{
  "best_metric": 2.9599573612213135,
  "best_model_checkpoint": "models/opt-babylm2-rewritten-clean-spacy-32k-earlystop-40epochs_seed-42_1e-3/checkpoint-48293",
  "epoch": 28.0,
  "eval_steps": 500,
  "global_step": 54089,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.5176653293645658,
      "grad_norm": 0.516268789768219,
      "learning_rate": 3.125e-05,
      "loss": 5.9216,
      "step": 1000
    },
    {
      "epoch": 0.9996117510029766,
      "eval_accuracy": 0.32528310960329715,
      "eval_loss": 4.013359069824219,
      "eval_runtime": 112.6568,
      "eval_samples_per_second": 463.265,
      "eval_steps_per_second": 7.243,
      "step": 1931
    },
    {
      "epoch": 1.0353306587291315,
      "grad_norm": 0.6370756030082703,
      "learning_rate": 6.25e-05,
      "loss": 4.1987,
      "step": 2000
    },
    {
      "epoch": 1.5529959880936974,
      "grad_norm": 0.6017232537269592,
      "learning_rate": 9.375e-05,
      "loss": 3.7977,
      "step": 3000
    },
    {
      "epoch": 1.9997411673353178,
      "eval_accuracy": 0.3639096213308086,
      "eval_loss": 3.544811725616455,
      "eval_runtime": 112.8804,
      "eval_samples_per_second": 462.348,
      "eval_steps_per_second": 7.229,
      "step": 3863
    },
    {
      "epoch": 2.070661317458263,
      "grad_norm": 0.5702515840530396,
      "learning_rate": 0.000125,
      "loss": 3.5582,
      "step": 4000
    },
    {
      "epoch": 2.588326646822829,
      "grad_norm": 0.48817870020866394,
      "learning_rate": 0.00015625,
      "loss": 3.3887,
      "step": 5000
    },
    {
      "epoch": 2.9998705836676587,
      "eval_accuracy": 0.3840780857274889,
      "eval_loss": 3.324249744415283,
      "eval_runtime": 112.7528,
      "eval_samples_per_second": 462.871,
      "eval_steps_per_second": 7.237,
      "step": 5795
    },
    {
      "epoch": 3.105991976187395,
      "grad_norm": 0.45357462763786316,
      "learning_rate": 0.0001875,
      "loss": 3.2719,
      "step": 6000
    },
    {
      "epoch": 3.6236573055519608,
      "grad_norm": 0.42539018392562866,
      "learning_rate": 0.00021875,
      "loss": 3.1805,
      "step": 7000
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.3949032381682315,
      "eval_loss": 3.2081618309020996,
      "eval_runtime": 112.6095,
      "eval_samples_per_second": 463.46,
      "eval_steps_per_second": 7.246,
      "step": 7727
    },
    {
      "epoch": 4.141322634916526,
      "grad_norm": 0.41741499304771423,
      "learning_rate": 0.00025,
      "loss": 3.1173,
      "step": 8000
    },
    {
      "epoch": 4.658987964281092,
      "grad_norm": 0.3810145854949951,
      "learning_rate": 0.00028125000000000003,
      "loss": 3.0632,
      "step": 9000
    },
    {
      "epoch": 4.999611751002977,
      "eval_accuracy": 0.401180377880219,
      "eval_loss": 3.143218517303467,
      "eval_runtime": 112.4876,
      "eval_samples_per_second": 463.962,
      "eval_steps_per_second": 7.254,
      "step": 9658
    },
    {
      "epoch": 5.176653293645658,
      "grad_norm": 0.3555419445037842,
      "learning_rate": 0.0003125,
      "loss": 3.0212,
      "step": 10000
    },
    {
      "epoch": 5.694318623010224,
      "grad_norm": 0.3318658173084259,
      "learning_rate": 0.00034375,
      "loss": 2.9865,
      "step": 11000
    },
    {
      "epoch": 5.999741167335317,
      "eval_accuracy": 0.4055885546400971,
      "eval_loss": 3.101013422012329,
