postgres-vectors

What problem does it solve?

Provides patterns for vector storage, distance operators, index strategies, and similarity queries in PostgreSQL using pgvector (384-d embeddings).

Core Features & Use Cases

• Store 384-d embeddings in a dedicated table
• Use cosine distance (<=>) for similarity queries
• Choose indexing strategy (HNSW vs IVFFlat) and tune parameters
• Find and batch-calculate similarity across multiple items

Quick Start

Create the embeddings table and run a cosine similarity query to find similar embeddings.