SMoLoRA: Exploring and Defying Dual Catastrophic Forgetting in Continual Visual Instruction Tuning

TL;DR

This paper introduces SMoLoRA, a novel framework with dual modules for visual understanding and instruction following, effectively mitigating dual catastrophic forgetting in continual visual instruction tuning of multimodal models.

Contribution

The paper proposes SMoLoRA, a dual-routing adaptation method that addresses dual forgetting in CVIT and introduces a new benchmark for evaluating generalization and instruction robustness.

Findings

SMoLoRA outperforms existing methods in mitigating dual forgetting.

It improves generalization to unseen tasks.

It enhances robustness in following diverse instructions.

Abstract

Visual instruction tuning (VIT) enables multimodal large language models (MLLMs) to effectively handle a wide range of vision tasks by framing them as language-based instructions. Building on this, continual visual instruction tuning (CVIT) extends the capability of MLLMs to incrementally learn new tasks, accommodating evolving functionalities. While prior work has advanced CVIT through the development of new benchmarks and approaches to mitigate catastrophic forgetting, these efforts largely follow traditional continual learning paradigms, neglecting the unique challenges specific to CVIT. We identify a dual form of catastrophic forgetting in CVIT, where MLLMs not only forget previously learned visual understanding but also experience a decline in instruction following abilities as they acquire new tasks. To address this, we introduce the Separable Mixture of Low-Rank Adaptation (SMoLoRA) framework, which employs separable routing through two distinct modules-one for visual understanding and another for instruction following. This dual-routing design enables specialized adaptation in both domains, preventing forgetting while improving performance. Furthermore, we propose a new CVIT benchmark that goes beyond existing benchmarks by additionally evaluating a model's ability to generalize to unseen tasks and handle diverse instructions across various tasks. Extensive experiments demonstrate that SMoLoRA outperforms existing methods in mitigating dual forgetting, improving generalization to unseen tasks, and ensuring robustness in following diverse instructions. Code is available at https://github.com/Minato-Zackie/SMoLoRA.