Shitao Xiao

UltraGist compresses long texts while keeping essential information intact.

VISTA improves how we find information by integrating text and visuals.

MLVU benchmark aims to improve machine understanding of long videos.

OmniGen simplifies image creation tasks into a single model for all users.

A novel model enhances text embeddings through in-context learning strategies.

MegaPairs connects images and text for better search results.