NoRA: Nested Low-Rank Adaptation for Efficient Fine-Tuning Large Models
This work addresses the need for more efficient fine-tuning methods for large models, offering a novel approach that is incremental but provides specific gains in parameter reduction and task performance.
The paper tackles the problem of parameter-efficient fine-tuning in large models by introducing NoRA, a nested low-rank adaptation method that reduces tunable parameters while improving task adaptation, achieving superior performance over LoRA and its variants on tasks like commonsense reasoning and vision-language fine-tuning.
In this paper, we introduce Nested Low-Rank Adaptation (NoRA), a novel approach to parameter-efficient fine-tuning that extends the capabilities of Low-Rank Adaptation (LoRA) techniques. Vanilla LoRA overlooks pre-trained weight inheritance and still requires fine-tuning numerous parameters. To addresses these issues, our NoRA adopts a dual-layer nested structure with Singular Value Decomposition (SVD), effectively leveraging original matrix knowledge while reducing tunable parameters. Specifically, NoRA freezes the outer LoRA weights and utilizes an inner LoRA design, providing enhanced control over model optimization. This approach allows the model to more precisely adapt to specific tasks while maintaining a compact parameter space. By freezing outer LoRA weights and using an inner LoRA design, NoRA enables precise task adaptation with a compact parameter space. Evaluations on tasks including commonsense reasoning with large language models, fine-tuning vision-language models, and subject-driven generation demonstrate NoRA's superiority over LoRA and its variants. Code will be released upon acceptance.