Challenging Decoder helps in Masked Auto-Encoder Pre-training for Dense Passage Retrieval

TL;DR

This paper introduces a challenging decoder strategy in masked auto-encoder pre-training that enhances dense passage retrieval by improving encoder representations without extra costs.

Contribution

It proposes a novel token importance aware masking method based on mutual information to make the decoder more demanding, strengthening encoder learning in an unsupervised way.

Findings

Improves dense passage retrieval performance on large-scale datasets.

Enhances zero-shot retrieval robustness.

No additional pre-training expenses incurred.

Abstract

Recently, various studies have been directed towards exploring dense passage retrieval techniques employing pre-trained language models, among which the masked auto-encoder (MAE) pre-training architecture has emerged as the most promising. The conventional MAE framework relies on leveraging the passage reconstruction of decoder to bolster the text representation ability of encoder, thereby enhancing the performance of resulting dense retrieval systems. Within the context of building the representation ability of the encoder through passage reconstruction of decoder, it is reasonable to postulate that a ``more demanding'' decoder will necessitate a corresponding increase in the encoder's ability. To this end, we propose a novel token importance aware masking strategy based on pointwise mutual information to intensify the challenge of the decoder. Importantly, our approach can be implemented in an unsupervised manner, without adding additional expenses to the pre-training phase. Our experiments verify that the proposed method is both effective and robust on large-scale supervised passage retrieval datasets and out-of-domain zero-shot retrieval benchmarks.