arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2401.01520cs.CVcs.LGeess.IV
收藏

S²-DMs:跳步扩散模型

S$^{2}$-DMs:Skip-Step Diffusion Models

South China Normal University(华南师范大学)

Yixuan Wang, Shuangyin Li

AI总结:

针对扩散模型训练与采样步数不对称导致未采样步骤信息丢失的问题,提出带 L_skip 损失的 S²-DMs 训练方法,在 CIFAR10 和 CELEBA 上显著提升多种采样器的生成质量。

摘要:

扩散模型已成为强大的生成工具,在样本质量上可与 GAN 媲美,并能反映自回归模型的似然分数。这些模型中的一个子类(以 DDIM 为代表)表现出一种固有的不对称性:它们在 T 个步骤上进行训练,但在生成过程中仅从 T 的一个子集中采样。这种选择性采样方法虽然针对速度进行了优化,却无意中遗漏了来自未采样步骤的关键信息,从而可能损害样本质量。为解决这一问题,我们提出了 S²-DMs,这是一种采用创新 L_skip 的全新训练方法,经过精心设计以重新整合选择性采样阶段被遗漏的信息。该方法具有多方面优势:显著提升样本质量,实现极其简单,只需极少的代码修改,并且足够灵活,可兼容多种采样算法。在 CIFAR10 数据集上,使用我们的算法训练的模型在各种采样算法(DDIM、PNDM、DEIS)和不同采样步数(10、20、……、1000)下,相比使用传统方法训练的模型提升了 3.27% 至 14.06%。在 CELEBA 数据集上,提升幅度为 8.97% 至 27.08%。代码和更多资源可在 GitHub 上获取。

英文摘要:

Diffusion models have emerged as powerful generative tools, rivaling GANs in sample quality and mirroring the likelihood scores of autoregressive models. A subset of these models, exemplified by DDIMs, exhibit an inherent asymmetry: they are trained over $T$ steps but only sample from a subset of $T$ during generation. This selective sampling approach, though optimized for speed, inadvertently misses out on vital information from the unsampled steps, leading to potential compromises in sample quality. To address this issue, we present the S$^{2}$-DMs, which is a new training method by using an innovative $L_{skip}$, meticulously designed to reintegrate the information omitted during the selective sampling phase. The benefits of this approach are manifold: it notably enhances sample quality, is exceptionally simple to implement, requires minimal code modifications, and is flexible enough to be compatible with various sampling algorithms. On the CIFAR10 dataset, models trained using our algorithm showed an improvement of 3.27% to 14.06% over models trained with traditional methods across various sampling algorithms (DDIMs, PNDMs, DEIS) and different numbers of sampling steps (10, 20, ..., 1000). On the CELEBA dataset, the improvement ranged from 8.97% to 27.08%. Access to the code and additional resources is provided in the github.

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。

↑