abhi26/cvbench_2d_curated

该数据集是一个多模态问答数据集，包含900个训练示例，每个示例包括图像、问题、多个选项和正确答案，并可能涉及不同类型的任务和提示，用于支持视觉语言模型的训练和评估。数据集结构包括索引、类型、任务、图像、问题、选项列表、答案和提示等字段。

This dataset is a multimodal question-answering dataset comprising 900 training examples, each including an image, a question, multiple choices, and the correct answer, and may involve different types of tasks and prompts for supporting the training and evaluation of vision-language models. The dataset structure includes fields such as idx, type, task, image, question, choices list, answer, and prompt.