RocketQAv2: A Joint Training Method for Dense Passage Retrieval and Passage Re-ranking

TL;DR

RocketQAv2 introduces a joint training method for dense passage retrieval and re-ranking, utilizing dynamic listwise distillation and hybrid data augmentation to improve performance on key NLP retrieval tasks.

Contribution

It presents a novel unified listwise training approach for both retrieval and re-ranking models, enabling mutual enhancement during training.

Findings

Significant performance improvements on MSMARCO and Natural Questions datasets.

Effective dynamic distillation enhances both retriever and re-ranker.

Hybrid data augmentation diversifies training instances for better generalization.

Abstract

In various natural language processing tasks, passage retrieval and passage re-ranking are two key procedures in finding and ranking relevant information. Since both the two procedures contribute to the final performance, it is important to jointly optimize them in order to achieve mutual improvement. In this paper, we propose a novel joint training approach for dense passage retrieval and passage re-ranking. A major contribution is that we introduce the dynamic listwise distillation, where we design a unified listwise training approach for both the retriever and the re-ranker. During the dynamic distillation, the retriever and the re-ranker can be adaptively improved according to each other's relevance information. We also propose a hybrid data augmentation strategy to construct diverse training instances for listwise training approach. Extensive experiments show the effectiveness of our approach on both MSMARCO and Natural Questions datasets. Our code is available at https://github.com/PaddlePaddle/RocketQA.