EQ-Bench
收藏资源简介:
EQ-Bench是由Samuel J. Paech开发的情感智能基准,旨在评估大型语言模型对复杂情感和社会互动的理解能力。该数据集包含60个英语问题,通过让模型预测对话中角色情感状态的强度来测试其情感理解能力。EQ-Bench的创建过程涉及使用GPT-4生成情感丰富的对话场景,并由作者精心选择参考答案。该数据集主要应用于评估模型的情感智能,特别是情感理解分支,以解决模型在自然语言对话中理解和处理情感复杂性的问题。
EQ-Bench is an emotional intelligence benchmark developed by Samuel J. Paech, designed to evaluate the capability of large language models (LLMs) to understand complex emotional and social interactions. This dataset contains 60 English questions that test models' emotional comprehension skills by requiring them to predict the intensity of characters' emotional states in conversations. The development of EQ-Bench involved generating emotion-rich dialogue scenarios using GPT-4, with reference answers carefully selected by the authors. This dataset is primarily used to evaluate models' emotional intelligence, specifically the emotional understanding branch, to address the challenge of models comprehending and processing emotional complexity in natural language conversations.




