Rehearsal-Free Modular and Compositional Continual Learning for Language Models
This addresses the problem of privacy and memory issues in rehearsal-based methods and lack of knowledge transfer in parameter-isolation approaches for continual learning in language models.
The authors tackled catastrophic forgetting in continual learning for language models by proposing MoCL, a rehearsal-free modular and compositional framework that adds new modules and composes them with existing ones, achieving state-of-the-art performance on various benchmarks.
Continual learning aims at incrementally acquiring new knowledge while not forgetting existing knowledge. To overcome catastrophic forgetting, methods are either rehearsal-based, i.e., store data examples from previous tasks for data replay, or isolate parameters dedicated to each task. However, rehearsal-based methods raise privacy and memory issues, and parameter-isolation continual learning does not consider interaction between tasks, thus hindering knowledge transfer. In this work, we propose MoCL, a rehearsal-free Modular and Compositional Continual Learning framework which continually adds new modules to language models and composes them with existing modules. Experiments on various benchmarks show that MoCL outperforms state of the art and effectively facilitates knowledge transfer.