PostgreSQL Embeddings With Pg_auto_embeddings
pg_auto_embeddings computes OpenAI model embeddings directly in PostgreSQL without extensions. Install with a single SQL script or use Docker for on-premise setup.
Overview You have a PostgreSQL database that stores a lot of text data. You want to use vector representations (embeddings), for example from OpenAI, to build a recommendation system, improved search, or implement RAG for working with LLMs. But you don’t want to install extensions (or maybe you can’t). For instance, on cloud Managed PostgreSQL, you often don’t have the required permissions. pg_auto_embeddings is a lightweight open-source solution that lets you compute embeddings through OpenAI models directly in PostgreSQL without installing external extensions. It uses the Foreign Data Wrapp...