VADER
收藏资源简介:
VADER数据集是一个由安全专家人工评估的基准数据集,旨在评估大型语言模型在软件漏洞评估、检测、解释和修复方面的性能。数据集包含174个真实世界软件漏洞案例,每个案例都是从GitHub存储库中精心挑选并由安全专家注释的。VADER数据集的创建过程经过严格的审核,以确保每个案例都包含准确的漏洞信息、修复方案和验证测试。数据集旨在推动漏洞感知型大型语言模型的发展,为软件安全领域提供可解释和可复现的基准。
The VADER dataset is a benchmark dataset manually evaluated by cybersecurity experts, aiming to assess the performance of large language models (LLMs) in software vulnerability assessment, detection, explanation, and remediation. It contains 174 real-world software vulnerability cases, each carefully selected from GitHub repositories and annotated by cybersecurity experts. The development process of the VADER dataset has undergone rigorous review to ensure that each case includes accurate vulnerability information, remediation solutions, and validation tests. This dataset is designed to advance the development of vulnerability-aware large language models, providing an explainable and reproducible benchmark for the field of software security.




