homebrewltd/instruction-text-only-full

该数据集包含多个特征，主要用于多轮对话任务。特征包括提示（prompt）、回答（answer）、多轮对话标志（multiturn）、对话内容列表（conversations，包含内容和角色）以及索引（index）。数据集分为一个训练集，包含2,501,692个样本，总大小为7,774,314,687字节。下载大小为4,088,033,617字节。数据集的配置文件中指定了默认配置，数据文件路径为data/train-*。

This dataset contains multiple features, primarily designed for multi-turn dialogue tasks. Features include prompt, answer, multiturn flag, conversations (containing content and role), and index. The dataset is divided into a training set containing 2,501,692 samples, with a total size of 7,774,314,687 bytes. The download size is 4,088,033,617 bytes. The datasets configuration file specifies the default configuration, with data file paths as data/train-*.