Using Style Ambiguity Loss to Improve Aesthetics of Diffusion Models

TMLR Paper3137 Authors

Abstract: Teaching text-to-image models to be creative involves using style ambiguity loss. In this work, we explore using the style ambiguity training objective, used to approximate creativity, on a diffusion model. We then experiment with forms of style ambiguity loss that do not require training a classifier or a labeled dataset, and find that the models trained with style ambiguity loss can generate better images than the baseline diffusion models and GANs.

Submission Length: Regular submission (no more than 12 pages of main content)

Previous TMLR Submission Url: https://openreview.net/forum?id=GqG4IvRyNl

Changes Since Last Submission:

1. User Study added to assess novelty
2. Comparisons with GANs
3. Model notation changed

Assigned Action Editor: ~Xuming_He3

Submission Number: 3137