{
  "best_global_step": 4400,
  "best_metric": 0.453235387802124,
  "best_model_checkpoint": "roberta-base-multi-head-4-directions/checkpoint-4400",
  "epoch": 3.259558164180321,
  "eval_steps": 200,
  "global_step": 5000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.13043123828156844,
      "eval_accuracy": 0.15051240893937523,
      "eval_f1_macro": 0.0654110324103885,
      "eval_f1_micro": 0.15051240893937523,
      "eval_loss": 0.6699725985527039,
      "eval_precision_macro": 0.03762810223484381,
      "eval_recall_macro": 0.25,
      "eval_roc_auc": 0.5065070998981289,
      "eval_runtime": 58.3231,
      "eval_samples_per_second": 138.864,
      "eval_steps_per_second": 69.441,
      "step": 200
    },
    {
      "epoch": 0.2608624765631369,
      "eval_accuracy": 0.3306581059390048,
      "eval_f1_macro": 0.21851269600524115,
      "eval_f1_micro": 0.3306581059390048,
      "eval_loss": 0.5907890796661377,
      "eval_precision_macro": 0.3477712230149137,
      "eval_recall_macro": 0.2628129027541947,
      "eval_roc_auc": 0.5264776449526523,
      "eval_runtime": 58.515,
      "eval_samples_per_second": 138.409,
      "eval_steps_per_second": 69.213,
      "step": 400
    },
    {
      "epoch": 0.3260780957039211,
      "grad_norm": 1.910775899887085,
      "learning_rate": 8.132333767926989e-07,
      "loss": 0.6418,
      "step": 500
    },
    {
      "epoch": 0.3912937148447053,
      "eval_accuracy": 0.39646869983948635,
      "eval_f1_macro": 0.24202854764905962,
      "eval_f1_micro": 0.39646869983948635,
      "eval_loss": 0.5681799650192261,
      "eval_precision_macro": 0.31314716750672944,
      "eval_recall_macro": 0.29579537486279056,
      "eval_roc_auc": 0.5825887308747437,
      "eval_runtime": 58.253,
      "eval_samples_per_second": 139.031,
      "eval_steps_per_second": 69.524,
      "step": 600
    },
    {
      "epoch": 0.5217249531262738,
      "eval_accuracy": 0.4446227929373997,
      "eval_f1_macro": 0.2752091664099515,
      "eval_f1_micro": 0.4446227929373997,
      "eval_loss": 0.5510929226875305,
      "eval_precision_macro": 0.3725228720549576,
      "eval_recall_macro": 0.3347772936726068,
      "eval_roc_auc": 0.6511260809997589,
      "eval_runtime": 58.5957,
      "eval_samples_per_second": 138.218,
      "eval_steps_per_second": 69.118,
      "step": 800
    },
    {
      "epoch": 0.6521561914078422,
      "grad_norm": 3.2853586673736572,
      "learning_rate": 1.6280964797913951e-06,
      "loss": 0.56,
      "step": 1000
    },
    {
      "epoch": 0.6521561914078422,
      "eval_accuracy": 0.47956537844178293,
      "eval_f1_macro": 0.339724330856578,
      "eval_f1_micro": 0.47956537844178293,
      "eval_loss": 0.5183268189430237,
      "eval_precision_macro": 0.5998469679206377,
      "eval_recall_macro": 0.37515425621657034,
      "eval_roc_auc": 0.7004256590184301,
      "eval_runtime": 58.4596,
      "eval_samples_per_second": 138.54,
      "eval_steps_per_second": 69.279,
      "step": 1000
    },
    {
      "epoch": 0.7825874296894106,
      "eval_accuracy": 0.4951228546734165,
      "eval_f1_macro": 0.3960300119997172,
      "eval_f1_micro": 0.4951228546734165,
      "eval_loss": 0.5027472972869873,
      "eval_precision_macro": 0.47649685990985047,
      "eval_recall_macro": 0.42432070581801545,
      "eval_roc_auc": 0.7210076403173139,
      "eval_runtime": 59.2756,
      "eval_samples_per_second": 136.633,
      "eval_steps_per_second": 68.325,
      "step": 1200
    },
    {
      "epoch": 0.9130186679709791,
      "eval_accuracy": 0.49364119027040376,
      "eval_f1_macro": 0.4202868285905695,
      "eval_f1_micro": 0.49364119027040376,
