基于逻辑闭环检验的大型视觉语言模型幻觉缓解

Logical Closed Loop: Uncovering Object Hallucinations in Large Vision-Language ModelsNew Laboratory of Pattern Recognition (NLPR), State Key Laboratory of Multimodal Artificial Intelligence Systems (MAIS), Institute of Automation, Chinese Academy of Sciences (CASIA)Qiang LiuYSSNLP2024YSSNLP2024YSSNLP2024YSSNLP2024YSSNLP2024YSSNLP2024YSSNLP2024YSSNLP2024YSSNLP20242Promising LVLMspBackgroundpLarge language models (LLMs) like GPT-4, Llama-3, and Claude, have showcased impressive abilities.pEmpowered by LLMs, large vision-language models (LVLMs) are facilitated to perform strong multimodal understanding and reasoning, e.g. MiniGPT-4, InstructBLIP, LLaVA, and QWEN-VL.InstructBLIP (NIPS 2023)LLaVA (NIPS 2023)YSSNLP2024YSSNLP2024YSSNLP2024YSSNLP2024YSSNLP2024YSSNLP2024YSSNLP2024YSSNLP2024YSSNLP20243Hallucinations in LVLMsYSSNLP2024YSSNLP2024YSSNLP2024YSSNLP2024YSSNLP2024YSSNLP2024YSSNLP2024YSSNLP2024YSSNLP20244Related WorkpRepresentative WorkpHigh-quality pre-trained and instruction tuning.pM