Concept
Dense Passage Retriever
Dense Passage Retriever (DPR) is a neural retrieval method that finds relevant text passages by comparing dense vector embeddings instead of matching keywords. Think of it like BM25 or Elasticsearch, except queries and documents are encoded into fixed-length vectors, and…
The rest of “Dense Passage Retriever” is a premium feature: every concept in the library gets a precise, practitioner-focused write-up like this one, cross-linked straight from the paper summaries that use it.
Log in to unlock→