leonidas123/gemma-4-pretokenized-traces

该数据集是一个预分词处理的数据集合，包含多个领域的子集：幼儿教育、功能调用、长文本Alpaca模型、小学数学、医学问答（MedQuad）、大规模科学（MegaScience）以及多语言数据（OPUS 3300）。每个子集都有训练分割，提供了输入ID和注意力掩码的特征。This dataset is a collection of pretokenized data subsets across various domains: education for young children, function calling, long-form alpaca, grade school mathematics, medical queries (MedQuad), large-scale science (MegaScience), and multilingual data (OPUS 3300). Each subset includes a training split with features such as input IDs and attention masks.