Devils in Middle Layers of Large Vision-Language Models: Interpreting, Detecting and Mitigating Object Hallucinations via Attention Lens
Zhangqi Jiang, Junkai Chen, Beier Zhu, Tingjin Luo, Yankun Shen, Xu Yang
Who it is from
- National University of Defense Technologyaffiliation declared by the authors, according to OpenAlex
Other papers from the same organizations
- Beyond the 80/20 Rule: High-Entropy Minority Tokens Drive Effective Reinforcement Learning for LLM ReasoningAlibaba Group · 722 citations
- A Survey of Knowledge Graph Reasoning on Graph Types: Static, Dynamic, and Multi-ModalNational University of Defense Technology · 555 citations
- Large Language Models are Few-Shot Summarizers: Multi-Intent Comment Generation via In-Context LearningNational University of Defense Technology · 296 citations
- AbsGS: Recovering Fine Details in 3D Gaussian SplattingNational University of Defense Technology · 251 citations
- Bilateral Reference for High-Resolution Dichotomous Image SegmentationNorthwestern Polytechnical University · 235 citations
- Calibrated One-Class Classification for Unsupervised Time Series Anomaly DetectionNational University of Defense Technology · 235 citations