mteb/NeuCLIR2023RetrievalHardNegatives
收藏Hugging Face2025-09-10 更新2025-10-25 收录
下载链接:
https://hf-mirror.com/datasets/mteb/NeuCLIR2023RetrievalHardNegatives
下载链接
链接失效反馈官方服务:
资源简介:
这是一个用于文本检索任务的数据集,包含Farsi、Russian和Chinese三种语言的文本数据。它包含了由BM25、e5-multilingual-large和e5-mistral-instruct模型生成的每个查询的250个顶级文档作为硬负样本,旨在测试和评估嵌入模型在检索相关文档方面的性能。
This is a dataset for text retrieval tasks, containing text data in Farsi, Russian, and Chinese. It includes 250 top documents per query generated by BM25, e5-multilingual-large, and e5-mistral-instruct as hard negatives, aiming to test and evaluate the performance of embedding models in retrieving relevant documents.
提供机构:
mteb



