تقرير
Diffusion Lens: Interpreting Text Encoders in Text-to-Image Pipelines
| العنوان: | Diffusion Lens: Interpreting Text Encoders in Text-to-Image Pipelines |
|---|---|
| المؤلفون: | Toker, Michael, Orgad, Hadas, Ventura, Mor, Arad, Dana, Belinkov, Yonatan |
| سنة النشر: | 2024 |
| المجموعة: | Computer Science |
| مصطلحات موضوعية: | Computer Science - Computer Vision and Pattern Recognition, Computer Science - Computation and Language, I.2.7, I.4.0 |
| الوصف: | 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 requires 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. Comment: Project webpage: tokeron.github.io/DiffusionLensWeb |
| نوع الوثيقة: | Working Paper |
| URL الوصول: | http://arxiv.org/abs/2403.05846 |
| رقم الأكسشن: | edsarx.2403.05846 |
| قاعدة البيانات: | arXiv |
| الوصف غير متاح. |