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 2410.22366cs.LGcs.AIcs.CV
收藏

一步足矣:面向文本生成图像扩散模型的稀疏自编码器

One-Step is Enough: Sparse Autoencoders for Text-to-Image Diffusion Models

EPFL(洛桑联邦理工学院);Northeastern University(东北大学)

Viacheslav Surkov, Chris Wendler, Antonio Mari, Mikhail Terekhov, Justin Deschenaux, Robert West, Caglar Gulcehre, David Bau

AI总结:

该研究首次将稀疏自编码器(SAE)用于文本生成图像扩散模型的可解释性研究,在SDXL Turbo的1步设置Transformer块更新上训练SAE,发现其可跨步数、跨模型泛化,还构建RIEBench基准验证特征可解释性与因果影响,为理解操控文生图模型内部机制提供新途径。

摘要:

对于大语言模型(LLM),稀疏自编码器(SAE)已被证明可将通常无法直接解释的中间表示分解为可解释特征的稀疏和,从而助力更好的控制与后续分析。然而,文本生成图像模型一直缺乏类似的分析与方法。我们探究了使用SAE为少步文本生成图像扩散模型SDXL Turbo学习可解释特征的可行性。为此,我们在SDXL Turbo去噪U-net内Transformer块在1步设置下执行的更新上训练SAE。有趣的是,我们发现它们无需额外训练就能泛化到4步SDXL Turbo,甚至泛化到多步SDXL基础模型(即另一种不同的模型)。此外,我们证明其学习到的特征具备可解释性,能对生成过程产生因果影响,还能揭示各块之间的专业化分工。我们通过构建RIEBench(一种基于表示的图像编辑基准)实现这一点,该基准用于在图像生成过程中通过开启和关闭单个SAE特征来编辑图像。这让我们能够根据编辑类别追踪哪些Transformer块的特征最具影响力。我们的工作是首次针对文本生成图像扩散模型可解释性开展的SAE研究,研究结果确立了SAE作为理解和操控文本生成图像模型内部机制的一种有前景的方法。

英文摘要:

For large language models (LLMs), sparse autoencoders (SAEs) have been shown to decompose intermediate representations that often are not interpretable directly into sparse sums of interpretable features, facilitating better control and subsequent analysis. However, similar analyses and approaches have been lacking for text-to-image models. We investigate the possibility of using SAEs to learn interpretable features for SDXL Turbo, a few-step text-to-image diffusion model. To this end, we train SAEs on the updates performed by transformer blocks within SDXL Turbo's denoising U-net in its 1-step setting. Interestingly, we find that they generalize to 4-step SDXL Turbo and even to the multi-step SDXL base model (i.e., a different model) without additional training. In addition, we show that their learned features are interpretable, causally influence the generation process, and reveal specialization among the blocks. We do so by creating RIEBench, a representation-based image editing benchmark, for editing images while they are generated by turning on and off individual SAE features. This allows us to track which transformer blocks' features are the most impactful depending on the edit category. Our work is the first investigation of SAEs for interpretability in text-to-image diffusion models and our results establish SAEs as a promising approach for understanding and manipulating the internal mechanisms of text-to-image models.

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。

↑