lmms-lab/mmau

该数据集是一个包含音频文件和相关问题的测试集，每个样本都包括一个唯一标识符、音频文件、问题文本、选项文本、答案、数据集名称、任务类型、数据集划分、类别、子类别和难度等级。数据集分为test和test_mini两个部分，分别包含9000和1000个样本，用于测试目的。

The dataset is a test set containing audio files and associated questions, with each sample including a unique identifier, an audio file, a question text, choice texts, an answer, the dataset name, task type, dataset split, category, sub-category, and difficulty level. The dataset is divided into two parts, test and test_mini, containing 9000 and 1000 samples respectively, for testing purposes.