      "eval_loss": 0.5012013912200928,
      "eval_precision_macro": 0.476097069319352,
      "eval_recall_macro": 0.44606990421659504,
      "eval_roc_auc": 0.7341331293715618,
      "eval_runtime": 58.8986,
      "eval_samples_per_second": 137.507,
      "eval_steps_per_second": 68.762,
      "step": 1400
    },
    {
      "epoch": 0.9782342871117633,
      "grad_norm": 6.466639041900635,
      "learning_rate": 2.4429595827900914e-06,
      "loss": 0.5136,
      "step": 1500
    },
    {
      "epoch": 1.0430423086329177,
      "eval_accuracy": 0.5227805901963205,
      "eval_f1_macro": 0.4272500725974152,
      "eval_f1_micro": 0.5227805901963205,
      "eval_loss": 0.4897969961166382,
      "eval_precision_macro": 0.511088683081628,
      "eval_recall_macro": 0.4450094829273592,
      "eval_roc_auc": 0.7377316174079231,
      "eval_runtime": 58.3834,
      "eval_samples_per_second": 138.721,
      "eval_steps_per_second": 69.369,
      "step": 1600
    },
    {
      "epoch": 1.1734735469144861,
      "eval_accuracy": 0.5304358562785529,
      "eval_f1_macro": 0.47228705140051763,
      "eval_f1_micro": 0.5304358562785529,
      "eval_loss": 0.48175758123397827,
      "eval_precision_macro": 0.49865343258508565,
      "eval_recall_macro": 0.4751571425140743,
      "eval_roc_auc": 0.7504044589822995,
      "eval_runtime": 59.0784,
      "eval_samples_per_second": 137.089,
      "eval_steps_per_second": 68.553,
      "step": 1800
    },
    {
      "epoch": 1.3039047851960546,
      "grad_norm": 7.449667453765869,
      "learning_rate": 3.257822685788788e-06,
      "loss": 0.4842,
      "step": 2000
    },
    {
      "epoch": 1.3039047851960546,
      "eval_accuracy": 0.5330287689838251,
      "eval_f1_macro": 0.49309253039408907,
      "eval_f1_micro": 0.5330287689838251,
      "eval_loss": 0.4809001386165619,
      "eval_precision_macro": 0.5034709810584846,
      "eval_recall_macro": 0.4966387871579345,
      "eval_roc_auc": 0.758014108468557,
      "eval_runtime": 58.3487,
      "eval_samples_per_second": 138.803,
      "eval_steps_per_second": 69.41,
      "step": 2000
    },
    {
      "epoch": 1.434336023477623,
      "eval_accuracy": 0.5411779232003952,
      "eval_f1_macro": 0.5090942441371813,
      "eval_f1_micro": 0.5411779232003952,
      "eval_loss": 0.47537535429000854,
      "eval_precision_macro": 0.5108825390909519,
      "eval_recall_macro": 0.5109581830122815,
      "eval_roc_auc": 0.7651162037942221,
      "eval_runtime": 58.7575,
      "eval_samples_per_second": 137.838,
      "eval_steps_per_second": 68.927,
      "step": 2200
    },
    {
      "epoch": 1.5647672617591915,
      "eval_accuracy": 0.5522904062229905,
      "eval_f1_macro": 0.5071849375719386,
      "eval_f1_micro": 0.5522904062229905,
      "eval_loss": 0.468904048204422,
      "eval_precision_macro": 0.5307207892146724,
      "eval_recall_macro": 0.5061303129931647,
      "eval_roc_auc": 0.7695441321425415,
      "eval_runtime": 58.5621,
      "eval_samples_per_second": 138.298,
      "eval_steps_per_second": 69.157,
      "step": 2400
    },
    {
      "epoch": 1.6299828808999757,
      "grad_norm": 6.51912260055542,
      "learning_rate": 4.072685788787484e-06,
      "loss": 0.4695,
      "step": 2500
    },
    {
      "epoch": 1.69519850004076,
      "eval_accuracy": 0.5379676503272008,
      "eval_f1_macro": 0.48087910671865897,
      "eval_f1_micro": 0.5379676503272008,
      "eval_loss": 0.4785490036010742,
      "eval_precision_macro": 0.5346246897655283,
      "eval_recall_macro": 0.4906654493419952,
      "eval_roc_auc": 0.7663202674809741,
      "eval_runtime": 58.4801,
      "eval_samples_per_second": 138.491,
