Model: microsoft/codebert-base
Trained on dataset: HID-APP/RD-Keystrokes-all-combined-clean-v3-1-training
Dataset version: v3.1-repaired
Full metadata repo: HID-APP/RD-Keystrokes-all-combined-clean-v3-1
Hub repo: HID-APP/codebert-rd-keystrokes-v3-1
Resume from checkpoint: /kaggle/working/outputs/codebert_rd_keystrokes_v3_1/checkpoint-3000

Validation metrics:
{
  "eval_loss": 0.09398876130580902,
  "eval_accuracy": 0.9817034700315458,
  "eval_precision": 0.9698461538461538,
  "eval_recall": 0.9943217665615142,
  "eval_f1": 0.9819314641744549
}

Test metrics:
Accuracy : 0.983596
Precision: 0.972856
Recall   : 0.994953
F1       : 0.983780

              precision    recall  f1-score   support

      benign       0.99      0.97      0.98      1585
   malicious       0.97      0.99      0.98      1585

    accuracy                           0.98      3170
   macro avg       0.98      0.98      0.98      3170
weighted avg       0.98      0.98      0.98      3170