SALT-NLP/MultiModalInstructionFollowing

该数据集是一个包含音频、文本和对应问题的多模态数据集。每个样本包括音频文件、音频的转录文本、相关的问题和上下文、一个或多个答案及其起始位置、表示答案是否可能的布尔值、样本的唯一标识符、一组可信答案、指令类型、指令提示、指令ID以及一组默认参数。数据集分为训练集，提供了详细的字节数和示例数。此外，还包括了数据集的下载大小和实际大小。数据集的配置文件提供了训练集的数据文件路径。

This dataset is a multimodal dataset containing audio, text, and corresponding questions. Each sample includes an audio file, a transcription of the audio, related questions and context, one or more answers with their start positions, a boolean indicating the possibility of the answers, a unique identifier for the sample, a set of plausible answers, instruction types, instruction prompts, instruction IDs, and a set of default arguments. The dataset is split into a training set, with detailed byte size and number of examples provided. In addition, the download size and actual size of the dataset are included. The dataset configuration file provides the data file paths for the training set.