{
  "best_metric": 0.11290808767080307,
  "best_model_checkpoint": "./hatebert_finetuned/checkpoint-15957",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 47871,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03133421069123269,
      "grad_norm": 0.1738063097000122,
      "learning_rate": 1.979110526205845e-05,
      "loss": 0.2021,
      "step": 500
    },
    {
      "epoch": 0.06266842138246538,
      "grad_norm": 5.1839799880981445,
      "learning_rate": 1.9582210524116897e-05,
      "loss": 0.1506,
      "step": 1000
    },
    {
      "epoch": 0.09400263207369806,
      "grad_norm": 2.0434365272521973,
      "learning_rate": 1.937331578617535e-05,
      "loss": 0.1376,
      "step": 1500
    },
    {
      "epoch": 0.12533684276493076,
      "grad_norm": 0.7909517884254456,
      "learning_rate": 1.9164421048233796e-05,
      "loss": 0.142,
      "step": 2000
    },
    {
      "epoch": 0.15667105345616344,
      "grad_norm": 0.14363446831703186,
      "learning_rate": 1.8955526310292244e-05,
      "loss": 0.1411,
      "step": 2500
    },
    {
      "epoch": 0.18800526414739613,
      "grad_norm": 13.209172248840332,
      "learning_rate": 1.8746631572350692e-05,
      "loss": 0.1187,
      "step": 3000
    },
    {
      "epoch": 0.2193394748386288,
      "grad_norm": 0.11167433857917786,
      "learning_rate": 1.8537736834409143e-05,
      "loss": 0.1458,
      "step": 3500
    },
    {
      "epoch": 0.2506736855298615,
      "grad_norm": 0.007185101043432951,
      "learning_rate": 1.832884209646759e-05,
      "loss": 0.1271,
      "step": 4000
    },
    {
      "epoch": 0.2820078962210942,
      "grad_norm": 14.943521499633789,
      "learning_rate": 1.811994735852604e-05,
      "loss": 0.1116,
      "step": 4500
    },
    {
      "epoch": 0.3133421069123269,
      "grad_norm": 0.0487419068813324,
      "learning_rate": 1.7911052620584487e-05,
      "loss": 0.1318,
      "step": 5000
    },
    {
      "epoch": 0.34467631760355955,
      "grad_norm": 0.08870559185743332,
      "learning_rate": 1.7702157882642938e-05,
      "loss": 0.1226,
      "step": 5500
    },
    {
      "epoch": 0.37601052829479226,
      "grad_norm": 3.435192346572876,
      "learning_rate": 1.7493263144701386e-05,
      "loss": 0.1302,
      "step": 6000
    },
    {
      "epoch": 0.40734473898602497,
      "grad_norm": 12.683052062988281,
      "learning_rate": 1.7284368406759833e-05,
      "loss": 0.1367,
      "step": 6500
    },
    {
      "epoch": 0.4386789496772576,
      "grad_norm": 18.9424991607666,
      "learning_rate": 1.7075473668818285e-05,
      "loss": 0.125,
      "step": 7000
    },
    {
      "epoch": 0.47001316036849033,
      "grad_norm": 0.24961325526237488,
      "learning_rate": 1.6866578930876732e-05,
      "loss": 0.1358,
      "step": 7500
    },
    {
      "epoch": 0.501347371059723,
      "grad_norm": 9.563410758972168,
      "learning_rate": 1.665768419293518e-05,
      "loss": 0.1238,
      "step": 8000
    },
    {
      "epoch": 0.5326815817509557,
      "grad_norm": 6.3511962890625,
      "learning_rate": 1.644878945499363e-05,
      "loss": 0.1278,
      "step": 8500
    },
    {
      "epoch": 0.5640157924421884,
      "grad_norm": 0.009973131120204926,
      "learning_rate": 1.623989471705208e-05,
      "loss": 0.1172,
      "step": 9000
    },
    {
      "epoch": 0.5953500031334211,
      "grad_norm": 1.0254442691802979,
      "learning_rate": 1.6030999979110527e-05,
      "loss": 0.1292,
      "step": 9500
    },
    {
      "epoch": 0.6266842138246538,
      "grad_norm": 0.6539679765701294,
