幻城网安--首个开源的超大网络安全数据集

网络安全数据集是一个全面的网络安全相关数据集合，由新疆幻城网安科技有限公司首个开源的全网最大的网络安全数据集旨在用于训练和微调网络安全应用中的大型语言模型 (LLM)。它专注于内部网络（离线）环境，解决漏洞检测、威胁分析、代码审计和 AI 驱动的防御机制等痛点。该数据集经过精心策划，旨在支持开发本地部署的 AI 模型，例如 DeepSeek-14B 和 Qwen-Code-30B MOE，从而实现敏感数据映射、零信任架构、AI 威胁预警、智能代码安全审计、网络流量分析和 WebShell 木马检测等功能。

This comprehensive cybersecurity dataset, the world's largest open-source cybersecurity collection first launched by Xinjiang Huancheng Cyberspace Security Technology Co., Ltd., is specifically developed for training and fine-tuning Large Language Models (LLMs) in cybersecurity applications. It focuses on internal network (offline) environments, addressing key pain points such as vulnerability detection, threat analysis, code auditing, and AI-driven defense mechanisms. This carefully curated dataset supports the development of locally deployed AI models including DeepSeek-14B and Qwen-Code-30B MOE, enabling functionalities like sensitive data mapping, zero-trust architecture, AI threat early warning, intelligent code security auditing, network traffic analysis, and WebShell trojan detection.

搜集汇总

背景与挑战

背景概述

幻城网安数据集是一个开源的超大网络安全数据集，整合了多种网络安全数据源和专有样本，适用于训练本地化AI安全大语言模型。数据集包含漏洞数据库、恶意软件样本、网络流量日志等多个类别，总规模超过3.53GB。