Modality-Parallel Persona Dataset

该数据集包含40个多样化的人格特征，这些特征在年龄、性别、职业和地域上各不相同，并通过四种方式呈现：仅图像、仅文本、图像与小文本结合，以及排版图像。每个人格都由一组独特的属性定义：年龄、性别、职业和地域，并以四种不同的方式呈现，以评估多模态大型语言模型的表现力。该项任务旨在评估多模态大型语言模型中人格特征的表达性。

This dataset contains 40 diverse personality traits, with samples varying across age, gender, occupation and geographic region, and presented in four formats: image-only, text-only, image combined with short text, and typographic image. Each personality trait is defined by a unique set of attributes including age, gender, occupation and geographic region, and is presented in four distinct modalities to evaluate the expressive capabilities of multimodal large language models. This task aims to assess the expressiveness of personality traits in multimodal large language models.