xTimeCrystal/TinyCorpus
收藏Hugging Face2025-06-05 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/xTimeCrystal/TinyCorpus
下载链接
链接失效反馈官方服务：
资源简介：
TinyCorpus是一个由Lucas S维护的高质量文本数据集，大小约250B字节。它主要用于教育和实验目的，训练参数小于50M的字节级英语模型。该数据集采用UTF-8编码，使用特定的字节标记文本的开始和结束。数据来源包括55%的FineWeb Edu，35%的DCLM Edu，10%的FineMath，1%经过筛选的StarCoderData Python（教育分数≥3），以及0.1%的WritingPrompts Curated。
TinyCorpus is a high-quality text dataset curated by Lucas S, approximately 250B bytes in size. It is primarily used for educational and experimental purposes to train byte-level English models with less than 50M parameters. The dataset is encoded in UTF-8, using specific bytes to mark the beginning and end of text. The sources of the data include 55% FineWeb Edu, 35% DCLM Edu, 10% FineMath, 1% Filtered StarCoderData Python (Educational Score ≥3), and 0.1% WritingPrompts Curated.
提供机构：
xTimeCrystal
