Wikipedia Dump (Filtered)
收藏arXiv2025-09-30 收录
下载链接：
https://storage.googleapis.com/lateral-datadumps/wikipedia_utf8_filtered_20pageviews.csv.gz
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是对2013年维基百科数据转储的筛选版本，包含了至少有20次页面浏览的页面，用于训练词嵌入。原始文本已经按照附录中概述的方法进行了预处理。该数据集的规模为463,000个文档，包含约4.98亿个单词，其任务是进行词嵌入训练。
This dataset is a filtered subset of the 2013 Wikipedia data dump, comprising pages that accumulated at least 20 page views, and is designed for word embedding training. The raw text has been preprocessed in accordance with the methodologies outlined in the appendix. This dataset encompasses 463,000 documents with approximately 498 million words in total, with its target task being word embedding training.
提供机构：
Wikipedia
