{
  "best_metric": 2.690006971359253,
  "best_model_checkpoint": "models/opt-babylm2-rewritten-clean-spacy_no-num-adj-earlystop-bpe_seed-42_1e-3/checkpoint-44480",
  "epoch": 19.99134539732494,
  "eval_steps": 500,
  "global_step": 44480,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.44958974935371476,
      "grad_norm": 0.5849416255950928,
      "learning_rate": 3.125e-05,
      "loss": 5.5771,
      "step": 1000
    },
    {
      "epoch": 0.8991794987074295,
      "grad_norm": 0.6677811145782471,
      "learning_rate": 6.25e-05,
      "loss": 4.0896,
      "step": 2000
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.35931850789096126,
      "eval_loss": 3.830660104751587,
      "eval_runtime": 73.9335,
      "eval_samples_per_second": 820.738,
      "eval_steps_per_second": 12.836,
      "step": 2225
    },
    {
      "epoch": 1.348432055749129,
      "grad_norm": 0.5427906513214111,
      "learning_rate": 9.375e-05,
      "loss": 3.6885,
      "step": 3000
    },
    {
      "epoch": 1.7980218051028438,
      "grad_norm": 0.5098850727081299,
      "learning_rate": 0.000125,
      "loss": 3.4325,
      "step": 4000
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.40814210192976336,
      "eval_loss": 3.3144962787628174,
      "eval_runtime": 73.7109,
      "eval_samples_per_second": 823.216,
      "eval_steps_per_second": 12.875,
      "step": 4450
    },
    {
      "epoch": 2.2472743621445432,
      "grad_norm": 0.47539061307907104,
      "learning_rate": 0.00015625,
      "loss": 3.2344,
      "step": 5000
    },
    {
      "epoch": 2.696864111498258,
      "grad_norm": 0.4557252824306488,
      "learning_rate": 0.0001875,
      "loss": 3.1208,
      "step": 6000
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.42951116108935333,
      "eval_loss": 3.1043691635131836,
      "eval_runtime": 73.4454,
      "eval_samples_per_second": 826.192,
      "eval_steps_per_second": 12.921,
      "step": 6675
    },
    {
      "epoch": 3.1461166685399573,
      "grad_norm": 0.4797649085521698,
      "learning_rate": 0.00021875,
      "loss": 3.0251,
      "step": 7000
    },
    {
      "epoch": 3.595706417893672,
      "grad_norm": 0.39974284172058105,
      "learning_rate": 0.00025,
      "loss": 2.957,
      "step": 8000
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.43958987682086675,
      "eval_loss": 2.997297763824463,
      "eval_runtime": 72.7828,
      "eval_samples_per_second": 833.713,
      "eval_steps_per_second": 13.039,
      "step": 8900
    },
    {
      "epoch": 4.044958974935372,
      "grad_norm": 0.37127092480659485,
      "learning_rate": 0.00028125000000000003,
      "loss": 2.91,
      "step": 9000
    },
    {
      "epoch": 4.4945487242890865,
      "grad_norm": 0.35168108344078064,
      "learning_rate": 0.0003125,
      "loss": 2.8553,
      "step": 10000
    },
    {
      "epoch": 4.944138473642801,
      "grad_norm": 0.33759039640426636,
      "learning_rate": 0.00034375,
      "loss": 2.8381,
      "step": 11000
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.4463711272247858,
      "eval_loss": 2.9338483810424805,
      "eval_runtime": 73.014,
      "eval_samples_per_second": 831.073,
      "eval_steps_per_second": 12.998,
      "step": 11125
    },
    {
      "epoch": 5.3933910306845005,
      "grad_norm": 0.3382190465927124,
      "learning_rate": 0.000375,
      "loss": 2.7866,
      "step": 12000
    },
    {
      "epoch": 5.842980780038215,
      "grad_norm": 0.30114060640335083,
      "learning_rate": 0.00040625000000000004,
      "loss": 2.7819,
      "step": 13000
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.4507991133170473,
      "eval_loss": 2.8903539180755615,
      "eval_runtime": 72.6589,
      "eval_samples_per_second": 835.135,
      "eval_steps_per_second": 13.061,
      "step": 13350
    },
    {
      "epoch": 6.292233337079915,
      "grad_norm": 0.29889941215515137,
      "learning_rate": 0.0004375,
      "loss": 2.7435,
      "step": 14000
    },
    {
      "epoch": 6.741823086433629,
      "grad_norm": 0.2726060450077057,
      "learning_rate": 0.00046871875,
      "loss": 2.7385,
      "step": 15000
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.4530601548463815,
      "eval_loss": 2.8666210174560547,
      "eval_runtime": 72.57,
      "eval_samples_per_second": 836.158,
      "eval_steps_per_second": 13.077,
