While reinforcement learning with verifiable rewards provides reliable outcome supervision for VideoLLMs, sequence-level rewards offer limited token-level guidance. On-policy self-distillation addresses this limitation by conditioning a self-teacher on privileged information to provide dense token-level supervision. Ho...
Wei-Jie Zhu, Han Fang, Han-Yun Fu et al.· 0 citations
TTRSD separates update direction from update position, determined by group-relative advantages, from update position, determined by visual sensitivity, without requiring ground-truth labels, external verifiers, or a separate teacher, demonstrating cross-dataset generalization while preserving inherent reasoning integri...
OVIBench, the first standardized benchmark for evaluating VLMs in this setting, is introduced and it is demonstrated that OVIBench effectively distinguishes models'interruption-handling abilities, especially in following correction requests.
Nai-Ming Liu, Zhi-Heng Wu, Shu-Ning Wang et al.· 1 citation
GenEvA uses a query-conditioned evidence distribution to focus aggregation on relevant frames, forming compact cross-frame latent evidence from their frame-specific information from their frame-specific information.