TerenceLau/sparrow
收藏Hugging Face2025-02-07 更新2025-02-15 收录
下载链接：
https://hf-mirror.com/datasets/TerenceLau/sparrow
下载链接
链接失效反馈官方服务：
资源简介：
Sparrow项目旨在帮助初学者从零开始理解大型语言模型的基础架构。该项目不仅包括模型本身，还包括用于缩短训练过程的优化方法。数据集从维基百科获取，使用联合国官方语言中的10%-20%数据进行训练，确保语料库内容丰富且易于获取。
The sparrow project aims to help beginners understand the base architecture of a large language model from scratch. Not only the model, but also the optimization methods that are widely used to shorten the training process. The dataset is collected from Wikipedia, using 10%-20% of the data in the six official languages of the United Nations for training, ensuring that the corpus is rich in content and easily accessible.
提供机构：
TerenceLau
