vix.ing · top · new · best · stats · spec

Merge then Realign: Simple and Effective Modality-Incremental Continual Learning for Multimodal LLMs

2025/03/08 by Dingkun Zhang, Zhang, Dingkun, Shuhan Qi +7 · 1 citation
Computer Science · #Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Natural Language Processing Techniques #Text and Document Classification Technologies

paper · pdf · doi:10.48550/arxiv.2503.07663

openalex publication_date 2025/03/08 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Recent advances in Multimodal Large Language Models (MLLMs) have enhanced their versatility as they integrate a growing number of modalities. Considering the heavy cost of training MLLMs, it is efficient to reuse the existing ones and extend them to more modalities through Modality-incremental Continual Learning (MCL). The exploration of MCL is in its early stages. In this work, we dive into the causes of performance degradation in MCL. We uncover that it suffers not only from forgetting as in traditional continual learning, but also from misalignment between the modality-agnostic and modality-specific components. To this end, we propose an elegantly simple MCL paradigm called "MErge then ReAlign" (MERA) to address both forgetting and misalignment. MERA avoids introducing heavy model budgets or modifying model architectures, hence is easy to deploy and highly reusable in the MLLM community. Extensive experiments demonstrate the impressive performance of MERA, holding an average of 99.84% Backward Relative Gain when extending to four modalities, achieving nearly lossless MCL performance. Our findings underscore the misalignment issue in MCL. More broadly, our work showcases how to adjust different components of MLLMs during continual learning.

Cited by

Related