EXMODD

EXMODD是一个解释性多模态开放领域对话数据集，由北京理工大学创建。该数据集包含9989条对话，旨在通过提供图像和文本的结合，增强对话系统的理解和响应能力。数据集通过设计合适的提示，利用大规模预训练语言模型生成高质量内容，并自动提供图像与对话内容之间的解释，以提高数据的可解释性和后续人工质量检查的便利性。EXMODD数据集特别适用于多模态开放领域对话任务，有助于模型生成多样化和连贯性强的响应，同时降低毒性。

EXMODD is an explanatory multimodal open-domain dialogue dataset created by Beijing Institute of Technology. It contains 9989 dialogues, aiming to enhance the understanding and response capabilities of dialogue systems by providing a combination of images and text. The dataset generates high-quality content via large-scale pre-trained language models with properly designed prompts, and automatically provides explanations between images and dialogue content, so as to improve the interpretability of the data and facilitate subsequent manual quality inspections. The EXMODD dataset is particularly suitable for multimodal open-domain dialogue tasks, helping models generate diverse and coherent responses while reducing toxicity.