← Search

Jakob Buhmann

2 accepted papers

2024

CADS: Unleashing the Diversity of Diffusion Models through Condition-Annealed Sampling

ICLR 2024spotlight

While conditional diffusion models are known to have good coverage of the data distribution, they still face limitations in output diversity, particularly when sampled with a high classifier-free guidance scale for optimal image quality or when trained on small datasets. We attribute this problem to…

Cited by 38SourcePDFScholar
2024

LiteVAE: Lightweight and Efficient Variational Autoencoders for Latent Diffusion Models

NeurIPS 2024poster

Advances in latent diffusion models (LDMs) have revolutionized high-resolution image generation, but the design space of the autoencoder that is central to these systems remains underexplored. In this paper, we introduce LiteVAE, a new autoencoder design for LDMs, which leverages the 2D discrete wav…

Cited by 10SourcePDFScholar