thng292/fineweb-subset-10M
收藏Hugging Face2025-08-18 更新2025-10-25 收录
下载链接:
https://hf-mirror.com/datasets/thng292/fineweb-subset-10M
下载链接
链接失效反馈官方服务:
资源简介:
该数据集来自HuggingFaceFW/fineweb-2和HuggingFaceFW/fineweb-edu,包含五种语言:越南语、英语、法语、日语和中文。每种语言都有200万个训练样本和10万个测试样本,总计训练样本1000万行,测试样本50万行。默认子集是其他子集的组合。
This dataset comes from HuggingFaceFW/fineweb-2 and HuggingFaceFW/fineweb-edu, including five languages: Vietnamese, English, French, Japanese, and Chinese. Each language has 2,000,000 training samples and 100,000 test samples, totaling 10 million rows for training and 500,000 rows for testing. The default subset is a combination of other subsets.
提供机构:
thng292



