CC-Stories
收藏OpenDataLab2026-05-17 更新2024-05-09 收录
下载链接：
https://opendatalab.org.cn/OpenDataLab/CC-Stories
下载链接
链接失效反馈官方服务：
资源简介：
This is a reproduction of the CC-stories dataset as it has been removed from its original source. To create this reproduction we process the English common crawl and only keep the top 0.1% of documents measured by their ngram overlap with a source document. The source document is created by joining the queries from PDP-60 and WSC273. Note, as the original dataset does not mention removing duplicate queries, neither do we.
提供机构：
OpenDataLab
创建时间：
2024-04-30
搜集汇总
数据集介绍
背景与挑战
背景概述
CC-Stories是一个复现的数据集，专为文本预训练语言模型设计，基于英文Common Crawl构建，通过筛选与特定查询组合文档ngram重叠度最高的0.1%文档而成。
以上内容由遇见数据集搜集并总结生成
