lmms-lab/Omni_Bench

该数据集包含了多种类型的信息，如任务类型、音频类型等，并划分为训练集。每个样本包含问题、答案、选项以及对应的音频和图像内容，适用于各种多媒体问答或识别任务。

The dataset contains various types of information such as task type, audio type, etc., and is split into training sets. Each sample includes questions, answers, options, as well as corresponding audio and image content, suitable for various multimedia question-answering or recognition tasks.