Generated Evaluation Dataset
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/LZY-the-boys/MIRACLE
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了大约4600个实例，这些实例是由ChatGPT生成的，旨在评估Miracle模型在个性化和生成能力方面的表现。该数据集涵盖了三个个人属性八种不同的组合，并作为评估的基准真值。其规模约为4600个实例，任务是对个性化对话生成的评估。
This dataset contains approximately 4,600 instances generated by ChatGPT, which are specifically created to assess the Miracle model's performance in personalization and generation capabilities. It encompasses eight distinct combinations of three personal attributes, serving as the ground truth for the evaluation task. With a total of around 4,600 instances, this dataset is tailored for the evaluation of personalized dialogue generation.
提供机构：
Authors of the paper
