m-a-p/OmniInstruct_v1

OmniBench是一个新颖的基准测试，旨在严格评估模型在同时处理视觉、听觉和文本输入时的识别、解读和推理能力。数据集包含文本、音频、图像等多种类型的数据，分为训练集和验证集。它包括问题、选项、答案、音频文件路径、图像文件路径等信息，旨在评估全语言模型（OLMs）的性能。

OmniBench is a novel benchmark designed to rigorously evaluate models ability to recognize, interpret, and reason across visual, acoustic, and textual inputs simultaneously. The dataset includes various types of data such as text, audio, and images, split into training and validation sets. It contains questions, options, answers, paths to audio files, paths to image files, etc., aiming to assess the performance of omni-language models (OLMs).