      "learning_rate": 1.582210524116898e-05,
      "loss": 0.0982,
      "step": 10000
    },
    {
      "epoch": 0.6580184245158864,
      "grad_norm": 0.045102544128894806,
      "learning_rate": 1.5613210503227426e-05,
      "loss": 0.1211,
      "step": 10500
    },
    {
      "epoch": 0.6893526352071191,
      "grad_norm": 0.23447087407112122,
      "learning_rate": 1.5404315765285874e-05,
      "loss": 0.1232,
      "step": 11000
    },
    {
      "epoch": 0.7206868458983519,
      "grad_norm": 0.21137630939483643,
      "learning_rate": 1.5195421027344322e-05,
      "loss": 0.1145,
      "step": 11500
    },
    {
      "epoch": 0.7520210565895845,
      "grad_norm": 0.10944367200136185,
      "learning_rate": 1.4986526289402771e-05,
      "loss": 0.1193,
      "step": 12000
    },
    {
      "epoch": 0.7833552672808172,
      "grad_norm": 6.194149971008301,
      "learning_rate": 1.4777631551461221e-05,
      "loss": 0.1273,
      "step": 12500
    },
    {
      "epoch": 0.8146894779720499,
      "grad_norm": 0.34108805656433105,
      "learning_rate": 1.4568736813519669e-05,
      "loss": 0.1237,
      "step": 13000
    },
    {
      "epoch": 0.8460236886632826,
      "grad_norm": 0.520793616771698,
      "learning_rate": 1.4359842075578116e-05,
      "loss": 0.1237,
      "step": 13500
    },
    {
      "epoch": 0.8773578993545152,
      "grad_norm": 0.08138167858123779,
      "learning_rate": 1.4150947337636568e-05,
      "loss": 0.1042,
      "step": 14000
    },
    {
      "epoch": 0.9086921100457479,
      "grad_norm": 0.2681790590286255,
      "learning_rate": 1.3942052599695016e-05,
      "loss": 0.1223,
      "step": 14500
    },
    {
      "epoch": 0.9400263207369807,
      "grad_norm": 0.027624983340501785,
      "learning_rate": 1.3733157861753463e-05,
      "loss": 0.1122,
      "step": 15000
    },
    {
      "epoch": 0.9713605314282133,
      "grad_norm": 18.958938598632812,
      "learning_rate": 1.3524263123811911e-05,
      "loss": 0.1247,
      "step": 15500
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9712674291085697,
      "eval_f1": 0.8566067240031274,
      "eval_loss": 0.11290808767080307,
      "eval_runtime": 139.1424,
      "eval_samples_per_second": 229.369,
      "eval_steps_per_second": 28.676,
      "step": 15957
    },
    {
      "epoch": 1.002694742119446,
      "grad_norm": 0.0361429825425148,
      "learning_rate": 1.3315368385870362e-05,
      "loss": 0.1112,
      "step": 16000
    },
    {
      "epoch": 1.0340289528106787,
      "grad_norm": 0.10898561030626297,
      "learning_rate": 1.310647364792881e-05,
      "loss": 0.065,
      "step": 16500
    },
    {
      "epoch": 1.0653631635019114,
      "grad_norm": 1.1375638246536255,
      "learning_rate": 1.2897578909987258e-05,
      "loss": 0.0781,
      "step": 17000
    },
    {
      "epoch": 1.096697374193144,
      "grad_norm": 48.0145378112793,
      "learning_rate": 1.2688684172045706e-05,
      "loss": 0.0776,
      "step": 17500
    },
    {
      "epoch": 1.1280315848843767,
      "grad_norm": 0.10827700048685074,
      "learning_rate": 1.2479789434104157e-05,
      "loss": 0.0798,
      "step": 18000
    },
    {
      "epoch": 1.1593657955756094,
      "grad_norm": 35.41124725341797,
      "learning_rate": 1.2270894696162605e-05,
      "loss": 0.0716,
      "step": 18500
    },
    {
      "epoch": 1.1907000062668422,
      "grad_norm": 0.01607622765004635,
      "learning_rate": 1.2061999958221053e-05,
      "loss": 0.074,
      "step": 19000
    },
    {