      "step": 15575
    },
    {
      "epoch": 7.191075643475329,
      "grad_norm": 0.2712298333644867,
      "learning_rate": 0.00049996875,
      "loss": 2.7141,
      "step": 16000
    },
    {
      "epoch": 7.640665392829043,
      "grad_norm": 0.2567969560623169,
      "learning_rate": 0.00053121875,
      "loss": 2.7061,
      "step": 17000
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.4559095609239081,
      "eval_loss": 2.8456263542175293,
      "eval_runtime": 72.544,
      "eval_samples_per_second": 836.458,
      "eval_steps_per_second": 13.082,
      "step": 17800
    },
    {
      "epoch": 8.089917949870744,
      "grad_norm": 0.24305634200572968,
      "learning_rate": 0.0005624687499999999,
      "loss": 2.6951,
      "step": 18000
    },
    {
      "epoch": 8.539507699224458,
      "grad_norm": 0.23454323410987854,
      "learning_rate": 0.0005936875,
      "loss": 2.6745,
      "step": 19000
    },
    {
      "epoch": 8.989097448578173,
      "grad_norm": 0.2262556552886963,
      "learning_rate": 0.0006249375000000001,
      "loss": 2.6855,
      "step": 20000
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.4575118590613569,
      "eval_loss": 2.833211898803711,
      "eval_runtime": 72.5848,
      "eval_samples_per_second": 835.988,
      "eval_steps_per_second": 13.074,
      "step": 20025
    },
    {
      "epoch": 9.438350005619872,
      "grad_norm": 0.2205825001001358,
      "learning_rate": 0.0006561562500000001,
      "loss": 2.6456,
      "step": 21000
    },
    {
      "epoch": 9.887939754973587,
      "grad_norm": 0.21933791041374207,
      "learning_rate": 0.00068740625,
      "loss": 2.6669,
      "step": 22000
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.45864309072343507,
      "eval_loss": 2.819795608520508,
      "eval_runtime": 72.9477,
      "eval_samples_per_second": 831.828,
      "eval_steps_per_second": 13.009,
      "step": 22250
    },
    {
      "epoch": 10.337192312015286,
      "grad_norm": 0.21199771761894226,
      "learning_rate": 0.000718625,
      "loss": 2.636,
      "step": 23000
    },
    {
      "epoch": 10.786782061369001,
      "grad_norm": 0.20166124403476715,
      "learning_rate": 0.000749875,
      "loss": 2.6499,
      "step": 24000
    },
    {
      "epoch": 11.0,
      "eval_accuracy": 0.4597305698812155,
      "eval_loss": 2.8118433952331543,
      "eval_runtime": 73.1507,
      "eval_samples_per_second": 829.521,
      "eval_steps_per_second": 12.973,
      "step": 24475
    },
    {
      "epoch": 11.2360346184107,
      "grad_norm": 0.20014500617980957,
      "learning_rate": 0.000781125,
      "loss": 2.6298,
      "step": 25000
    },
    {
      "epoch": 11.685624367764415,
      "grad_norm": 0.19325494766235352,
      "learning_rate": 0.00081234375,
      "loss": 2.6351,
      "step": 26000
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.4601314513940052,
      "eval_loss": 2.807219982147217,
      "eval_runtime": 72.6854,
      "eval_samples_per_second": 834.831,
      "eval_steps_per_second": 13.056,
      "step": 26700
    },
    {
      "epoch": 12.134876924806115,
      "grad_norm": 0.19165903329849243,
      "learning_rate": 0.00084359375,
      "loss": 2.6265,
      "step": 27000
    },
    {
      "epoch": 12.58446667415983,
      "grad_norm": 0.1863769292831421,
      "learning_rate": 0.0008748437500000001,
      "loss": 2.6204,
      "step": 28000
    },
    {
      "epoch": 13.0,
      "eval_accuracy": 0.4611660010081818,
      "eval_loss": 2.802619218826294,
      "eval_runtime": 72.6835,
      "eval_samples_per_second": 834.852,
      "eval_steps_per_second": 13.057,
      "step": 28925
    },
    {
      "epoch": 13.033719231201529,
      "grad_norm": 0.19991189241409302,
      "learning_rate": 0.00090609375,
      "loss": 2.6286,
      "step": 29000
    },
    {
      "epoch": 13.483308980555243,
      "grad_norm": 0.18545052409172058,
      "learning_rate": 0.0009373125,
      "loss": 2.6068,
      "step": 30000
    },
    {
      "epoch": 13.932898729908958,
      "grad_norm": 0.17478196322917938,
      "learning_rate": 0.0009685625,
      "loss": 2.6277,
      "step": 31000
    },
    {
      "epoch": 14.0,
      "eval_accuracy": 0.4612847208758256,
      "eval_loss": 2.801252841949463,
      "eval_runtime": 72.4155,
      "eval_samples_per_second": 837.942,
