EleutherAI/djinn-problems-v0.3

该数据集包含了漏洞相关的详细信息，包括漏洞描述、函数名称、测试用例、真实情况、利用代码、关键字、评估难度等。每个记录还包括了不同模型对真实情况和利用情况的判断结果，以及关于漏洞欺骗性、利用公平性、问题质量等指标。数据集分为训练集和评估集，可用于机器学习模型的训练和评估。

The dataset contains detailed information related to vulnerabilities, including vulnerability descriptions, function names, test cases, ground truth, exploit code, keywords, evaluated difficulty, and more. Each record also includes the results of different models judgments on the ground truth and exploit situations, as well as metrics on vulnerability cheatiness, exploit fairness, and problem quality. The dataset is split into training and evaluation sets, which can be used for training and evaluating machine learning models.