Not All Layers Are Created Equal: Adaptive LoRA Ranks for Personalized Image Generation

TL;DR

This paper introduces LoRA$^2$, an adaptive method for fine-tuning diffusion models that automatically adjusts layer ranks during training, improving personalization quality while reducing memory use.

Contribution

It proposes a novel adaptive rank selection mechanism for LoRA, allowing each layer's rank to be learned during fine-tuning, unlike fixed-rank approaches.

Findings

Achieves competitive performance with lower ranks and memory usage.

Outperforms fixed-rank LoRA across 29 subjects.

Reduces the need for manual rank tuning.

Abstract

Low Rank Adaptation (LoRA) is the de facto fine-tuning strategy to generate personalized images from pre-trained diffusion models. Choosing a good rank is extremely critical, since it trades off performance and memory consumption, but today the decision is often left to the community's consensus, regardless of the personalized subject's complexity. The reason is evident: the cost of selecting a good rank for each LoRA component is combinatorial, so we opt for practical shortcuts such as fixing the same rank for all components. In this paper, we take a first step to overcome this challenge. Inspired by variational methods that learn an adaptive width of neural networks, we let the ranks of each layer freely adapt during fine-tuning on a subject. We achieve it by imposing an ordering of importance on the rank's positions, effectively encouraging the creation of higher ranks when strictly needed. Qualitatively and quantitatively, our approach, LoRA$^2$, achieves a competitive trade-off between DINO, CLIP-I, and CLIP-T across 29 subjects while requiring much less memory and lower rank than high rank LoRA versions. Code: https://github.com/donaldssh/NotAllLayersAreCreatedEqual.