What is ‘hypothetical document embedding’ (HyDE) and how does it improve RAG?

Question

AI Automation Specialist — Hard

What is ‘hypothetical document embedding’ (HyDE) and how does it improve RAG?

Accepted Answer

HyDE bridges the query-document embedding gap by having the LLM generate a hypothetical answer, whose dense vector often better matches actual document embeddings than the sparse query.