Consistency Flow Matching: Defining Straight Flows with Velocity Consistency

Ling Yang,Zixiang Zhang,Zhilong Zhang,Xingchao Liu,Minkai Xu,Wentao Zhang,Chenlin Meng,Stefano Ermon,Bin Cui

2024-07-03

Abstract:Flow matching (FM) is a general framework for defining probability paths via Ordinary Differential Equations (ODEs) to transform between noise and data samples. Recent approaches attempt to straighten these flow trajectories to generate high-quality samples with fewer function evaluations, typically through iterative rectification methods or optimal transport solutions. In this paper, we introduce Consistency Flow Matching (Consistency-FM), a novel FM method that explicitly enforces self-consistency in the velocity field. Consistency-FM directly defines straight flows starting from different times to the same endpoint, imposing constraints on their velocity values. Additionally, we propose a multi-segment training approach for Consistency-FM to enhance expressiveness, achieving a better trade-off between sampling quality and speed. Preliminary experiments demonstrate that our Consistency-FM significantly improves training efficiency by converging 4.4x faster than consistency models and 1.7x faster than rectified flow models while achieving better generation quality. Our code is available at: <a class="link-external link-https" href="https://github.com/YangLing0818/consistency_flow_matching" rel="external noopener nofollow">this https URL</a>

Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

The paper presents a new method called Consistency Flow Matching (Consistency-FM) to address the problem of probabilistic flow linearization for improved efficiency and quality in noise-to-sample transformation. Current methods such as iterative correction and optimization transport solutions suffer from cumulative errors or high computational complexity.

Within the framework of Continuous Normalizing Flows (CNFs), Flow Matching defines probabilistic paths through ordinary differential equations (ODEs). Straight-line paths are ideal as they are the shortest and easiest to simulate. The introduced Consistency-FM directly defines straight-line flows from different time points to the same end point by enforcing self-consistency in the velocity field and imposing constraints on the velocity values of these flows. Furthermore, the expressive power of Consistency-FM is enhanced through a multi-stage training approach to achieve better transfer between complex distributions.

Compared to existing methods, Consistency-FM improves training efficiency, with convergence speed 4.4 times faster than consistency models and 1.7 times faster than correction flow models, while maintaining better generation quality. Experimental results demonstrate the outstanding performance of Consistency-FM in image generation tasks, particularly for high-resolution images, highlighting its potential in modeling complex data distributions.

In summary, this paper attempts to address the problem of improving training efficiency and sample quality in generative models through linearized flows, proposing a novel flow matching method with self-consistency.