Jiaqi Li

LongVPO: from anchored cues to self-reasoning for long-form video preference optimization

Stochastic layer-wise shuffle for improving Vision Mamba training

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