Stratified Dataset of 26M Files

该数据集精心筛选自113种不同内容类型，旨在为深度学习内容类型检测器提供一个基准和训练资源。它包含了来自多种内容类型的文件，特别注重在各类环境中均衡地呈现不同类型。该数据集规模宏大，包含2600万个文件，其任务是对内容类型进行检测。

This dataset is carefully curated from 113 distinct content types, serving as both a benchmark and training resource for deep learning-based content type detectors. It includes files spanning multiple content types, with particular emphasis on the balanced representation of each type across diverse contexts. Boasting a substantial scale with 26 million files, this dataset is tailored for content type detection tasks.