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 2605.16147cs.CV
收藏

注册信息对像素空间扩散变换器的重要性

Registers Matter for Pixel-Space Diffusion Transformers

Yandex Research(Yandex研究院)

Nikita Starodubcev, Ilia Sudakov, Ilya Drobyshevskiy, Artem Babenko, Dmitry Baranchuk

AI总结:

研究发现扩散变换器与视觉变换器在处理像素空间时存在差异,注册令牌显著提升了生成质量,通过分析中间表示发现其在高噪声水平下产生更清晰的特征图,进而提出了一种高效的双流架构以提升生成效果。

摘要:

视觉变换器(ViTs)已知会表现出高范数补丁-令牌异常,这会降低特征图的质量,这一问题通过注册令牌得到有效的缓解。随着扩散模型越来越多地采用变换器架构并朝向像素空间训练迈进,它们在形式上越来越接近ViTs,从而引发了注册令牌是否也对扩散变换器(DiTs)有用的问题。在本文中,我们发现DiTs与ViTs在关键方面存在差异:它们不表现出补丁-令牌异常。有趣的是,注册令牌显著提高了像素空间DiTs的收敛性和生成质量。通过分析中间表示,我们发现注册令牌在高噪声水平下会产生更清晰的特征图,这可能解释了它们在像素空间生成中的有效性。我们进一步观察到,最近的像素空间DiT架构隐式地包含了注册-like机制,这可能部分解释了其强大的经验表现。受这些见解的启发,我们研究了一种参数高效的双流架构,专门处理注册令牌,并通过几乎不增加运行时间开销的方式提高了像素空间生成质量。

英文摘要:

Vision Transformers (ViTs) are known to exhibit high-norm patch-token outliers that degrade feature map quality, a problem effectively mitigated by register tokens. As diffusion models increasingly adopt transformer architectures and move toward pixel-space training, they become closer in form to ViTs, raising the question of whether register tokens are also useful for Diffusion Transformers (DiTs). In this work, we show that DiTs differ from ViTs in a key respect: they do not exhibit patch-token outliers but still benefit from registers. Interestingly, registers are more effective in pixel-space DiTs than in latent-space DiTs. By analyzing intermediate representations, we find that register tokens produce cleaner feature maps at high noise levels, which may contribute to their effectiveness in pixel-space generation. We further observe that recent pixel-space DiT architectures implicitly incorporate register-like mechanisms, which may partially account for their strong empirical performance. Motivated by these observations, we propose Register Guidance, a technique that amplifies the contribution of register tokens responsible for improving visual structure and coherence.

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。

↑