      "eval_steps_per_second": 69.254,
      "step": 2600
    },
    {
      "epoch": 1.8256297383223283,
      "eval_accuracy": 0.5545129028275095,
      "eval_f1_macro": 0.5144296310304717,
      "eval_f1_micro": 0.5545129028275095,
      "eval_loss": 0.46492108702659607,
      "eval_precision_macro": 0.5301331529844335,
      "eval_recall_macro": 0.5080649601410273,
      "eval_roc_auc": 0.7745130299239165,
      "eval_runtime": 58.563,
      "eval_samples_per_second": 138.295,
      "eval_steps_per_second": 69.156,
      "step": 2800
    },
    {
      "epoch": 1.9560609766038968,
      "grad_norm": 6.679250717163086,
      "learning_rate": 4.88754889178618e-06,
      "loss": 0.4655,
      "step": 3000
    },
    {
      "epoch": 1.9560609766038968,
      "eval_accuracy": 0.5494505494505495,
      "eval_f1_macro": 0.5210019068338745,
      "eval_f1_micro": 0.5494505494505495,
      "eval_loss": 0.46518564224243164,
      "eval_precision_macro": 0.5276079708493782,
      "eval_recall_macro": 0.5245104582824609,
      "eval_roc_auc": 0.7757459310311635,
      "eval_runtime": 58.703,
      "eval_samples_per_second": 137.966,
      "eval_steps_per_second": 68.991,
      "step": 3000
    },
    {
      "epoch": 2.0860846172658354,
      "eval_accuracy": 0.5653784417829362,
      "eval_f1_macro": 0.528637637687615,
      "eval_f1_micro": 0.5653784417829362,
      "eval_loss": 0.46096640825271606,
      "eval_precision_macro": 0.5443547583807372,
      "eval_recall_macro": 0.5223821110706193,
      "eval_roc_auc": 0.781316894157911,
      "eval_runtime": 58.6208,
      "eval_samples_per_second": 138.159,
      "eval_steps_per_second": 69.088,
      "step": 3200
    },
    {
      "epoch": 2.216515855547404,
      "eval_accuracy": 0.5498209655513027,
      "eval_f1_macro": 0.5186060540296338,
      "eval_f1_micro": 0.5498209655513027,
      "eval_loss": 0.46535903215408325,
      "eval_precision_macro": 0.5399973417347204,
      "eval_recall_macro": 0.5255215009903015,
      "eval_roc_auc": 0.7808947543899514,
      "eval_runtime": 58.434,
      "eval_samples_per_second": 138.601,
      "eval_steps_per_second": 69.309,
      "step": 3400
    },
    {
      "epoch": 2.281731474688188,
      "grad_norm": 7.935737609863281,
      "learning_rate": 4.963030947642902e-06,
      "loss": 0.4414,
      "step": 3500
    },
    {
      "epoch": 2.3469470938289723,
      "eval_accuracy": 0.5530312384244969,
      "eval_f1_macro": 0.5192037737144601,
      "eval_f1_micro": 0.5530312384244969,
      "eval_loss": 0.4659627676010132,
      "eval_precision_macro": 0.542720199525297,
      "eval_recall_macro": 0.5266595526270865,
      "eval_roc_auc": 0.7817212427653404,
      "eval_runtime": 58.8426,
      "eval_samples_per_second": 137.638,
      "eval_steps_per_second": 68.828,
      "step": 3600
    },
    {
      "epoch": 2.4773783321105407,
      "eval_accuracy": 0.564390665514261,
      "eval_f1_macro": 0.5393273157888367,
      "eval_f1_micro": 0.564390665514261,
      "eval_loss": 0.45927414298057556,
      "eval_precision_macro": 0.5430622655929049,
      "eval_recall_macro": 0.5388167782580582,
      "eval_roc_auc": 0.785497056666471,
      "eval_runtime": 60.3532,
      "eval_samples_per_second": 134.193,
      "eval_steps_per_second": 67.105,
      "step": 3800
    },
    {
      "epoch": 2.607809570392109,
      "grad_norm": 6.252172946929932,
      "learning_rate": 4.920143415906128e-06,
      "loss": 0.433,
      "step": 4000
    },
    {
      "epoch": 2.607809570392109,
      "eval_accuracy": 0.5583405358686258,
      "eval_f1_macro": 0.5308754875566113,
      "eval_f1_micro": 0.5583405358686258,
      "eval_loss": 0.4604816436767578,
