arXiv preprint / cs.CV

GeoReward: Mitigating Contextual Variable Overestimation in Vision-Language Models for Cross-Market Preference Prediction

作者：Shuo Liu, Huixiang Cai, Weiru Zhang, Xiaoyi Zeng · 会议/期刊：arXiv preprint · 方向：cs.CV · 发布日期：2026-08-06

2026-08-06视觉语言视觉感知强化学习
新标签打开 AlphaXiv新标签打开 Code
GeoReward: Mitigating Contextual Variable Overestimation in Vision-Language Models for Cross-Market Preference Prediction figure
AlphaXiv 中文概览（可滚动查看）