allenai/dolma3_pool
收藏Hugging Face2025-12-11 更新2025-12-20 收录
下载链接:
https://hf-mirror.com/datasets/allenai/dolma3_pool
下载链接
链接失效反馈官方服务:
资源简介:
Dolma 3 Pool是一个包含超过9万亿token的数据集,数据来源多样,包括网页内容(Common Crawl)、学术文档(olmOCR Science PDFs)、GitHub代码(StackEdu)、LaTeX论文(arXiv)、数学网页(FineMath 3+)以及百科全书内容(Wikipedia & Wikibooks)。该数据集主要用于训练Olmo 3 7B模型的第一阶段,适用于文本生成任务。数据集语言为英语,采用Open Data Commons Attribution License v1.0 (ODC-By)许可,供研究和教育使用。
The Dolma 3 pool is a dataset of over 9 trillion tokens from a diverse mix of web content (Common Crawl), academic documents (olmOCR Science PDFs), GitHub code (StackEdu), LaTeX papers (arXiv), math web pages (FineMath 3+), and encyclopedic content (Wikipedia & Wikibooks). This dataset is primarily used to train the first stage of the Olmo 3 7B model and is suitable for text-generation tasks. The dataset is in English and is licensed under the Open Data Commons Attribution License v1.0 (ODC-By) for research and educational use.
提供机构:
allenai



