Controlled Low-Rank Adaptation with Subspace Regularization for Continued Training on Large Language Models

TL;DR

This paper introduces CLoRA, a subspace regularization technique for Low-Rank Adaptation in large language models, which effectively reduces catastrophic forgetting during continual learning while maintaining model capacity.

Contribution

It proposes a novel regularization method, CLoRA, that constrains null space updates to mitigate forgetting with minimal impact on model capacity.

Findings

CLoRA outperforms existing methods in finetuning tasks.

It effectively reduces catastrophic forgetting in continual learning.

Balances model capacity and forgetting mitigation.

Abstract

Large language models (LLMs) exhibit remarkable capabilities in natural language processing but face catastrophic forgetting when learning new tasks, where adaptation to a new domain leads to a substantial decline in performance on previous tasks. In this paper, we propose Controlled LoRA (CLoRA), a sub-space regularization method on LoRA structure. Aiming to reduce the scale of output change while introduce minimal constraint on model capacity, CLoRA imposes constraint on the direction of updating matrix's null space. Experimental results on one-stage LLM finetuning tasks and continual learning settings highlight the superority of CLoRA as a effective parameter efficient finetuning method with catastrophic forgetting mitigating.Further investigation for model parameters indicates that CLoRA effectively balances the trade-off between model capacity and degree of forgetting.