad_norm": 0.03255374729633331,
      "learning_rate": 4.919293469262752e-06,
      "loss": 0.0011,
      "step": 10500
    },
    {
      "epoch": 5.3,
      "grad_norm": 0.013030334375798702,
      "learning_rate": 4.867516882911181e-06,
      "loss": 0.0,
      "step": 10600
    },
    {
      "epoch": 5.35,
      "grad_norm": 0.006988250184804201,
      "learning_rate": 4.815740296559609e-06,
      "loss": 0.0,
      "step": 10700
    },
    {
      "epoch": 5.4,
      "grad_norm": 0.19381622970104218,
      "learning_rate": 4.763963710208039e-06,
      "loss": 0.0,
      "step": 10800
    },
    {
      "epoch": 5.45,
      "grad_norm": 0.32009562849998474,
      "learning_rate": 4.7121871238564676e-06,
      "loss": 0.0,
      "step": 10900
    },
    {
      "epoch": 5.5,
      "grad_norm": 0.018149327486753464,
      "learning_rate": 4.660410537504898e-06,
      "loss": 0.0,
      "step": 11000
    },
    {
      "epoch": 5.55,
      "grad_norm": 0.02935118041932583,
      "learning_rate": 4.608633951153326e-06,
      "loss": 0.0,
      "step": 11100
    },
    {
      "epoch": 5.6,
      "grad_norm": 0.005307406652718782,
      "learning_rate": 4.556857364801755e-06,
      "loss": 0.0,
      "step": 11200
    },
    {
      "epoch": 5.65,
      "grad_norm": 0.017713243141770363,
      "learning_rate": 4.505080778450184e-06,
      "loss": 0.0,
      "step": 11300
    },
    {
      "epoch": 5.7,
      "grad_norm": 0.4359346330165863,
      "learning_rate": 4.4533041920986134e-06,
      "loss": 0.0,
      "step": 11400
    },
    {
      "epoch": 5.75,
      "grad_norm": 0.007976237684488297,
      "learning_rate": 4.401527605747043e-06,
      "loss": 0.0,
      "step": 11500
    },
    {
      "epoch": 5.8,
      "grad_norm": 0.013159127905964851,
      "learning_rate": 4.349751019395472e-06,
      "loss": 0.0,
      "step": 11600
    },
    {
      "epoch": 5.85,
      "grad_norm": 0.00992471445351839,
      "learning_rate": 4.297974433043901e-06,
      "loss": 0.0,
      "step": 11700
    },
    {
      "epoch": 5.9,
      "grad_norm": 0.030109956860542297,
      "learning_rate": 4.246197846692329e-06,
      "loss": 0.0,
      "step": 11800
    },
    {
      "epoch": 5.95,
      "grad_norm": 1.787908673286438,
      "learning_rate": 4.194421260340759e-06,
      "loss": 0.0,
      "step": 11900
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.026327166706323624,
      "learning_rate": 4.142644673989188e-06,
      "loss": 0.0,
      "step": 12000
    },
    {
      "epoch": 6.0,
      "eval_economic_inequality_accuracy": 0.818639798488665,
      "eval_economic_inequality_f1": 0.8122124279169094,
      "eval_economic_policy_benefits_accuracy": 0.8686868686868687,
      "eval_economic_policy_benefits_f1": 0.8566211399694749,
      "eval_ethnic_boundaries_accuracy": 0.9362745098039216,
      "eval_ethnic_boundaries_f1": 0.9357377620451811,
      "eval_language_policy_accuracy": 0.7352941176470589,
      "eval_language_policy_f1": 0.6586056644880174,
      "eval_loss": NaN,
      "eval_mother_tongue_education_accuracy": 0.7777777777777778,
      "eval_mother_tongue_education_f1": 0.7407407407407408,
      "eval_overall_accuracy": 0.836769818966157,
      "eval_overall_f1": 0.8138010681888227,
      "eval_religion_ethnic_identity_accuracy": 0.8839458413926499,
      "eval_religion_ethnic_identity_f1": 0.8788886739726133,
      "eval_runtime": 4.4805,
      "eval_samples_per_second": 892.749,
      "eval_steps_per_second": 55.797,
      "step": 12000
    },
    {
      "epoch": 6.05,
      "grad_norm": 0.14624720811843872,
      "learning_rate": 4.090868087637617e-06,
      "loss": 0.0,
      "step": 12100
    },
    {
      "epoch": 6.1,
      "grad_norm": 0.035003069788217545,
      "learning_rate": 4.039091501286046e-06,
      "loss": 0.0,
      "step": 12200
    },
    {
      "epoch": 6.15,
      "grad_norm": 0.006480562500655651,
      "learning_rate": 3.987314914934475e-06,
      "loss": 0.0,
      "step": 12300
    },
    {
      "epoch": 6.2,
      "grad_norm": 16.347896575927734,
      "learning_rate": 3.935538328582904e-06,
