Vector Db

Index setups, embeddings models, distances metrics, and cosine similarity.

Core Guidelines

• Index Optimization: Use HNSW indices for faster similarity searches.
• Metric Match: Ensure the query metric (Cosine, L2, Dot Product) matches the embedding training parameters.
• Metadata Filters: Apply metadata pre-filtering to scope queries efficiently.