Liang Li

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

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