(venv) PS D:\Zidan\1. Machine Learning\sum_liputan6> python train_evaluate.py --model tfidf_lr --train_limit 5000
=== Training TFIDF_LR Model ===
Loading training data...
Training data: 5000 articles
Training completed in 32.52 seconds

Evaluating on dev set...

=== Evaluating on DEV set ===
Evaluating on 100 articles...
Extractive Performance:
  accuracy: 0.7920
  precision: 0.3950
  recall: 0.3692
  f1: 0.3816
ROUGE Scores:
  rouge-1: 0.3564
  rouge-2: 0.1958
  rouge-l: 0.2981

Model saved to saved_models/tfidf_lr_model.pkl