jhu-clsp/ettin-decay-data
收藏Hugging Face2025-07-18 更新2025-08-09 收录
下载链接:
https://hf-mirror.com/datasets/jhu-clsp/ettin-decay-data
下载链接
链接失效反馈官方服务:
资源简介:
Ettin衰减阶段数据集是用于训练Ettin编码器和解码器模型的数据集,包含高质量的长期内容和教育材料。数据集由多种数据源组成,包括最高质量的网络爬取数据、优质代码库、书籍、数学内容、高质量问答内容、学术预印本、百科全书文章和教科书。数据集在训练特征上包括积极的学习率衰减、长上下文、较低的掩码比例和注重质量而非数量的数据源。
The Ettin Decay Phase Data is a dataset used for training Ettin encoder and decoder models, containing high-quality long-form content and educational materials. The dataset is composed of various sources, including the highest-quality web crawl data, premium code repositories, books, mathematical content, high-quality Q&A content, academic preprints, encyclopedia articles, and textbooks. The dataset features aggressive learning rate decay, long context, lower masking ratio, and a focus on quality over quantity in its training characteristics.
提供机构:
jhu-clsp



