AI Glossary

Bi-encoder retrieval

Also called dual encoder

What does “Bi-encoder retrieval” mean?

Bi-encoder retrieval: Encoding queries and candidates separately so their vectors can be compared efficiently.

Bi-encoder retrieval visual guide: Encoding queries and candidates separately so their vectors can be compared efficiently.
Visual guide Open full size (new tab)

What is an example of “Bi-encoder retrieval”?

A document index stores embeddings ahead of time and only encodes the incoming query at search time.

Keep learning

See also

Semantic searchTermRetrieval-augmented generationTermHybrid searchTermGraph RAGTermCross-encoder rerankerTerm

Same topic

Relevance feedbackTermRetrieval recallTermBM25TermBatch APITermChain of thoughtTerm