Awesome RL with Human Feedback / 2024 / arXiv

Aligning Large Multimodal Models with Factually Augmented RLHF

作者：Zhiqing Sun, Sheng Shen, Shengcao Cao, Haotian Liu, Chunyuan Li, Yikang Shen, Chuang Gan, Liang-Yan Gui, Yu-Xiong Wang, Yiming Yang, Kurt Keutzer, Trevor Darrell · 单位：Microsoft Research, MIT-IBM Watson AI Lab · 来源：arXiv · 发布日期：2024

事实性/幻觉偏好学习对齐目标设计数据筛选多模态反馈
新标签打开 AlphaXiv新标签打开 Code
Aligning Large Multimodal Models with Factually Augmented RLHF figure
AlphaXiv 中文概览（可滚动查看）