      "eval_runtime": 112.8154,
      "eval_samples_per_second": 462.614,
      "eval_steps_per_second": 7.233,
      "step": 11590
    },
    {
      "epoch": 6.21198395237479,
      "grad_norm": 0.3243854343891144,
      "learning_rate": 0.000375,
      "loss": 2.9568,
      "step": 12000
    },
    {
      "epoch": 6.729649281739356,
      "grad_norm": 0.3086845874786377,
      "learning_rate": 0.00040625000000000004,
      "loss": 2.9347,
      "step": 13000
    },
    {
      "epoch": 6.999870583667659,
      "eval_accuracy": 0.4087078510269791,
      "eval_loss": 3.071547746658325,
      "eval_runtime": 113.6377,
      "eval_samples_per_second": 459.267,
      "eval_steps_per_second": 7.181,
      "step": 13522
    },
    {
      "epoch": 7.2473146111039215,
      "grad_norm": 0.29632025957107544,
      "learning_rate": 0.0004375,
      "loss": 2.9084,
      "step": 14000
    },
    {
      "epoch": 7.764979940468487,
      "grad_norm": 0.28605663776397705,
      "learning_rate": 0.0004686875,
      "loss": 2.8953,
      "step": 15000
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.4107785654978604,
      "eval_loss": 3.053938388824463,
      "eval_runtime": 112.9423,
      "eval_samples_per_second": 462.094,
      "eval_steps_per_second": 7.225,
      "step": 15454
    },
    {
      "epoch": 8.282645269833052,
      "grad_norm": 0.2786637246608734,
      "learning_rate": 0.0004999375,
      "loss": 2.8698,
      "step": 16000
    },
    {
      "epoch": 8.80031059919762,
      "grad_norm": 0.2667602002620697,
      "learning_rate": 0.00053115625,
      "loss": 2.8689,
      "step": 17000
    },
    {
      "epoch": 8.999611751002977,
      "eval_accuracy": 0.4122456033572655,
      "eval_loss": 3.039193868637085,
      "eval_runtime": 112.7699,
      "eval_samples_per_second": 462.801,
      "eval_steps_per_second": 7.236,
      "step": 17385
    },
    {
      "epoch": 9.317975928562184,
      "grad_norm": 0.25813835859298706,
      "learning_rate": 0.00056240625,
      "loss": 2.8401,
      "step": 18000
    },
    {
      "epoch": 9.835641257926751,
      "grad_norm": 0.2392367571592331,
      "learning_rate": 0.00059365625,
      "loss": 2.8456,
      "step": 19000
    },
    {
      "epoch": 9.999741167335317,
      "eval_accuracy": 0.4133619617611367,
      "eval_loss": 3.0309925079345703,
      "eval_runtime": 112.8237,
      "eval_samples_per_second": 462.58,
      "eval_steps_per_second": 7.233,
      "step": 19317
    },
    {
      "epoch": 10.353306587291316,
      "grad_norm": 0.2465026080608368,
      "learning_rate": 0.00062490625,
      "loss": 2.8163,
      "step": 20000
    },
    {
      "epoch": 10.870971916655883,
      "grad_norm": 0.21547040343284607,
      "learning_rate": 0.000656125,
      "loss": 2.8298,
      "step": 21000
    },
    {
      "epoch": 10.99987058366766,
      "eval_accuracy": 0.41438394403555634,
      "eval_loss": 3.0251340866088867,
      "eval_runtime": 112.6781,
      "eval_samples_per_second": 463.178,
      "eval_steps_per_second": 7.242,
      "step": 21249
    },
    {
      "epoch": 11.388637246020448,
      "grad_norm": 0.23142270743846893,
      "learning_rate": 0.0006873749999999999,
      "loss": 2.798,
      "step": 22000
    },
    {
      "epoch": 11.906302575385013,
      "grad_norm": 0.2184012234210968,
      "learning_rate": 0.00071859375,
      "loss": 2.817,
      "step": 23000
