mteb/Quora-PLHardNegatives
收藏Hugging Face2025-09-10 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/mteb/Quora-PLHardNegatives
下载链接
链接失效反馈官方服务：
资源简介：
Quora-PLHardNegatives是一个基于Quora平台波兰语问题的数据集，用于文本检索任务。该数据集是MTEB（Massive Text Embedding Benchmark）的一部分，包含了给定查询找到其他重复问题的任务。它使用了BM25、e5-multilingual-large和e5-mistral-instruct模型选出的每个查询的前250个文档作为硬负样本。
Quora-PLHardNegatives is a dataset based on Polish questions from the Quora platform, designed for text retrieval tasks. It is a part of MTEB (Massive Text Embedding Benchmark) and includes the task of finding duplicate questions for a given query. The dataset uses the top 250 documents per query selected by BM25, e5-multilingual-large, and e5-mistral-instruct models as hard negative samples.
提供机构：
mteb
