← Search

Zefang Wang

2 accepted papers

2026

OBS-Diff: Accurate Pruning For Diffusion Models in One-Shot

ICLR 2026poster

Large-scale text-to-image diffusion models, while powerful, suffer from prohibitive computational cost. Existing one-shot network pruning methods can hardly be directly applied to them due to the iterative denoising nature of diffusion models. To bridge the gap, this paper presents \textit{OBS-Diff}…

Cited by 0SourcecodeScholar
2026

Prism-MoE: Efficient Dense-to-MoE Conversion for Visual Autoregressive Generation

ICML 2026poster

Scaling up visual autoregressive models improves generation quality but incurs substantial inference costs. Mixture-of-Experts (MoE) architectures mitigate this issue through sparse activation and have proven effective in large language models. However, training MoE models from scratch remains prohi…

Cited by 0SourceScholar