通过跨批次优化噪声-数据配对，提升流模型采样速度与质量。
Faster Inference of Flow-Based Generative Models via Improved Data-Noise Coupling
- 利用全局优化的噪声-数据配对，突破单批次限制
- 在多个数据集上实现更快采样且不损失生成质量
- 适合需要高效生成的图像视频任务开发者
条件流匹配（CFM）是一种无需模拟的连续归一化流训练方法，为图像和视频生成等关键任务提供了比扩散模型更高效的替代方案。CFM 的性能取决于数据与噪声的耦合方式。近期方法采用小批量最优传输（minibatch OT）在每轮训练中重新分配噪声-数据对，以简化采样路径并加速推理。然而，该方法仅限于单个小批量优化，难以在大规模数据集上发挥效能。为此，本文提出 LOOM-CFM（Looking Out Of Minibatch-CFM），通过跨训练过程保持并优化这些配对关系，扩展了小批量 OT 的作用范围。实验表明，该方法在多个数据集上均显著提升了采样速度与生成质量的权衡表现。此外，LOOM-CFM 还改善了蒸馏初始化效果，并支持潜空间训练中的高分辨率合成。
原文摘要 · Abstract (English)
Conditional Flow Matching (CFM), a simulation-free method for training continuous normalizing flows, provides an efficient alternative to diffusion models for key tasks like image and video generation. The performance of CFM in solving these tasks depends on the way data is coupled with noise. A recent approach uses minibatch optimal transport (OT) to reassign noise-data pairs in each training step to streamline sampling trajectories and thus accelerate inference. However, its optimization is restricted to individual minibatches, limiting its effectiveness on large datasets. To address this shortcoming, we introduce LOOM-CFM (Looking Out Of Minibatch-CFM), a novel method to extend the scope of minibatch OT by preserving and optimizing these assignments across minibatches over training time. Our approach demonstrates consistent improvements in the sampling speed-quality trade-off across multiple datasets. LOOM-CFM also enhances distillation initialization and supports high-resolution synthesis in latent space training.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。