Mixture-of-Subspaces in Low-Rank Adaptation
Abstract: In this paper, we introduce a subspace-inspired Low-Rank Adaptation (LoRA) method, which is computationally efficient, easy to implement, and readily applicable to large language, multimodal, and diffusion models.
Initially, we equivalently decompose the weights of LoRA into two subspaces, and find that simply mixing them can enhance performance.
To study such a phenomenon, we revisit it through a fine-grained subspace lens, showing that such modification is equivalent to employing a fixed mixer to fuse the subspaces.
To be more flexible, we jointly learn the mixer with the original LoRA weights, and term the method as Mixture-of-Subspaces LoRA (MoSLoRA).
MoSLoRA consistently outperforms LoRA on tasks in different modalities, including commonsense reasoning, visual instruction tuning, and subject-driven text-to-image generation, demonstrating its effectiveness and robustness.
Paper Type: Long
Research Area: Language Modeling
Research Area Keywords: data-efficient fine-tuning
Contribution Types: Approaches to low-resource settings, Approaches low compute settings-efficiency
Languages Studied: English
Submission Number: 5722
Loading