mteb/WITT2IRetrieval
收藏Hugging Face2025-10-21 更新2025-10-25 收录
下载链接:
https://hf-mirror.com/datasets/mteb/WITT2IRetrieval
下载链接
链接失效反馈官方服务:
资源简介:
WITT2IRetrieval是一个多语言图像检索数据集,可用于训练和评估嵌入模型,以便根据多语言描述检索图像。该数据集支持多种语言,包括阿拉伯语、保加利亚语、丹麦语、希腊语、英语、爱沙尼亚语、印度尼西亚语、日语、韩语、土耳其语和越南语。数据集根据MTEB基准获得,并采用CC BY-SA 4.0许可证。
WITT2IRetrieval is a multilingual image retrieval dataset that can be used to train and evaluate embedding models for the task of retrieving images based on multilingual descriptions. The dataset supports multiple languages including Arabic, Bulgarian, Danish, Greek, English, Estonian, Indonesian, Japanese, Korean, Turkish, and Vietnamese. The dataset is part of the MTEB benchmark and is licensed under CC BY-SA 4.0.
提供机构:
mteb



