SciRE-Solver:通过基于递归差分的Score-integrand求解器加速扩散模型采样
SciRE-Solver: Accelerating Diffusion Models Sampling by Score-integrand Solver with Recursive Difference
AI总结:
本文提出基于递归差分(RD)方法的SciRE-Solver和SciREI-Solver,通过计算score函数导数并利用截断Taylor展开,在多种预训练扩散模型上以极少的NFE实现了SOTA的FID。
摘要:
扩散模型(DMs)在图像、音频和视频生成领域取得了显著进展。DMs的一个缺点是其缓慢的迭代过程。近期用于快速采样的算法是从微分方程的角度设计的。然而,在基于Taylor展开的高阶算法中,由于大规模、训练良好的神经网络的复杂性,估计score函数的导数变得难以处理。受此动机驱使,在这项工作中,我们引入了递归差分(RD)方法来计算DMs领域中score函数的导数。基于RD方法和score-integrand的截断Taylor展开,我们提出了SciRE-Solver,具有收敛阶保证,用于加速DMs的采样。为了进一步研究RD方法的有效性,我们还提出了一个名为SciREI-Solver的变体,它基于RD方法和指数积分器。我们提出的带有RD方法的采样算法在与现有的免训练采样算法的比较中,在离散时间和连续时间预训练DMs上,在各种score函数评估次数(NFE)下,均获得了最先进(SOTA)的FIDs。值得注意的是,使用少量NFEs的SciRE-Solver展现出超越某些预训练模型在其原始论文中使用不少于$1000$ NFEs所达到FID的有望潜力。例如,在CIFAR-10上,对于连续时间DM,我们以$100$ NFE达到了$2.40$ FID的SOTA值;对于离散时间DM,以$84$ NFE达到了$3.15$ FID。在CelebA 64$\ imes$64上,对于离散时间DM,以$18$($50$)NFE达到了$2.17$($2.02$)FID。
英文摘要:
Diffusion models (DMs) have made significant progress in the fields of image, audio, and video generation. One downside of DMs is their slow iterative process. Recent algorithms for fast sampling are designed from the perspective of differential equations. However, in higher-order algorithms based on Taylor expansion, estimating the derivative of the score function becomes intractable due to the complexity of large-scale, well-trained neural networks. Driven by this motivation, in this work, we introduce the recursive difference (RD) method to calculate the derivative of the score function in the realm of DMs. Based on the RD method and the truncated Taylor expansion of score-integrand, we propose SciRE-Solver with the convergence order guarantee for accelerating sampling of DMs. To further investigate the effectiveness of the RD method, we also propose a variant named SciREI-Solver based on the RD method and exponential integrator. Our proposed sampling algorithms with RD method attain state-of-the-art (SOTA) FIDs in comparison to existing training-free sampling algorithms, across both discrete-time and continuous-time pre-trained DMs, under various number of score function evaluations (NFE). Remarkably, SciRE-Solver using a small NFEs demonstrates promising potential to surpass the FID achieved by some pre-trained models in their original papers using no fewer than $1000$ NFEs. For example, we reach SOTA value of $2.40$ FID with $100$ NFE for continuous-time DM and of $3.15$ FID with $84$ NFE for discrete-time DM on CIFAR-10, as well as of $2.17$ (2.02) FID with $18$ (50) NFE for discrete-time DM on CelebA 64$\times$64.
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。