Misraj/mudd
收藏官方服务：
资源简介：
MUDD是一个包含4,758,338行未经处理的阿拉伯语文本的大型数据集，专为预训练大型语言模型和推动阿拉伯自然语言处理研究而设计。每条文本记录都保留了其UUID。该数据集是从SlimPajama-627B英文数据集的精选子集翻译而来，使用了Mutarjim翻译模型。
MUDD is a large-scale dataset consisting of 4,758,338 rows of unprocessed Arabic text, specifically designed for pretraining large language models and advancing Arabic natural language processing research. Each text entry retains its UUID. The dataset is translated from a carefully selected subset of the English SlimPajama-627B dataset using the Mutarjim translation model.
提供机构：
Misraj
