qihoo360/DOCCI-CN
收藏官方服务:
资源简介:
FG-CLIP 2是一个双语的细粒度视觉-语言对齐模型,适用于英语和中文。它在29个数据集和8个不同的任务中表现优于最近的强基线模型,实现了英语和中文的最佳报告性能。
FG-CLIP 2 is a bilingual fine-grained vision-language alignment model suitable for both English and Chinese. It outperforms recent strong baselines across 29 datasets and 8 diverse tasks, achieving the best reported performance in both languages.
提供机构:
qihoo360


