MLLM-CL/DCL

该数据集包含五个配置（AD、Fin、Med、RS、Sci），每个配置用于多模态任务，结合文本对话、问题和答案以及图像数据。特征包括：conversations（对话列表，每个对话有from和value字段）、problem（问题文本）、answer（答案文本）、images（图像列表）和_qid（唯一标识符）。数据集分为训练集和测试集，适用于训练和评估模型，可能涉及广告、金融、医疗、遥感或科学领域的问答或对话任务。

This dataset includes five configurations (AD, Fin, Med, RS, Sci), each designed for multimodal tasks that combine text conversations, questions and answers, along with image data. Features include: conversations (a list of dialogues with from and value fields), problem (question text), answer (answer text), images (a list of images), and _qid (unique identifier). The dataset is split into training and test sets, suitable for training and evaluating models, potentially involving question-answering or dialogue tasks in domains such as advertising, finance, medical, remote sensing, or science.