      "eval_steps_per_second": 13.105,
      "step": 31150
    },
    {
      "epoch": 14.382151286950657,
      "grad_norm": 0.17678463459014893,
      "learning_rate": 0.00099978125,
      "loss": 2.5975,
      "step": 32000
    },
    {
      "epoch": 14.831741036304372,
      "grad_norm": 0.17033128440380096,
      "learning_rate": 0.0009204326923076923,
      "loss": 2.6136,
      "step": 33000
    },
    {
      "epoch": 15.0,
      "eval_accuracy": 0.46383548541367764,
      "eval_loss": 2.779118061065674,
      "eval_runtime": 72.723,
      "eval_samples_per_second": 834.399,
      "eval_steps_per_second": 13.05,
      "step": 33375
    },
    {
      "epoch": 15.280993593346071,
      "grad_norm": 0.1788545548915863,
      "learning_rate": 0.0008403044871794871,
      "loss": 2.5726,
      "step": 34000
    },
    {
      "epoch": 15.730583342699786,
      "grad_norm": 0.17410264909267426,
      "learning_rate": 0.0007602564102564103,
      "loss": 2.5687,
      "step": 35000
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.4676111908177905,
      "eval_loss": 2.75136661529541,
      "eval_runtime": 72.8808,
      "eval_samples_per_second": 832.592,
      "eval_steps_per_second": 13.021,
      "step": 35600
    },
    {
      "epoch": 16.179835899741487,
      "grad_norm": 0.17518466711044312,
      "learning_rate": 0.0006801282051282051,
      "loss": 2.5356,
      "step": 36000
    },
    {
      "epoch": 16.6294256490952,
      "grad_norm": 0.17421123385429382,
      "learning_rate": 0.0006000801282051283,
      "loss": 2.5184,
      "step": 37000
    },
    {
      "epoch": 17.0,
      "eval_accuracy": 0.4707787558002766,
      "eval_loss": 2.728271245956421,
      "eval_runtime": 72.8051,
      "eval_samples_per_second": 833.458,
      "eval_steps_per_second": 13.035,
      "step": 37825
    },
    {
      "epoch": 17.0786782061369,
      "grad_norm": 0.18264968693256378,
      "learning_rate": 0.0005199519230769231,
      "loss": 2.4989,
      "step": 38000
    },
    {
      "epoch": 17.528267955490616,
      "grad_norm": 0.18594865500926971,
      "learning_rate": 0.00043990384615384616,
      "loss": 2.4571,
      "step": 39000
    },
    {
      "epoch": 17.97785770484433,
      "grad_norm": 0.17989173531532288,
      "learning_rate": 0.00035977564102564105,
      "loss": 2.4613,
      "step": 40000
    },
    {
      "epoch": 18.0,
      "eval_accuracy": 0.47402006023239884,
      "eval_loss": 2.705965518951416,
      "eval_runtime": 72.7008,
      "eval_samples_per_second": 834.654,
      "eval_steps_per_second": 13.054,
      "step": 40050
    },
    {
      "epoch": 18.42711026188603,
      "grad_norm": 0.19071288406848907,
      "learning_rate": 0.0002797275641025641,
      "loss": 2.3913,
      "step": 41000
    },
    {
      "epoch": 18.876700011239745,
      "grad_norm": 0.19211626052856445,
      "learning_rate": 0.00019959935897435898,
      "loss": 2.3966,
      "step": 42000
    },
    {
      "epoch": 19.0,
      "eval_accuracy": 0.4765565421949927,
      "eval_loss": 2.6947293281555176,
      "eval_runtime": 72.857,
      "eval_samples_per_second": 832.864,
      "eval_steps_per_second": 13.026,
      "step": 42275
    },
    {
      "epoch": 19.325952568281444,
      "grad_norm": 0.19902721047401428,
      "learning_rate": 0.00011947115384615386,
      "loss": 2.3404,
      "step": 43000
    },
    {
      "epoch": 19.775542317635157,
      "grad_norm": 0.20530302822589874,
      "learning_rate": 3.942307692307692e-05,
      "loss": 2.3227,
      "step": 44000
    },
    {
      "epoch": 19.99134539732494,
      "eval_accuracy": 0.4781093360218181,
      "eval_loss": 2.690006971359253,
      "eval_runtime": 72.887,
      "eval_samples_per_second": 832.521,
      "eval_steps_per_second": 13.02,
      "step": 44480
    },
    {
      "epoch": 19.99134539732494,
      "step": 44480,
      "total_flos": 1.487139158163456e+18,
      "train_loss": 2.8016022716494775,
      "train_runtime": 30047.0507,
      "train_samples_per_second": 379.002,
      "train_steps_per_second": 1.48
    }
  ],
  "logging_steps": 1000,
  "max_steps": 44480,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 20,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 3,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.487139158163456e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}