Retrieval Augmented Generation (RAG) › Reranking Retrieved Documents

What is Reranking Retrieved Documents

Beginner

Answer

Reranking retrieved documents is an extra step in a RAG pipeline that reorders the initially retrieved documents to make sure the most genuinely relevant ones are placed first.

Real-world example A RAG system uses a reranking step to move the single most relevant support article to the top before generating a final answer.

Common follow-ups: What is Retrieval Augmented Generation (RAG), How is Reranking Retrieved Documents evaluated in practice, What tools are commonly used for Reranking Retrieved Documents

Retrieval Augmented Generation (RAG) topics: Introduction to Retrieval Augmented Generation Vector Databases for RAG Embedding Models for Retrieval