training_checkpoints/checkpoint-2572/trainer_state.json

{
  "best_metric": 0.7407336443567221,
  "best_model_checkpoint": "InLegalBERT/checkpoint-2572",
  "epoch": 4.0,
  "eval_steps": 500,
  "global_step": 2572,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.78,
      "grad_norm": 6.225058078765869,
      "learning_rate": 4.742871954380509e-05,
      "loss": 1.065,
      "step": 500
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7993803253292022,
      "eval_f1_macro": 0.6028764374506596,
      "eval_f1_micro": 0.7993803253292022,
      "eval_f1_weighted": 0.780434475079005,
      "eval_loss": 0.6395189166069031,
      "eval_macro_fpr": 0.01853296480715729,
      "eval_macro_sensitivity": 0.6307714529979154,
      "eval_macro_specificity": 0.9847175193097543,
      "eval_precision": 0.7817573111259606,
      "eval_precision_macro": 0.6194286814679265,
      "eval_recall": 0.7993803253292022,
      "eval_recall_macro": 0.6307714529979154,
      "eval_runtime": 29.4696,
      "eval_samples_per_second": 43.808,
      "eval_steps_per_second": 5.497,
      "eval_weighted_fpr": 0.017610661589719183,
      "eval_weighted_sensitivity": 0.7993803253292022,
      "eval_weighted_specificity": 0.9713824643171124,
      "step": 643
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.756747245788574,
      "learning_rate": 4.483670295489891e-05,
      "loss": 0.5866,
      "step": 1000
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8187451587916343,
      "eval_f1_macro": 0.7276084501581802,
      "eval_f1_micro": 0.8187451587916342,
      "eval_f1_weighted": 0.8152450677099363,
      "eval_loss": 0.6906521916389465,
      "eval_macro_fpr": 0.01610571025604924,
      "eval_macro_sensitivity": 0.7365748487318147,
      "eval_macro_specificity": 0.9863504064994671,
      "eval_precision": 0.8198905458065566,
      "eval_precision_macro": 0.7284980957351811,
      "eval_recall": 0.8187451587916343,
      "eval_recall_macro": 0.7365748487318147,
      "eval_runtime": 57.1075,
      "eval_samples_per_second": 22.606,
      "eval_steps_per_second": 2.837,
      "eval_weighted_fpr": 0.015566790846194785,
      "eval_weighted_sensitivity": 0.8187451587916343,
      "eval_weighted_specificity": 0.976510938700373,
      "step": 1286
    },
    {
      "epoch": 2.33,
      "grad_norm": 8.241662979125977,
      "learning_rate": 4.224468636599274e-05,
      "loss": 0.4622,
      "step": 1500
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8179705654531371,
      "eval_f1_macro": 0.7282676953256336,
      "eval_f1_micro": 0.817970565453137,
      "eval_f1_weighted": 0.8150036178041801,
      "eval_loss": 0.8055915236473083,
      "eval_macro_fpr": 0.016168738933500626,
      "eval_macro_sensitivity": 0.737597935006265,
      "eval_macro_specificity": 0.9862931306285473,
      "eval_precision": 0.8137371356291269,
      "eval_precision_macro": 0.7226669118270935,
      "eval_recall": 0.8179705654531371,
      "eval_recall_macro": 0.737597935006265,
      "eval_runtime": 56.5581,
      "eval_samples_per_second": 22.826,
      "eval_steps_per_second": 2.864,
      "eval_weighted_fpr": 0.01564684732671949,
      "eval_weighted_sensitivity": 0.8179705654531371,
      "eval_weighted_specificity": 0.9764263939750704,
      "step": 1929
    },
    {
      "epoch": 3.11,
      "grad_norm": 5.718050003051758,
      "learning_rate": 3.9652669777086575e-05,
      "loss": 0.346,
      "step": 2000
    },
    {
      "epoch": 3.89,
      "grad_norm": 28.698440551757812,
      "learning_rate": 3.7060653188180404e-05,
      "loss": 0.2398,
      "step": 2500
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8171959721146398,
      "eval_f1_macro": 0.7407336443567221,
      "eval_f1_micro": 0.8171959721146398,
      "eval_f1_weighted": 0.8160506897887181,
      "eval_loss": 0.9310164451599121,
      "eval_macro_fpr": 0.016134598971936556,
      "eval_macro_sensitivity": 0.7424877811517157,
      "eval_macro_specificity": 0.9862275325034797,
      "eval_precision": 0.8234764041889034,
      "eval_precision_macro": 0.7660803499273104,
      "eval_recall": 0.8171959721146398,
      "eval_recall_macro": 0.7424877811517157,
      "eval_runtime": 58.71,
      "eval_samples_per_second": 21.989,
      "eval_steps_per_second": 2.759,
      "eval_weighted_fpr": 0.015727042516326804,
      "eval_weighted_sensitivity": 0.8171959721146398,
      "eval_weighted_specificity": 0.9762170154375561,
      "step": 2572
    }
  ],
  "logging_steps": 500,
  "max_steps": 9645,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 15,
  "save_steps": 500,
  "total_flos": 5408089577791488.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}