mario-rc/aif-emotional-generation

这是一个新颖的、多样化的多轮对话数据集，旨在训练和评估聊天机器人生成情感细微反应的能力。数据集通过使用GPT-4作为生成引擎，并遵循CoE（一致性同理心）方法来创建。这个方法结构化了对话，以反映一致的同理心理解和情感色调对齐，以及开放式回应，以促进对话参与。数据集包含了针对1000个不同话题生成的多样化情感反应，每个话题都通过单次推断方法结合CoE方法进行提示。

This is a novel, diverse multi-turn dialogues dataset developed to train and evaluate a chatbot’s ability to generate emotionally nuanced responses. The dataset was created using GPT-4 as a generative engine, following the CoE methodology. This methodology structures dialogues to reflect consistent empathetic understanding and emotional tone alignment, along with open-ended responses, to promote conversational engagement.