      "epoch": 1.222034216958075,
      "grad_norm": 0.016460325568914413,
      "learning_rate": 1.1853105220279502e-05,
      "loss": 0.084,
      "step": 19500
    },
    {
      "epoch": 1.2533684276493076,
      "grad_norm": 0.025399167090654373,
      "learning_rate": 1.164421048233795e-05,
      "loss": 0.0841,
      "step": 20000
    },
    {
      "epoch": 1.2847026383405402,
      "grad_norm": 13.243876457214355,
      "learning_rate": 1.14353157443964e-05,
      "loss": 0.0867,
      "step": 20500
    },
    {
      "epoch": 1.3160368490317729,
      "grad_norm": 14.900182723999023,
      "learning_rate": 1.1226421006454849e-05,
      "loss": 0.077,
      "step": 21000
    },
    {
      "epoch": 1.3473710597230055,
      "grad_norm": 0.02405312843620777,
      "learning_rate": 1.1017526268513297e-05,
      "loss": 0.0791,
      "step": 21500
    },
    {
      "epoch": 1.3787052704142382,
      "grad_norm": 0.5990450978279114,
      "learning_rate": 1.0808631530571745e-05,
      "loss": 0.0808,
      "step": 22000
    },
    {
      "epoch": 1.4100394811054708,
      "grad_norm": 0.013339116238057613,
      "learning_rate": 1.0599736792630196e-05,
      "loss": 0.0848,
      "step": 22500
    },
    {
      "epoch": 1.4413736917967037,
      "grad_norm": 0.023200716823339462,
      "learning_rate": 1.0390842054688644e-05,
      "loss": 0.0714,
      "step": 23000
    },
    {
      "epoch": 1.4727079024879364,
      "grad_norm": 0.007010191213339567,
      "learning_rate": 1.0181947316747092e-05,
      "loss": 0.0797,
      "step": 23500
    },
    {
      "epoch": 1.504042113179169,
      "grad_norm": 29.841516494750977,
      "learning_rate": 9.973052578805541e-06,
      "loss": 0.0864,
      "step": 24000
    },
    {
      "epoch": 1.5353763238704017,
      "grad_norm": 0.004155378323048353,
      "learning_rate": 9.764157840863989e-06,
      "loss": 0.0726,
      "step": 24500
    },
    {
      "epoch": 1.5667105345616343,
      "grad_norm": 18.432579040527344,
      "learning_rate": 9.555263102922438e-06,
      "loss": 0.0801,
      "step": 25000
    },
    {
      "epoch": 1.5980447452528672,
      "grad_norm": 0.5066466927528381,
      "learning_rate": 9.346368364980886e-06,
      "loss": 0.0746,
      "step": 25500
    },
    {
      "epoch": 1.6293789559440999,
      "grad_norm": 0.6675248146057129,
      "learning_rate": 9.137473627039336e-06,
      "loss": 0.0785,
      "step": 26000
    },
    {
      "epoch": 1.6607131666353325,
      "grad_norm": 0.062432803213596344,
      "learning_rate": 8.928578889097784e-06,
      "loss": 0.0855,
      "step": 26500
    },
    {
      "epoch": 1.6920473773265652,
      "grad_norm": 0.005082368850708008,
      "learning_rate": 8.719684151156233e-06,
      "loss": 0.0885,
      "step": 27000
    },
    {
      "epoch": 1.7233815880177978,
      "grad_norm": 0.010900082997977734,
      "learning_rate": 8.510789413214683e-06,
      "loss": 0.0945,
      "step": 27500
    },
    {
      "epoch": 1.7547157987090305,
      "grad_norm": 11.707688331604004,
      "learning_rate": 8.30189467527313e-06,
      "loss": 0.0722,
      "step": 28000
    },
    {
      "epoch": 1.7860500094002631,
      "grad_norm": 0.0018975611310452223,
      "learning_rate": 8.09299993733158e-06,
      "loss": 0.0761,
      "step": 28500
    },
    {
      "epoch": 1.8173842200914958,
      "grad_norm": 0.029523102566599846,
      "learning_rate": 7.884105199390028e-06,
      "loss": 0.0732,
      "step": 29000
    },
    {
      "epoch": 1.8487184307827285,
      "grad_norm": 0.03987930715084076,
