Dense Passage Retrieval for Open-Domain Question Answering

Code

• github.com/facebookresearch/DPR
  OfficialIn paperpytorch★ 1,864
• github.com/huggingface/transformers
  pytorch★ 158,292
• github.com/deepset-ai/haystack
  pytorch★ 24,592
• github.com/texttron/tevatron
  jax★ 734
• github.com/luyug/GC-DPR
  pytorch★ 136
• github.com/DevSinghSachan/unsupervised-passage-reranking
  pytorch★ 100
• github.com/AkariAsai/XORQA
  pytorch★ 80
• github.com/oriram/spider
  pytorch★ 54
• github.com/Hannibal046/nanoDPR
  pytorch★ 54
• github.com/efficientqa/retrieval-based-baselines
  tf★ 51

Abstract

Open-domain question answering relies on efficient passage retrieval to select candidate contexts, where traditional sparse vector space models, such as TF-IDF or BM25, are the de facto method. In this work, we show that retrieval can be practically implemented using dense representations alone, where embeddings are learned from a small number of questions and passages by a simple dual-encoder framework. When evaluated on a wide range of open-domain QA datasets, our dense retriever outperforms a strong Lucene-BM25 system largely by 9%-19% absolute in terms of top-20 passage retrieval accuracy, and helps our end-to-end QA system establish new state-of-the-art on multiple open-domain QA benchmarks.

Tasks

Benchmark Results

Reproductions