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.4152235609706615,
      "eval_loss": 3.0175206661224365,
      "eval_runtime": 112.4939,
      "eval_samples_per_second": 463.936,
      "eval_steps_per_second": 7.254,
      "step": 23181
    },
    {
      "epoch": 12.42396790474958,
      "grad_norm": 0.211518794298172,
      "learning_rate": 0.0007498437500000001,
      "loss": 2.7828,
      "step": 24000
    },
    {
      "epoch": 12.941633234114144,
      "grad_norm": 0.2092377245426178,
      "learning_rate": 0.00078109375,
      "loss": 2.8069,
      "step": 25000
    },
    {
      "epoch": 12.999611751002977,
      "eval_accuracy": 0.41580334298885296,
      "eval_loss": 3.0118961334228516,
      "eval_runtime": 112.5807,
      "eval_samples_per_second": 463.578,
      "eval_steps_per_second": 7.248,
      "step": 25112
    },
    {
      "epoch": 13.459298563478711,
      "grad_norm": 0.20688970386981964,
      "learning_rate": 0.00081234375,
      "loss": 2.7707,
      "step": 26000
    },
    {
      "epoch": 13.976963892843276,
      "grad_norm": 0.20183704793453217,
      "learning_rate": 0.00084353125,
      "loss": 2.7996,
      "step": 27000
    },
    {
      "epoch": 13.999741167335317,
      "eval_accuracy": 0.4162821365373128,
      "eval_loss": 3.005990743637085,
      "eval_runtime": 112.5919,
      "eval_samples_per_second": 463.533,
      "eval_steps_per_second": 7.247,
      "step": 27044
    },
    {
      "epoch": 14.494629222207843,
      "grad_norm": 0.19293886423110962,
      "learning_rate": 0.00087478125,
      "loss": 2.7615,
      "step": 28000
    },
    {
      "epoch": 14.99987058366766,
      "eval_accuracy": 0.4170801257847458,
      "eval_loss": 3.0038492679595947,
      "eval_runtime": 112.6692,
      "eval_samples_per_second": 463.214,
      "eval_steps_per_second": 7.242,
      "step": 28976
    },
    {
      "epoch": 15.012294551572408,
      "grad_norm": 0.1958475559949875,
      "learning_rate": 0.0009060312499999999,
      "loss": 2.7934,
      "step": 29000
    },
    {
      "epoch": 15.529959880936975,
      "grad_norm": 0.18981003761291504,
      "learning_rate": 0.00093728125,
      "loss": 2.7575,
      "step": 30000
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.4168810041740398,
      "eval_loss": 3.0022430419921875,
      "eval_runtime": 112.632,
      "eval_samples_per_second": 463.367,
      "eval_steps_per_second": 7.245,
      "step": 30908
    },
    {
      "epoch": 16.04762521030154,
      "grad_norm": 0.21319861710071564,
      "learning_rate": 0.00096853125,
      "loss": 2.7826,
      "step": 31000
    },
    {
      "epoch": 16.565290539666105,
      "grad_norm": 0.188828244805336,
      "learning_rate": 0.00099975,
      "loss": 2.7573,
      "step": 32000
    },
    {
      "epoch": 16.999611751002977,
      "eval_accuracy": 0.4178921662552739,
      "eval_loss": 2.9961841106414795,
      "eval_runtime": 112.6097,
      "eval_samples_per_second": 463.459,
      "eval_steps_per_second": 7.246,
      "step": 32839
    },
    {
      "epoch": 17.082955869030673,
      "grad_norm": 0.1903599202632904,
      "learning_rate": 0.0009780725022104334,
      "loss": 2.7729,
      "step": 33000
    },
    {
      "epoch": 17.60062119839524,
      "grad_norm": 0.18594609200954437,
      "learning_rate": 0.0009559902740937224,
      "loss": 2.7451,
      "step": 34000
    },
    {
      "epoch": 17.99974116733532,
