RocketQA: An Optimized Training Approach to Dense Passage Retrieval for Open-Domain Question Answering

TL;DR

RocketQA introduces an optimized training method for dense passage retrieval in open-domain QA, significantly improving retrieval accuracy by addressing training challenges with innovative strategies.

Contribution

The paper presents RocketQA, a novel training approach incorporating cross-batch negatives, denoised hard negatives, and data augmentation to enhance dense passage retrieval.

Findings

RocketQA outperforms previous models on MSMARCO and Natural Questions datasets.

The three strategies significantly improve retrieval performance.

End-to-end QA performance is enhanced using RocketQA retriever.

Abstract

In open-domain question answering, dense passage retrieval has become a new paradigm to retrieve relevant passages for finding answers. Typically, the dual-encoder architecture is adopted to learn dense representations of questions and passages for semantic matching. However, it is difficult to effectively train a dual-encoder due to the challenges including the discrepancy between training and inference, the existence of unlabeled positives and limited training data. To address these challenges, we propose an optimized training approach, called RocketQA, to improving dense passage retrieval. We make three major technical contributions in RocketQA, namely cross-batch negatives, denoised hard negatives and data augmentation. The experiment results show that RocketQA significantly outperforms previous state-of-the-art models on both MSMARCO and Natural Questions. We also conduct extensive experiments to examine the effectiveness of the three strategies in RocketQA. Besides, we demonstrate that the performance of end-to-end QA can be improved based on our RocketQA retriever.