Moyan AI Training Institution LogoMoyan AI

RAG & Vector Search · Established · Intermediate

Cross-Encoder Re-Ranking

Also known as: Cross-Encoder Reranker, Passage Rescoring

A retrieval technique that passes query-document pairs jointly through a transformer to score exact semantic relevance.

What Cross-Encoder Re-Ranking is

Cross-Encoder Re-Ranking drastically improves RAG retrieval precision compared to initial vector cosine similarity searches.

How it works

Performs full cross-attention between query and candidate passages retrieved by fast bi-encoder vector algorithms.

Why it matters

Deploys in RAG pipelines as a second-stage rescorer before constructing the final prompt context.

Common uses

  • RAG precision optimization
  • Search result re-ranking
  • Enterprise knowledge retrieval

More in this collection

Browse all AI Concepts