rSDE-Bench
收藏arXiv2025-09-30 收录
下载链接：
https://yuzhu-cai.github.io/rSDE-Bench/
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是一个面向需求导向的软件开发基准，包含了复杂多样的软件需求，并具备对需求正确性进行自动评估的功能。此外，该自动评估指标（准确性）与人工评估高度一致，相关系数达到了0.9922。该数据集的任务是对软件需求进行评估，同时考察编码能力。
This dataset is a requirement-oriented software development benchmark that encompasses diverse and complex software requirements and supports automatic evaluation of requirement correctness. Moreover, the automatic evaluation metric (accuracy) is highly consistent with manual evaluation, with a correlation coefficient of 0.9922. The tasks of this dataset involve evaluating software requirements and assessing coding capabilities.
提供机构：
Yuzhu Cai