      "learning_rate": 7.675210461448477e-06,
      "loss": 0.0576,
      "step": 29500
    },
    {
      "epoch": 1.8800526414739611,
      "grad_norm": 0.03522760421037674,
      "learning_rate": 7.466315723506925e-06,
      "loss": 0.071,
      "step": 30000
    },
    {
      "epoch": 1.911386852165194,
      "grad_norm": 0.1319192349910736,
      "learning_rate": 7.2574209855653745e-06,
      "loss": 0.0863,
      "step": 30500
    },
    {
      "epoch": 1.9427210628564267,
      "grad_norm": 0.02413085289299488,
      "learning_rate": 7.048526247623822e-06,
      "loss": 0.0718,
      "step": 31000
    },
    {
      "epoch": 1.9740552735476593,
      "grad_norm": 0.009825543500483036,
      "learning_rate": 6.839631509682272e-06,
      "loss": 0.0824,
      "step": 31500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9668807770640765,
      "eval_f1": 0.8449464573859469,
      "eval_loss": 0.14600731432437897,
      "eval_runtime": 139.0182,
      "eval_samples_per_second": 229.574,
      "eval_steps_per_second": 28.701,
      "step": 31914
    },
    {
      "epoch": 2.005389484238892,
      "grad_norm": 0.09895268082618713,
      "learning_rate": 6.63073677174072e-06,
      "loss": 0.0756,
      "step": 32000
    },
    {
      "epoch": 2.036723694930125,
      "grad_norm": 0.04020028933882713,
      "learning_rate": 6.421842033799169e-06,
      "loss": 0.0425,
      "step": 32500
    },
    {
      "epoch": 2.0680579056213575,
      "grad_norm": 0.0006764501449652016,
      "learning_rate": 6.212947295857618e-06,
      "loss": 0.0371,
      "step": 33000
    },
    {
      "epoch": 2.09939211631259,
      "grad_norm": 0.07980203628540039,
      "learning_rate": 6.0040525579160665e-06,
      "loss": 0.0395,
      "step": 33500
    },
    {
      "epoch": 2.130726327003823,
      "grad_norm": 9.280925750732422,
      "learning_rate": 5.795157819974515e-06,
      "loss": 0.0437,
      "step": 34000
    },
    {
      "epoch": 2.1620605376950555,
      "grad_norm": 0.052823714911937714,
      "learning_rate": 5.586263082032965e-06,
      "loss": 0.0433,
      "step": 34500
    },
    {
      "epoch": 2.193394748386288,
      "grad_norm": 18.749326705932617,
      "learning_rate": 5.3773683440914125e-06,
      "loss": 0.0365,
      "step": 35000
    },
    {
      "epoch": 2.2247289590775208,
      "grad_norm": 0.05787142366170883,
      "learning_rate": 5.168473606149862e-06,
      "loss": 0.0323,
      "step": 35500
    },
    {
      "epoch": 2.2560631697687534,
      "grad_norm": 0.08446703106164932,
      "learning_rate": 4.95957886820831e-06,
      "loss": 0.0488,
      "step": 36000
    },
    {
      "epoch": 2.287397380459986,
      "grad_norm": 0.0023540444672107697,
      "learning_rate": 4.7506841302667585e-06,
      "loss": 0.038,
      "step": 36500
    },
    {
      "epoch": 2.3187315911512187,
      "grad_norm": 0.020784469321370125,
      "learning_rate": 4.541789392325208e-06,
      "loss": 0.0396,
      "step": 37000
    },
    {
      "epoch": 2.3500658018424514,
      "grad_norm": 0.00681657437235117,
      "learning_rate": 4.332894654383657e-06,
      "loss": 0.0352,
      "step": 37500
    },
    {
      "epoch": 2.3814000125336845,
      "grad_norm": 1.1381922960281372,
      "learning_rate": 4.123999916442105e-06,
      "loss": 0.0377,
      "step": 38000
    },
    {
      "epoch": 2.412734223224917,
      "grad_norm": 0.003584115067496896,
      "learning_rate": 3.915105178500554e-06,
      "loss": 0.0452,
      "step": 38500
    },
    {
      "epoch": 2.44406843391615,
