SolidiFI-benchmark
收藏arXiv2025-09-30 收录
下载链接:
https://github.com/Mirror-Tang/Evaluation-of-ChatGPT-s-Smart-Contract-Auditing-Capabilities-Based-on-Chain-of-Thought/tree/master
下载链接
链接失效反馈官方服务:
资源简介:
该数据集名为SolidiFI-benchmark,旨在评估智能合约中各种类型漏洞的检测成果,并通过此评估GPT-4在识别这些漏洞方面的性能。该数据集包含了多种智能合约审计报告,并以GPT-4与多种漏洞检测工具的性能对比作为评估依据。该数据集的规模涉及对多种类型漏洞的评估,其任务专注于智能合约的漏洞检测。
The dataset named SolidiFI-benchmark is designed to evaluate the detection performance of various types of vulnerabilities in smart contracts, and assess GPT-4's capability in identifying these vulnerabilities. It includes multiple smart contract audit reports, and the evaluation is based on the performance comparison between GPT-4 and a variety of vulnerability detection tools. The scope of this dataset covers evaluations of multiple types of vulnerabilities, with its core task focusing on smart contract vulnerability detection.
搜集汇总
数据集介绍

背景与挑战
背景概述
SolidiFI-benchmark是一个包含35个智能合约和732个漏洞的数据集,用于评估GPT-4在智能合约安全审计中的漏洞检测、代码解析和PoC编写能力。该数据集支持基于思维链(CoT)提示的审计过程模拟,并与其他漏洞检测工具进行对比研究。
以上内容由遇见数据集搜集并总结生成



