RECALL: REpresentation-aligned Catastrophic-forgetting ALLeviation via Hierarchical Model Merging
This work addresses the challenge of updating LLMs over time without forgetting previous knowledge, offering a scalable, data-free solution for continual learning in NLP.
The authors tackled the problem of catastrophic forgetting in continual learning for large language models by proposing RECALL, a representation-aware model merging framework that aligns knowledge across models without needing historical data, achieving superior performance in knowledge retention and generalization across five NLP tasks.
We unveil that internal representations in large language models (LLMs) serve as reliable proxies of learned knowledge, and propose RECALL, a novel representation-aware model merging framework for continual learning without access to historical data. RECALL computes inter-model similarity from layer-wise hidden representations over clustered typical samples, and performs adaptive, hierarchical parameter fusion to align knowledge across models. This design enables the preservation of domain-general features in shallow layers while allowing task-specific adaptation in deeper layers. Unlike prior methods that require task labels or incur performance trade-offs, RECALL achieves seamless multi-domain integration and strong resistance to catastrophic forgetting. Extensive experiments across five NLP tasks and multiple continual learning scenarios show that RECALL outperforms baselines in both knowledge retention and generalization, providing a scalable and data-free solution for evolving LLMs.