DyMO: Training-Free Diffusion Model Alignment with Dynamic Multi-Objective Scheduling

16
citations
#512
in CVPR 2025
of 2873 papers
2
Top Authors
7
Data Points

Top Authors

Abstract

Text-to-image diffusion model alignment is critical for improving the alignment between the generated images and human preferences. While training-based methods are constrained by high computational costs and dataset requirements, training-free alignment methods remain underexplored and are often limited by inaccurate guidance. We propose a plug-and-play training-free alignment method, DyMO, for aligning the generated images and human preferences during inference. Apart from text-aware human preference scores, we introduce a semantic alignment objective for enhancing the semantic alignment in the early stages of diffusion, relying on the fact that the attention maps are effective reflections of the semantics in noisy images. We propose dynamic scheduling of multiple objectives and intermediate recurrent steps to reflect the requirements at different steps. Experiments with diverse pre-trained diffusion models and metrics demonstrate the effectiveness and robustness of the proposed method.

Citation History

Jan 26, 2026
0
Jan 27, 2026
0
Jan 27, 2026
0
Feb 2, 2026
14+14
Feb 13, 2026
16+2
Feb 13, 2026
16
Feb 13, 2026
16