      "eval_accuracy": 0.41887905804207104,
      "eval_loss": 2.98665452003479,
      "eval_runtime": 112.7053,
      "eval_samples_per_second": 463.066,
      "eval_steps_per_second": 7.24,
      "step": 34771
    },
    {
      "epoch": 18.118286527759803,
      "grad_norm": 0.19103111326694489,
      "learning_rate": 0.0009338859416445623,
      "loss": 2.7475,
      "step": 35000
    },
    {
      "epoch": 18.63595185712437,
      "grad_norm": 0.17397448420524597,
      "learning_rate": 0.0009118037135278515,
      "loss": 2.7275,
      "step": 36000
    },
    {
      "epoch": 18.999870583667658,
      "eval_accuracy": 0.4201490782172229,
      "eval_loss": 2.98036527633667,
      "eval_runtime": 112.4975,
      "eval_samples_per_second": 463.921,
      "eval_steps_per_second": 7.253,
      "step": 36703
    },
    {
      "epoch": 19.153617186488933,
      "grad_norm": 0.18536260724067688,
      "learning_rate": 0.0008896993810786914,
      "loss": 2.7238,
      "step": 37000
    },
    {
      "epoch": 19.671282515853502,
      "grad_norm": 0.17299328744411469,
      "learning_rate": 0.0008676171529619805,
      "loss": 2.7099,
      "step": 38000
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 0.42075012492063313,
      "eval_loss": 2.9760243892669678,
      "eval_runtime": 112.7131,
      "eval_samples_per_second": 463.034,
      "eval_steps_per_second": 7.24,
      "step": 38635
    },
    {
      "epoch": 20.188947845218067,
      "grad_norm": 0.19630002975463867,
      "learning_rate": 0.0008455128205128205,
      "loss": 2.7028,
      "step": 39000
    },
    {
      "epoch": 20.706613174582632,
      "grad_norm": 0.18252891302108765,
      "learning_rate": 0.0008234084880636605,
      "loss": 2.693,
      "step": 40000
    },
    {
      "epoch": 20.999611751002977,
      "eval_accuracy": 0.4216216764536817,
      "eval_loss": 2.968324899673462,
      "eval_runtime": 112.7059,
      "eval_samples_per_second": 463.064,
      "eval_steps_per_second": 7.24,
      "step": 40566
    },
    {
      "epoch": 21.224278503947197,
      "grad_norm": 0.19568035006523132,
      "learning_rate": 0.0008013262599469496,
      "loss": 2.6802,
      "step": 41000
    },
    {
      "epoch": 21.741943833311765,
      "grad_norm": 0.2092587798833847,
      "learning_rate": 0.0007792219274977895,
      "loss": 2.6785,
      "step": 42000
    },
    {
      "epoch": 21.99974116733532,
      "eval_accuracy": 0.4221156483286934,
      "eval_loss": 2.96663761138916,
      "eval_runtime": 112.5823,
      "eval_samples_per_second": 463.572,
      "eval_steps_per_second": 7.248,
      "step": 42498
    },
    {
      "epoch": 22.25960916267633,
      "grad_norm": 0.2045181393623352,
      "learning_rate": 0.0007571175950486296,
      "loss": 2.6616,
      "step": 43000
    },
    {
      "epoch": 22.777274492040895,
      "grad_norm": 0.19560641050338745,
      "learning_rate": 0.0007350132625994696,
      "loss": 2.6628,
      "step": 44000
    },
    {
      "epoch": 22.999870583667658,
      "eval_accuracy": 0.4226662007972378,
      "eval_loss": 2.9646129608154297,
      "eval_runtime": 112.7148,
      "eval_samples_per_second": 463.027,
      "eval_steps_per_second": 7.24,
      "step": 44430
    },
    {
      "epoch": 23.29493982140546,
      "grad_norm": 0.19825534522533417,
      "learning_rate": 0.0007129089301503095,
      "loss": 2.6395,
      "step": 45000
    },
    {
