arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2405.20320cs.CVcs.AIcs.LG
收藏

改进整流流(Rectified Flows)的训练

Improving the Training of Rectified Flows

Carnegie Mellon University(卡内基梅隆大学);Microsoft Research(微软研究院)

Sangyun Lee, Zinan Lin, Giulia Fanti

AI总结:

针对整流流采样需较多函数评估的问题,本研究提出U形时间步分布、LPIPS-Huber预度量等单轮训练改进技术,在低NFE设置下实现了优于主流蒸馏方法的生成性能。

摘要:

扩散模型在图像和视频生成领域展现出巨大潜力,但从当前最优模型中采样需要对生成式常微分方程(ODE)进行计算成本高昂的数值积分。解决这一问题的一种方法是整流流(rectified flows),它通过迭代学习平滑的ODE路径,降低截断误差的影响。然而,整流流仍然需要相对较多的函数评估次数(NFEs)。在本研究中,我们提出了改进的整流流训练技术,使其即使在低NFE设置下也能与知识蒸馏(knowledge distillation)方法相媲美。我们的核心发现是,在实际场景下,仅需一次Reflow算法迭代训练整流流就足以学习到近乎笔直的轨迹,因此当前使用多次Reflow迭代的做法是不必要的。据此,我们提出了改进整流流单轮训练的技术,包括U形时间步分布和LPIPS-Huber预度量。借助这些技术,在CIFAR-10数据集的1 NFE设置下,我们将此前2-整流流的FID指标提升了高达75%。在64×64分辨率的ImageNet数据集上,我们改进后的整流流在单步和两步设置下均优于一致性蒸馏、渐进蒸馏等当前最优的蒸馏方法,且在FID指标上可与改进一致性训练(iCT)的性能相匹敌。代码可在https://github.com/sangyun884/rfpp获取。

英文摘要:

Diffusion models have shown great promise for image and video generation, but sampling from state-of-the-art models requires expensive numerical integration of a generative ODE. One approach for tackling this problem is rectified flows, which iteratively learn smooth ODE paths that are less susceptible to truncation error. However, rectified flows still require a relatively large number of function evaluations (NFEs). In this work, we propose improved techniques for training rectified flows, allowing them to compete with \emph{knowledge distillation} methods even in the low NFE setting. Our main insight is that under realistic settings, a single iteration of the Reflow algorithm for training rectified flows is sufficient to learn nearly straight trajectories; hence, the current practice of using multiple Reflow iterations is unnecessary. We thus propose techniques to improve one-round training of rectified flows, including a U-shaped timestep distribution and LPIPS-Huber premetric. With these techniques, we improve the FID of the previous 2-rectified flow by up to 75\% in the 1 NFE setting on CIFAR-10. On ImageNet 64$\times$64, our improved rectified flow outperforms the state-of-the-art distillation methods such as consistency distillation and progressive distillation in both one-step and two-step settings and rivals the performance of improved consistency training (iCT) in FID. Code is available at https://github.com/sangyun884/rfpp.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