arXiv 2505.13447
Mean Flows for One-step Generative Modeling
By Zhengyang Geng, Mingyang Deng, et al.
Published 2025-05-19
Mindmap
Browse the paper's core ideas, clusters, and relationships in a structured outline.
We propose a principled and effective framework for one-step generative modeling. We introduce the notion of average velocity to characterize flow fields, in contrast to instantaneous velocity modeled by Flow Matching methods. A well-defined identity between average and instantaneous velocities is derived and used to guide neural network training. Our method, termed the MeanFlow model, is self-contained and requires…