HA-DPO
收藏OpenXLab2026-04-18 收录
官方服务：
资源简介：
Hallucination-aware positive-negative data for HA-DPO (Hallucination-aware Direct Preference Optimization), which are used for LVLM hallucination mitigation. HA-DPO data consists positive-negative data for 3 LVLMs (MiniGPT-4, InstructBLIP, and LLaVA-1.5) in 2 formats (dense image description and question answering).
提供机构：
zzy8782180创建时间：
2024-01-30
