bcml/bclm-char-200k
收藏Hugging Face2025-10-16 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/bcml/bclm-char-200k
下载链接
链接失效反馈官方服务：
资源简介：
BCLM-1字符数据集（2048上下文，200k样本）包含200k个样本，每个样本的最大字符长度为2046个，由各种教育文本和文章组成，用于训练小型语言模型。数据集的分布是50%来自fineweb-edu，50%来自Cosmopedia。超过2046个字符的样本将被截断。
BCLM-1 Character Dataset (2048 ctx, 200k samples) consists of 200k samples, with a maximum character length of 2046, made up of various educational texts and articles for training a small language model. The dataset is distributed equally between 50% fineweb-edu and 50% Cosmopedia. Samples longer than 2046 characters are truncated.
提供机构：
bcml
