Dense Passage Retrieval (DPR)
Set of tools and models from Meta AI Research for open domain Q&A using dense representations, outperforming BM25 by 9%-19% in passage retrieval accuracy with a dual-encoder BERT framework.
About this tool
Overview
Dense Passage Retriever (DPR) is a set of tools and models for open domain Q&A task developed by Facebook AI Research (now Meta AI). The research shows that retrieval can be practically implemented using dense representations alone, where embeddings are learned from a small number of questions and passages.
Architecture
DPR uses:
- Dense encoder EP(·) which maps any text passage to a d-dimensional real-valued vector
- Different encoder EQ(·) that maps the input question to a d-dimensional vector
- Two independent BERT networks (base, uncased)
- FAISS for efficient inference-time encoding and indexing
Performance
When evaluated on a wide range of open-domain QA datasets, the dense retriever outperforms a strong Lucene-BM25 system by 9%-19% absolute in terms of top-20 passage retrieval accuracy.
Retrieval Process
At run-time, DPR applies the question encoder to map the input question to a d-dimensional vector, and retrieves k passages of which vectors are the closest to the question vector using efficient similarity search.
Resources
- GitHub Repository: github.com/facebookresearch/DPR
- Models: Pre-trained models available on Hugging Face
- Paper: Available on arXiv (2004.04906)
Pricing
Free and open-source.
Loading more......
Information
Categories
Tags
Similar Products
6 result(s)