STEAR: Layer-Aware Spatiotemporal Evidence Intervention for Hallucination Mitigation in Video Large Language Models
STEAR:面向视频大语言模型的层感知时空证据干预以缓解幻觉
机构 * Gaoling School of Artificial Intelligence, Renmin University of China(中国人民大学高瓴人工智能学院) ; King Abdullah University of Science and Technology(阿卜杜拉国王科技大学)
专题命中 知识编辑与模型理解 :large language model(title,abstract);language model(title,abstract);LLM(abstract)
AI总结 STEAR通过层感知的时空证据干预框架,缓解视频大语言模型中的空间和时间幻觉,提升解码的准确性与一致性。
Comments Preprint