Continual Training of Language Models for Few-Shot Learning
This addresses the need for adaptable language models in various domains without forgetting prior skills, though it appears incremental as it builds on existing post-training methods.
The paper tackles the problem of expanding language model knowledge by incrementally post-training on multiple unlabeled domain corpora to improve few-shot learning, achieving verified effectiveness in experiments.
Recent work on applying large language models (LMs) achieves impressive performance in many NLP applications. Adapting or posttraining an LM using an unlabeled domain corpus can produce even better performance for end-tasks in the domain. This paper proposes the problem of continually extending an LM by incrementally post-train the LM with a sequence of unlabeled domain corpora to expand its knowledge without forgetting its previous skills. The goal is to improve the few-shot end-task learning in these domains. The resulting system is called CPT (Continual PostTraining), which to our knowledge, is the first continual post-training system. Experimental results verify its effectiveness.