PhD student, University of North Carolina at Chapel Hill
3 papers at NeurIPS 2025
We introduce MJ-Bench-Video, a large-scale video preference dataset for comprehensively evaluating the reward models of text-to-video generation as well as a MoE-based video reward model
A novel benchmark using a comprehensive preference dataset to evaluate multimodal judges across multiple key perspectives
ReAgent-V enables reward-driven, multi-agent video understanding with dynamic reflection and frame selection.