MixEval
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/psycoy/mixeval/
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是一个评估对话能力的基准，包含了4000个通用问题和1000个难题。评估遵循评估模型建立的协议进行。总体而言，该数据集共有5000个问题，旨在评估模型的问题解决能力。
This dataset is a benchmark for evaluating conversational competence. It comprises 4,000 general questions and 1,000 challenging problems. Evaluations are conducted in accordance with the protocol established for model assessment. Overall, this dataset contains a total of 5,000 questions, which is designed to assess the problem-solving capabilities of models.
