Adapting a Language Model While Preserving its General Knowledge

TL;DR

This paper introduces a novel domain-adaptive pre-training method that selectively preserves general knowledge in language models while effectively integrating domain-specific information, leading to improved task performance.

Contribution

It proposes a new approach that uses soft-masking of attention heads and contrasting representations to better preserve and adapt general knowledge during domain-specific training.

Findings

Improved performance on domain-specific tasks.

Effective preservation of general knowledge.

Enhanced integration of domain-specific information.

Abstract

Domain-adaptive pre-training (or DA-training for short), also known as post-training, aims to train a pre-trained general-purpose language model (LM) using an unlabeled corpus of a particular domain to adapt the LM so that end-tasks in the domain can give improved performances. However, existing DA-training methods are in some sense blind as they do not explicitly identify what knowledge in the LM should be preserved and what should be changed by the domain corpus. This paper shows that the existing methods are suboptimal and proposes a novel method to perform a more informed adaptation of the knowledge in the LM by (1) soft-masking the attention heads based on their importance to best preserve the general knowledge in the LM and (2) contrasting the representations of the general and the full (both general and domain knowledge) to learn an integrated representation with both general and domain-specific knowledge. Experimental results will demonstrate the effectiveness of the proposed approach.