Abstract
Currently, applying diffusion models in pixel space of high resolution images is difficult. Instead, existing approaches focus on diffusion in lower dimensional spaces (latent diffusion), or have multiple super-resolution levels of generation referred to as cascades. The downside is that these approaches add additional complexity to the diffusion framework. This paper aims to improve denoising diffusion for high resolution images while keeping the model as simple as possible. The paper is centered around the research question: How can one train a standard denoising diffusion models on high resolution images, and still obtain performance comparable to these alternate approaches? The four main findings are: 1) the noise schedule should be adjusted for high resolution images, 2) It is sufficient to scale only a particular part of the architecture, 3) dropout should be added at specific locations in the architecture, and 4) downsampling is an effective strategy to avoid high resolution feature maps. Combining these simple yet effective techniques, we achieve state-of-the-art on image generation among diffusion models without sampling modifiers on ImageNet.
Abstract (translated)
目前,在高分辨率图像的像素空间应用扩散模型比较困难。相反,现有的方法主要关注扩散在较低维度空间(潜在扩散)或具有多个超分辨率级别生成的现象,被称为 cascades。这种方法的缺点在于,它们增加了扩散框架的复杂性。本文的目标是改善高分辨率图像的去噪扩散,同时保持模型尽可能简单。本文的核心问题是:如何在高分辨率图像上训练标准去噪扩散模型,同时仍然获得与这些替代方法相当的性能?这四个主要发现是:1) 对高分辨率图像进行调整噪声时间表是必要的,2)仅扩大特定部分架构即可,3)在架构中添加 dropout 是有效的策略,4)减少采样是避免高分辨率特征图的有效策略。将这些方法简单但有效的技术结合起来,我们在扩散模型中生成图像的技术领域达到当前最高水平,而在 ImageNet 上无需采样增强器。
URL
https://arxiv.org/abs/2301.11093
