Haonan Zhang

Visuomotor control can now generalize to unseen environments and instructions by grounding world models in a vision-language latent space, outperforming standard vision-language approaches by a large margin.

VLLMs can be made much faster without sacrificing accuracy by intelligently merging redundant tokens across space and time using optimal transport.

Achieve state-of-the-art low-field MRI enhancement by explicitly modeling and correcting the intensity distribution shift between low-field and high-field domains using a differentiable Sinkhorn optimal transport module within a diffusion framework.