MIA-DPO: Multi-Image Augmented Direct Preference Optimization For Large Vision-Language Models
Research paper by Ziyu Liu and collaborators. Published 2024-10-23. Topics: Large Vision-Language Models, LVLMs, direct preference optimization, DPO, Multi-Image Augmented Direct Preference Optimization.
Resource information
- Authors
- Ziyu Liu, Yuhang Zang, Xiaoyi Dong, Pan Zhang, Yuhang Cao, Haodong Duan, Conghui He, Yuanjun Xiong, Dahua Lin, Jiaqi Wang
- Research topics
- Large Vision-Language Models, LVLMs, direct preference optimization, DPO, Multi-Image Augmented Direct Preference Optimization, MIA-DPO, attention values, LLaVA-v1.5, InternLM-XC2.5
- Publication date
- 2024-10-23
- Publication
- arXiv
- Last record update
- Nov 3, 2026