      "eval_precision_macro": 0.5405007742715062,
      "eval_recall_macro": 0.5347270499333006,
      "eval_roc_auc": 0.7849083537765603,
      "eval_runtime": 58.4178,
      "eval_samples_per_second": 138.639,
      "eval_steps_per_second": 69.328,
      "step": 4000
    },
    {
      "epoch": 2.7382408086736776,
      "eval_accuracy": 0.5485862452154587,
      "eval_f1_macro": 0.5244598378559915,
      "eval_f1_micro": 0.5485862452154587,
      "eval_loss": 0.46733224391937256,
      "eval_precision_macro": 0.5429571340400278,
      "eval_recall_macro": 0.5359068311148909,
      "eval_roc_auc": 0.7850860010461288,
      "eval_runtime": 58.5291,
      "eval_samples_per_second": 138.376,
      "eval_steps_per_second": 69.196,
      "step": 4200
    },
    {
      "epoch": 2.868672046955246,
      "eval_accuracy": 0.5685887146561304,
      "eval_f1_macro": 0.5358137721542776,
      "eval_f1_micro": 0.5685887146561304,
      "eval_loss": 0.453235387802124,
      "eval_precision_macro": 0.5498552760907015,
      "eval_recall_macro": 0.531919882945018,
      "eval_roc_auc": 0.7869039137806253,
      "eval_runtime": 58.2827,
      "eval_samples_per_second": 138.961,
      "eval_steps_per_second": 69.489,
      "step": 4400
    },
    {
      "epoch": 2.9338876660960302,
      "grad_norm": 7.882796287536621,
      "learning_rate": 4.8772558841693544e-06,
      "loss": 0.435,
      "step": 4500
    },
    {
      "epoch": 2.9991032852368145,
      "eval_accuracy": 0.5695764909248056,
      "eval_f1_macro": 0.5394199448286808,
      "eval_f1_micro": 0.5695764909248056,
      "eval_loss": 0.4586756229400635,
      "eval_precision_macro": 0.5469928436264997,
      "eval_recall_macro": 0.5385537439613218,
      "eval_roc_auc": 0.7865007504045092,
      "eval_runtime": 58.5001,
      "eval_samples_per_second": 138.444,
      "eval_steps_per_second": 69.231,
      "step": 4600
    },
    {
      "epoch": 3.1291269258987526,
      "eval_accuracy": 0.5659958019508581,
      "eval_f1_macro": 0.5352002738500827,
      "eval_f1_micro": 0.5659958019508581,
      "eval_loss": 0.46012088656425476,
      "eval_precision_macro": 0.5485092746099708,
      "eval_recall_macro": 0.537293222573151,
      "eval_roc_auc": 0.7868513938139832,
      "eval_runtime": 58.3581,
      "eval_samples_per_second": 138.781,
      "eval_steps_per_second": 69.399,
      "step": 4800
    },
    {
      "epoch": 3.259558164180321,
      "grad_norm": 6.890503406524658,
      "learning_rate": 4.834368352432582e-06,
      "loss": 0.4056,
      "step": 5000
    },
    {
      "epoch": 3.259558164180321,
      "eval_accuracy": 0.5674774663538709,
      "eval_f1_macro": 0.5441156714097114,
      "eval_f1_micro": 0.5674774663538709,
      "eval_loss": 0.47018805146217346,
      "eval_precision_macro": 0.5454139749449086,
      "eval_recall_macro": 0.5461421293276678,
      "eval_roc_auc": 0.7877411895075941,
      "eval_runtime": 58.8315,
      "eval_samples_per_second": 137.664,
      "eval_steps_per_second": 68.841,
      "step": 5000
    },
    {
      "epoch": 3.259558164180321,
      "step": 5000,
      "total_flos": 1.052068446713856e+16,
      "train_loss": 0.48495769653320314,
      "train_runtime": 3435.9427,
      "train_samples_per_second": 285.616,
      "train_steps_per_second": 17.858
    }
  ],
  "logging_steps": 500,
  "max_steps": 61360,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 40,
  "save_steps": 400,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 3,
        "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": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.052068446713856e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}