@inproceedings{6416a036e54144d4b2cc1903e956e7df,
title = "Diffusion Lens: Interpreting Text Encoders in Text-to-Image Pipelines",
abstract = "Text-to-image diffusion models (T2I) use a latent representation of a text prompt to guide the image generation process. However, the process by which the encoder produces the text representation is unknown. We propose the DIFFUSION LENS, a method for analyzing the text encoder of T2I models by generating images from its intermediate representations. Using the DIFFUSION LENS, we perform an extensive analysis of two recent T2I models. Exploring compound prompts, we find that complex scenes describing multiple objects are composed progressively and more slowly compared to simple scenes; Exploring knowledge retrieval, we find that representation of uncommon concepts require further computation compared to common concepts, and that knowledge retrieval is gradual across layers. Overall, our findings provide valuable insights into the text encoder component in T2I pipelines.",
author = "Michael Toker and Hadas Orgad and Mor Ventura and Dana Arad and Yonatan Belinkov",
note = "Publisher Copyright: {\textcopyright} 2024 Association for Computational Linguistics.; 62nd Annual Meeting of the Association for Computational Linguistics, ACL 2024 ; Conference date: 11-08-2024 Through 16-08-2024",
year = "2024",
language = "الإنجليزيّة",
series = "Proceedings of the Annual Meeting of the Association for Computational Linguistics",
pages = "9713--9728",
editor = "Lun-Wei Ku and Martins, {Andre F. T.} and Vivek Srikumar",
booktitle = "Long Papers",
}