Adapting a Language Model While Preserving its General Knowledge
This addresses the issue of knowledge loss during domain adaptation for language models, which is incremental but important for applications requiring both general and specialized knowledge.
The paper tackles the problem of domain-adaptive pre-training for language models, which can degrade general knowledge, by proposing a method that soft-masks attention heads and uses contrastive learning to preserve general knowledge while integrating domain-specific knowledge, resulting in improved performance on end-tasks.
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.