Latent Diffusion U-Net Representations Contain Positional Embeddings and Anomalies
Fuente:
arXiv
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908309905211392 |
|---|---|
| author | Loos, Jonas Linhardt, Lorenz |
| author_facet | Loos, Jonas Linhardt, Lorenz |
| contents | Diffusion models have demonstrated remarkable capabilities in synthesizing realistic images, spurring interest in using their representations for various downstream tasks. To better understand the robustness of these representations, we analyze popular Stable Diffusion models using representational similarity and norms. Our findings reveal three phenomena: (1) the presence of a learned positional embedding in intermediate representations, (2) high-similarity corner artifacts, and (3) anomalous high-norm artifacts. These findings underscore the need to further investigate the properties of diffusion model representations before considering them for downstream tasks that require robust features. Project page: https://jonasloos.github.io/sd-representation-anomalies |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_07008 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Latent Diffusion U-Net Representations Contain Positional Embeddings and Anomalies Loos, Jonas Linhardt, Lorenz Computer Vision and Pattern Recognition Diffusion models have demonstrated remarkable capabilities in synthesizing realistic images, spurring interest in using their representations for various downstream tasks. To better understand the robustness of these representations, we analyze popular Stable Diffusion models using representational similarity and norms. Our findings reveal three phenomena: (1) the presence of a learned positional embedding in intermediate representations, (2) high-similarity corner artifacts, and (3) anomalous high-norm artifacts. These findings underscore the need to further investigate the properties of diffusion model representations before considering them for downstream tasks that require robust features. Project page: https://jonasloos.github.io/sd-representation-anomalies |
| title | Latent Diffusion U-Net Representations Contain Positional Embeddings and Anomalies |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2504.07008 |