      "loss": 0.0,
      "step": 12400
    },
    {
      "epoch": 6.25,
      "grad_norm": 0.009049425832927227,
      "learning_rate": 3.8837617422313335e-06,
      "loss": 0.0,
      "step": 12500
    },
    {
      "epoch": 6.3,
      "grad_norm": 0.16999471187591553,
      "learning_rate": 3.831985155879763e-06,
      "loss": 0.0,
      "step": 12600
    },
    {
      "epoch": 6.35,
      "grad_norm": 0.010557026602327824,
      "learning_rate": 3.780208569528192e-06,
      "loss": 0.0,
      "step": 12700
    },
    {
      "epoch": 6.4,
      "grad_norm": 0.46991246938705444,
      "learning_rate": 3.7284319831766206e-06,
      "loss": 0.0,
      "step": 12800
    },
    {
      "epoch": 6.45,
      "grad_norm": 0.01426410861313343,
      "learning_rate": 3.6766553968250498e-06,
      "loss": 0.0,
      "step": 12900
    },
    {
      "epoch": 6.5,
      "grad_norm": 0.19090694189071655,
      "learning_rate": 3.624878810473479e-06,
      "loss": 0.0001,
      "step": 13000
    },
    {
      "epoch": 6.55,
      "grad_norm": 0.014205949380993843,
      "learning_rate": 3.573102224121908e-06,
      "loss": 0.2072,
      "step": 13100
    },
    {
      "epoch": 6.6,
      "grad_norm": 0.05406644195318222,
      "learning_rate": 3.521325637770337e-06,
      "loss": 0.0,
      "step": 13200
    },
    {
      "epoch": 6.65,
      "grad_norm": 0.004384028259664774,
      "learning_rate": 3.4695490514187665e-06,
      "loss": 0.0,
      "step": 13300
    },
    {
      "epoch": 6.7,
      "grad_norm": 0.00611688569188118,
      "learning_rate": 3.417772465067195e-06,
      "loss": 0.0,
      "step": 13400
    },
    {
      "epoch": 6.75,
      "grad_norm": 0.0035077868960797787,
      "learning_rate": 3.365995878715625e-06,
      "loss": 0.0,
      "step": 13500
    },
    {
      "epoch": 6.8,
      "grad_norm": 0.013541534543037415,
      "learning_rate": 3.3142192923640536e-06,
      "loss": 0.0018,
      "step": 13600
    },
    {
      "epoch": 6.85,
      "grad_norm": 0.011575588956475258,
      "learning_rate": 3.2624427060124823e-06,
      "loss": 0.0,
      "step": 13700
    },
    {
      "epoch": 6.9,
      "grad_norm": 0.012222085148096085,
      "learning_rate": 3.210666119660912e-06,
      "loss": 0.0,
      "step": 13800
    },
    {
      "epoch": 6.95,
      "grad_norm": 0.18693706393241882,
      "learning_rate": 3.1588895333093407e-06,
      "loss": 0.0,
      "step": 13900
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.04562217742204666,
      "learning_rate": 3.10711294695777e-06,
      "loss": 0.0,
      "step": 14000
    },
    {
      "epoch": 7.0,
      "eval_economic_inequality_accuracy": 0.8035264483627204,
      "eval_economic_inequality_f1": 0.8050953778684486,
      "eval_economic_policy_benefits_accuracy": 0.8417508417508418,
      "eval_economic_policy_benefits_f1": 0.8363986338896733,
      "eval_ethnic_boundaries_accuracy": 0.9183006535947712,
      "eval_ethnic_boundaries_f1": 0.9256093210513574,
      "eval_language_policy_accuracy": 0.7058823529411765,
      "eval_language_policy_f1": 0.6587850826605153,
      "eval_loss": NaN,
      "eval_mother_tongue_education_accuracy": 0.7777777777777778,
      "eval_mother_tongue_education_f1": 0.7407407407407408,
      "eval_overall_accuracy": 0.816061278039622,
      "eval_overall_f1": 0.8032276539443766,
      "eval_religion_ethnic_identity_accuracy": 0.8491295938104448,
      "eval_religion_ethnic_identity_f1": 0.8527367674555238,
      "eval_runtime": 4.4694,
      "eval_samples_per_second": 894.976,
      "eval_steps_per_second": 55.936,
      "step": 14000
    }
  ],
  "logging_steps": 100,
  "max_steps": 20000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 2,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 2
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.9472671563776e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": {
    "gradient_accumulation_steps": 1,
    "learning_rate": 1.0001165419669436e-05,
    "num_train_epochs": 10,
    "per_device_train_batch_size": 8,
    "warmup_steps": 684,
    "weight_decay": 0.07522641729243515
  }
}