Deep Dive Into Self-Query Retriever In Langchain : Exploring Pros of Building SQ Retriever with PaLM

SelfQueryRetriever queries itself, as the Langchain Docs say.

In this discussion we dive into how it is doing, and how to build one. Given a VectorStore (Chroma / Weaviat / Pinecone at the moment) retriever uses a query-constructing LLM chain to write a structured query and then applies that structured query to it’s underlying VectorStore.

SQ Retriever uses user-input query for semantic similarity search and extracts filters, limits from the user query. Uses them on the metadata of stored documents and to execute those filters.

The code can be found here. I believe you will learn something interesting

python_de_learners_data/selfQueryingRetriever_QAChains.ipynb at main ·…