Merge then Realign: Simple and Effective Modality-Incremental Continual Learning for Multimodal LLMs
원문 초록
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 necessary to reuse the existing ones and further extend them to more modalities through Modality-incremental Continual Learning (MCL). However, this often comes with a performance degradation in the previously learned modalities. In this work, we revisit the MCL and investigate a more severe issue it faces in contrast to traditional continual learning, that its degradation comes not only from catastrophic forgetting but also from the misalignment between the modality-agnostic and modality-specific components. To address this problem, we propose an elegantly simple MCL paradigm called "MErge then ReAlign" (MERA). Our method avoids introducing heavy training overhead or modifying the model architecture, hence is easy to deploy and highly reusable in the MLLM community. Extensive experiments demonstrate that, despite the simplicity of MERA, it shows impressive performance, holding up to a 99.84% Backward Relative Gain when extending to four modalities, achieving a nearly lossless MCL performance.
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
Questioning Representational Optimism in Deep Learning: The Fractured Entangled Representation Hypothesis
LifelongPR: Lifelong knowledge fusion for point cloud place recognition based on replay and prompt learning
Rethinking Query-based Transformer for Continual Image Segmentation
Addressing Imbalanced Domain-Incremental Learning through Dual-Balance Collaborative Experts
PSAT: Pediatric Segmentation Approaches via Adult Augmentations and Transfer Learning
Remember Past, Anticipate Future: Learning Continual Multimodal Misinformation Detectors
토론
이 논문과 연결된 포럼 글이 아직 없습니다.