Xiangyu Zeng

VideoChat-r1.5: visual test-time scaling to reinforce multimodal reasoning by iterative perception

StreamForest: efficient online video understanding with persistent event memory

Make your training flexible: towards deployment-efficient video models

Task preference optimization: improving multimodal large language models with vision task alignment

Online video understanding: a comprehensive benchmark and memory-augmented method

TimeSuite: Improving MLLMs for Long Video Understanding via Grounded Tuning