多模态持续指令微调与动态梯度指导
Multimodal Continual Instruction Tuning with Dynamic Gradient Guidance
AI总结：
本文提出通过几何参数空间方向向量近似缺失梯度，结合有限回放缓冲区和伯努利采样策略，有效缓解多模态持续指令微调中的灾难性遗忘问题。
摘要：
多模态持续指令微调使多模态大语言模型能够依次适应新任务，同时基于已获得的知识。然而，这种持续学习范式面临灾难性遗忘的显著挑战，即学习新任务会导致对先前任务的性能下降。在本文中，我们通过将灾难性遗忘概念化为在学习新任务时旧任务梯度缺失的问题，引入了新的见解。我们的方法通过利用参数空间的几何特性，具体使用当前参数与之前最优参数之间的方向向量作为梯度指导，近似这些缺失的梯度。此近似梯度可以进一步与有限回放缓冲区中的真实梯度结合，并通过伯努利采样策略动态平衡模型的稳定性与可塑性。在多模态持续指令微调数据集上的广泛实验表明，我们的方法在不扩展模型的情况下实现了最先进的性能，有效缓解了灾难性遗忘，同时保持了紧凑的架构。
英文摘要：
Multimodal continual instruction tuning enables multimodal large language models to sequentially adapt to new tasks while building upon previously acquired knowledge. However, this continual learning paradigm faces the significant challenge of catastrophic forgetting, where learning new tasks leads to performance degradation on previous ones. In this paper, we introduce a novel insight into catastrophic forgetting by conceptualizing it as a problem of missing gradients from old tasks during new task learning. Our approach approximates these missing gradients by leveraging the geometric properties of the parameter space, specifically using the directional vector between current parameters and previously optimal parameters as gradient guidance. This approximated gradient can be further integrated with real gradients from a limited replay buffer and regulated by a Bernoulli sampling strategy that dynamically balances model stability and plasticity. Extensive experiments on multimodal continual instruction tuning datasets demonstrate that our method achieves state-of-the-art performance without model expansion, effectively mitigating catastrophic forgetting while maintaining a compact architecture.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
