activity
20242026
collaborators

6 papers

cs.CV2026

Improved Baselines with Representation Autoencoders

Jaskirat Singh, Boyang Zheng, Zongze Wu +3

Representation Autoencoders (RAE) replace traditional VAE with pretrained vision encoders. In this paper, we systematically investigate several design choices and find three insigh…

cs.CV2026

Improved Mean Flows: On the Challenges of Fastforward Generative Models

Zhengyang Geng, Yiyang Lu, Zongze Wu +3

MeanFlow (MF) has recently been established as a framework for one-step generative modeling. However, its ``fastforward'' nature introduces key challenges in both the training obje…

cs.CV2026

End-to-End Training for Unified Tokenization and Latent Denoising

Shivam Duggal, Xingjian Bai, Zongze Wu +5

Latent diffusion models (LDMs) enable high-fidelity synthesis by operating in learned latent spaces. However, training state-of-the-art LDMs requires complex staging: a tokenizer m…

cs.CV2026

Causality in Video Diffusers is Separable from Denoising

Xingjian Bai, Guande He, Zhengqi Li +3

Causality -- referring to temporal, uni-directional cause-effect relationships between components -- underlies many complex generative processes, including videos, language, and ro…

cs.CV2025

What matters for Representation Alignment: Global Information or Spatial Structure?

Jaskirat Singh, Xingjian Leng, Zongze Wu +4

Representation alignment (REPA) guides generative training by distilling representations from a strong, pretrained vision encoder to intermediate diffusion features. We investigate…

cs.CV2025

SliderSpace: Decomposing the Visual Capabilities of Diffusion Models

Rohit Gandikota, Zongze Wu, Richard Zhang +3

We present SliderSpace, a framework for automatically decomposing the visual capabilities of diffusion models into controllable and human-understandable directions. Unlike existing…