      "epoch": 23.812605150770025,
      "grad_norm": 0.1957850456237793,
      "learning_rate": 0.0006908267020335986,
      "loss": 2.6501,
      "step": 46000
    },
    {
      "epoch": 24.0,
      "eval_accuracy": 0.42281715752022214,
      "eval_loss": 2.9626119136810303,
      "eval_runtime": 111.9607,
      "eval_samples_per_second": 466.146,
      "eval_steps_per_second": 7.288,
      "step": 46362
    },
    {
      "epoch": 24.330270480134594,
      "grad_norm": 0.2085314244031906,
      "learning_rate": 0.0006687223695844385,
      "loss": 2.6181,
      "step": 47000
    },
    {
      "epoch": 24.84793580949916,
      "grad_norm": 0.21406565606594086,
      "learning_rate": 0.0006466401414677277,
      "loss": 2.6343,
      "step": 48000
    },
    {
      "epoch": 24.999611751002977,
      "eval_accuracy": 0.42334742212654364,
      "eval_loss": 2.9599573612213135,
      "eval_runtime": 111.9994,
      "eval_samples_per_second": 465.985,
      "eval_steps_per_second": 7.286,
      "step": 48293
    },
    {
      "epoch": 25.365601138863724,
      "grad_norm": 0.20139683783054352,
      "learning_rate": 0.0006245358090185677,
      "loss": 2.598,
      "step": 49000
    },
    {
      "epoch": 25.88326646822829,
      "grad_norm": 0.20590080320835114,
      "learning_rate": 0.0006024535809018568,
      "loss": 2.6198,
      "step": 50000
    },
    {
      "epoch": 25.99974116733532,
      "eval_accuracy": 0.42357351907998303,
      "eval_loss": 2.9637885093688965,
      "eval_runtime": 112.1406,
      "eval_samples_per_second": 465.398,
      "eval_steps_per_second": 7.277,
      "step": 50225
    },
    {
      "epoch": 26.400931797592857,
      "grad_norm": 0.2109968364238739,
      "learning_rate": 0.0005803492484526968,
      "loss": 2.5789,
      "step": 51000
    },
    {
      "epoch": 26.918597126957422,
      "grad_norm": 0.21926765143871307,
      "learning_rate": 0.0005582670203359858,
      "loss": 2.604,
      "step": 52000
    },
    {
      "epoch": 26.999870583667658,
      "eval_accuracy": 0.4239798023060537,
      "eval_loss": 2.9604480266571045,
      "eval_runtime": 112.1419,
      "eval_samples_per_second": 465.392,
      "eval_steps_per_second": 7.276,
      "step": 52157
    },
    {
      "epoch": 27.436262456321987,
      "grad_norm": 0.2280665636062622,
      "learning_rate": 0.0005361626878868259,
      "loss": 2.5576,
      "step": 53000
    },
    {
      "epoch": 27.953927785686552,
      "grad_norm": 0.218441903591156,
      "learning_rate": 0.000514080459770115,
      "loss": 2.5876,
      "step": 54000
    },
    {
      "epoch": 28.0,
      "eval_accuracy": 0.42452238991016983,
      "eval_loss": 2.960141658782959,
      "eval_runtime": 111.7102,
      "eval_samples_per_second": 467.191,
      "eval_steps_per_second": 7.305,
      "step": 54089
    },
    {
      "epoch": 28.0,
      "step": 54089,
      "total_flos": 1.808986925039616e+18,
      "train_loss": 2.9159927132606205,
      "train_runtime": 57317.9169,
      "train_samples_per_second": 345.105,
      "train_steps_per_second": 1.348
    }
  ],
  "logging_steps": 1000,
  "max_steps": 77240,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 40,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 3,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 3
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.808986925039616e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}