EmoBench-M
收藏资源简介:
EmoBench-M是由深圳大学计算机科学与软件工程学院等机构创建的一个新型基准数据集,旨在评估大型多模态语言模型在情感智能方面的能力。该数据集基于心理学的情感理论,包含13个评估场景,涵盖了基础情感识别、对话情感理解和复杂社会情感分析三个维度。数据集采用视频、音频和文本等多模态数据,为评估大型多模态语言模型在真实世界交互中的情感智能提供了全面的基准。
EmoBench-M is a novel benchmark dataset created by institutions including the School of Computer Science and Software Engineering, Shenzhen University, aiming to evaluate the emotional intelligence capabilities of large multimodal language models. Based on psychological emotion theories, it encompasses 13 evaluation scenarios covering three core dimensions: basic emotion recognition, conversational emotion understanding, and complex social emotion analysis. Leveraging multi-modal data such as video, audio and text, this dataset provides a comprehensive benchmark for assessing the emotional intelligence of large multimodal language models in real-world interactions.

- 1EmoBench-M: Benchmarking Emotional Intelligence for Multimodal Large Language Models深圳大学计算机科学与软件工程学院 · 2025年