      "grad_norm": 0.0009662588709034026,
      "learning_rate": 3.7062104405590027e-06,
      "loss": 0.0383,
      "step": 39000
    },
    {
      "epoch": 2.4754026446073825,
      "grad_norm": 0.02164757065474987,
      "learning_rate": 3.4973157026174514e-06,
      "loss": 0.0385,
      "step": 39500
    },
    {
      "epoch": 2.506736855298615,
      "grad_norm": 0.10245607048273087,
      "learning_rate": 3.2884209646759e-06,
      "loss": 0.0322,
      "step": 40000
    },
    {
      "epoch": 2.5380710659898478,
      "grad_norm": 0.0011107695754617453,
      "learning_rate": 3.079526226734349e-06,
      "loss": 0.0415,
      "step": 40500
    },
    {
      "epoch": 2.5694052766810804,
      "grad_norm": 0.06932328641414642,
      "learning_rate": 2.870631488792798e-06,
      "loss": 0.0394,
      "step": 41000
    },
    {
      "epoch": 2.600739487372313,
      "grad_norm": 0.10022483021020889,
      "learning_rate": 2.6617367508512465e-06,
      "loss": 0.0274,
      "step": 41500
    },
    {
      "epoch": 2.6320736980635457,
      "grad_norm": 22.11353874206543,
      "learning_rate": 2.452842012909695e-06,
      "loss": 0.032,
      "step": 42000
    },
    {
      "epoch": 2.6634079087547784,
      "grad_norm": 0.001642558490857482,
      "learning_rate": 2.243947274968144e-06,
      "loss": 0.0399,
      "step": 42500
    },
    {
      "epoch": 2.694742119446011,
      "grad_norm": 0.024312496185302734,
      "learning_rate": 2.0350525370265925e-06,
      "loss": 0.0388,
      "step": 43000
    },
    {
      "epoch": 2.7260763301372437,
      "grad_norm": 0.0015839878469705582,
      "learning_rate": 1.8261577990850414e-06,
      "loss": 0.0387,
      "step": 43500
    },
    {
      "epoch": 2.7574105408284764,
      "grad_norm": 0.006028833799064159,
      "learning_rate": 1.61726306114349e-06,
      "loss": 0.0438,
      "step": 44000
    },
    {
      "epoch": 2.788744751519709,
      "grad_norm": 0.002367444569244981,
      "learning_rate": 1.4083683232019387e-06,
      "loss": 0.0263,
      "step": 44500
    },
    {
      "epoch": 2.8200789622109417,
      "grad_norm": 10.050576210021973,
      "learning_rate": 1.1994735852603874e-06,
      "loss": 0.0459,
      "step": 45000
    },
    {
      "epoch": 2.8514131729021743,
      "grad_norm": 0.06076907739043236,
      "learning_rate": 9.905788473188362e-07,
      "loss": 0.0451,
      "step": 45500
    },
    {
      "epoch": 2.8827473835934074,
      "grad_norm": 0.0017307106172665954,
      "learning_rate": 7.816841093772848e-07,
      "loss": 0.0358,
      "step": 46000
    },
    {
      "epoch": 2.91408159428464,
      "grad_norm": 0.1369352489709854,
      "learning_rate": 5.727893714357336e-07,
      "loss": 0.0304,
      "step": 46500
    },
    {
      "epoch": 2.9454158049758727,
      "grad_norm": 0.030802803114056587,
      "learning_rate": 3.638946334941823e-07,
      "loss": 0.0262,
      "step": 47000
    },
    {
      "epoch": 2.9767500156671054,
      "grad_norm": 0.0022679055109620094,
      "learning_rate": 1.5499989555263104e-07,
      "loss": 0.0346,
      "step": 47500
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9687294375685415,
      "eval_f1": 0.8493357487922706,
      "eval_loss": 0.18749012053012848,
      "eval_runtime": 137.5346,
      "eval_samples_per_second": 232.051,
      "eval_steps_per_second": 29.011,
      "step": 47871
    }
  ],
  "logging_steps": 500,
  "max_steps": 47871,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 1.0076311464910848e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}