# Cached Multi-Lora Composition for Multi-Concept Image Generation

arXiv: https://arxiv.org/abs/2502.04923 · HTML: https://mlsanity.com/paper/2502.04923/

## Key facts

- Cited by 13 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 7 Feb 2025.

## Built on by (most cited first, 13 of 13)

- [QR-LoRA: Efficient and Disentangled Fine-tuning via QR Decomposition for Customized Generation](https://mlsanity.com/paper/2507.04599/) (2025)
- [Diffusion Blend: Inference-Time Multi-Preference Alignment for Diffusion Models](https://mlsanity.com/paper/2505.18547/) (2025)
- [MultLFG: Training-free Multi-LoRA composition using Frequency-domain Guidance](https://mlsanity.com/paper/2505.20525/) (2025)
- [LoRAtorio: An intrinsic approach to LoRA Skill Composition](https://mlsanity.com/paper/2508.11624/) (2025)
- [Preference-Aligned LoRA Merging: Preserving Subspace Coverage and Addressing Directional Anisotropy](https://mlsanity.com/paper/2603.26299/) (2026)
- [Semantic-guided LoRA Parameters Generation](https://mlsanity.com/paper/2509.10535/) (2025)
- [VMDiff: Visual Mixing Diffusion for Limitless Cross-Object Synthesis](https://mlsanity.com/paper/2509.23605/) (2025)
- [Training-Free Multi-Concept Image Editing](https://mlsanity.com/paper/2602.20839/) (2026)
- [FREE-Switch: Frequency-based Dynamic LoRA Switch for Style Transfer](https://mlsanity.com/paper/2604.10023/) (2026)
- [Training-Free Multi-Concept LoRA Composition with Prompt-Aware Weighting](https://mlsanity.com/paper/2606.03792/) (2026)
- [SDO: Subspace Deconflicting Operator for Multi-Adapter Composition](https://mlsanity.com/paper/2608.13820/) (2026)
- [Frequency-Aware Continual Learning for Smart Contract Vulnerability Detection with Large Language Models](https://mlsanity.com/paper/2608.19680/) (2026)
- [PoEM: Predicting RL Outcomes from Existing Policies](https://mlsanity.com/paper/2609.30226/) (2026)

## Abstract (arXiv)

Low-Rank Adaptation (LoRA) has emerged as a widely adopted technique in text-to-image models, enabling precise rendering of multiple distinct elements, such as characters and styles, in multi-concept image generation. However, current approaches face significant challenges when composing these LoRAs for multi-concept image generation, resulting in diminished generated image quality. In this paper, we initially investigate the role of LoRAs in the denoising process through the lens of the Fourier frequency domain. Based on the hypothesis that applying multiple LoRAs could lead to "semantic conflicts", we find that certain LoRAs amplify high-frequency features such as edges and textures, whereas others mainly focus on low-frequency elements, including the overall structure and smooth color gradients. Building on these insights, we devise a frequency domain based sequencing strategy to determine the optimal order in which LoRAs should be integrated during inference. This strategy offers a methodical and generalizable solution compared to the naive integration commonly found in existing LoRA fusion techniques. To fully leverage our proposed LoRA order sequence determination method in multi-LoRA composition tasks, we introduce a novel, training-free framework, Cached Multi-LoRA (CMLoRA), designed to efficiently integrate multiple LoRAs while maintaining cohesive image generation. With its flexible backbone for multi-LoRA fusion and a non-uniform caching strategy tailored to individual LoRAs, CMLoRA has the potential to reduce semantic conflicts in LoRA composition and improve computational efficiency. Our experimental evaluations demonstrate that CMLoRA outperforms state-of-the-art training-free LoRA fusion methods by a significant margin -- it achieves an average improvement of 2.19% in CLIPScore, and 11.25% in MLLM win rate compared to LoraHub, LoRA Composite, and LoRA Switch.