What is Reranking?

RAG

Reranking — A second-stage retrieval step that re-scores and reorders initial search results using a more powerful model. Improves RAG accuracy by 10-30% over embedding-only retrieval.

FAQ

What is reranking?

A second pass that re-scores retrieved documents using a cross-encoder model, improving relevance ranking.

Does reranking help?

Yes, typically 10-30% improvement in retrieval accuracy. Use Cohere Rerank or cross-encoder models.

Related Terms

Learn Reranking in depth

Free hands-on course with code examples and Google Colab notebooks.

Start Course →