src/config/config.yaml

data:
  raw_path: artifacts/raw_data/electronics_sample_50k.json
  cleaned_path: artifacts/cleaned_data/clean_reviews.parquet
  embeddings_path: artifacts/embeddings/review_embeddings.npy
  cluster_assignments_path: artifacts/clustering/cluster_labels.csv
  summaries_path: artifacts/summaries/cluster_summaries.json
models:
  embedding_model: sentence-transformers/all-MiniLM-L6-v2
  summarizer_model: google/pegasus-xsum
clustering:
  n_clusters: 20
  pca_components: 50
app:
  results_cache: artifacts/summaries/cluster_summaries.json