产品简介 Product Overview

驭码 CodeRider 是极狐GitLab推出的基于人工智能和生成式内容技术（AIGC）的新一代软件生产工具，与极狐GitLab 深度融合，提供开发者 AI 辅助编程和 DevOps 流程支持，包括代码补全、生成、解释、单元测试生成、议题处理、 MR 处理加速、 AI 多轮对话以及企业知识库等功能。

CodeRider is an innovative software based on Artificial Intelligence and Generative Content Technology (AIGC) launched by JiHu GitLab. It deeply integrates with JiHu GitLab, offering developers AI coding assistant and AI DevOps assistant. CodeRider encompasses various features, including code autocompletion, generation, explanation, unit-testing generation, issue management, acceleration of Merge Request (MR) processing, AI multi-turn chat, and enterprise knowledge base capabilities.

核心场景 Core Scenarios

  1. 智能代码辅助 AI Coding Assistant

    CodeRider 在您研发过程中，根据项目代码仓库实时生成代码推荐、单元测试、代码解释、注释、优化建议等。CodeRider 旨在为您提供沉浸式代码辅助体验，借助其流畅的代码生成速度，帮助您提升编程效率。

    CodeRider generates real-time code recommendations, unit tests, code explanations, comments, and optimization suggestions based on your project’s code repository during the development process. CodeRider aims to provide you with an immersive code assistance experience, utilizing its smooth code generation speed to help enhance your programming efficiency.

  2. 智能 DevOps 辅助 AI DevOps Assistant

    CodeRider 与极狐GitLab 的 DevOps 流程深度结合，打造简单易用的智能辅助研发流程，提升您的组织合作研发效率。

    CodeRider is deeply integrated with GitLab’s DevOps workflow to create a user-friendly intelligent development process, improving your organization’s collaborative development efficiency.

  3. 智能研发对话 AI R&D Chat

    CodeRider 支持多轮对话技术问答，并结合企业私有文档资源，为您高效解决研发过程中的问题，提升企业内部文档资源的应用价值。

    CodeRider supports multi-turn technical Q&A and combines internal document resources to efficiently resolve issues encountered during the development process, enhancing the application value of your company’s internal documentation resources.

产品优势 Product Advantages

  • 灵活的模型算力选择： CodeRider 提供混合模型算力配置，企业管理员和用户可根据实际使用场景，灵活选择算力资源，实现成本与性能的最优平衡，满足不同研发场景要求。

  • Flexible Model Compute Options: CodeRider provides a hybrid model compute power configuration, allowing enterprise administrators and users to flexibly choose compute resources based on actual usage scenarios. This achieves an optimal balance between cost and performance, meeting the requirements of different development scenarios.

  • 全方位覆盖研发场景：CodeRider 与极狐GitLab 的 DevOps 深度结合，提供需求理解、代码编写、合并请求处理等全方位研发智能辅助，提高整个 DevOps 流程的效率和可靠性，以满足各行业的研发提效需求。

  • Comprehensive Coverage of Development Scenarios: CodeRider is deeply integrated with GitLab’s DevOps, offering comprehensive intelligent assistance for requirement understanding, coding, merge request processing, and more. This improves the efficiency and reliability of the entire DevOps process, meeting the development efficiency needs of various industries.

  • 强大的智能生成功能： CodeRider 及时更新并采用最先进的代码大模型，智能生成高质量的代码片段，完成项目任务，有效减少开发人员的重复编程工作，提高团队工作效率。

  • Powerful Intelligent Code Generation Capabilities. CodeRider timely updates and utilizes state-of-the-art code large models to intelligently generate high-quality code snippets, completing project tasks and effectively reducing developers’ repetitive programming work, thereby enhancing team efficiency.

  • 支持私有化部署，保障数据安全：CodeRider 支持私有化部署，满足企业对数据隐私和安全性的严格要求，并在遵循法规和合规性方面获得更高的可信度和安全性，为业务提供可信度和安全性保障。

  • Private Deployment Ensures Data Security: CodeRider supports enterprise self-management, meeting strict requirements for data privacy and security. It enhances trust and security compliance with regulations, providing reliable safeguards for business operations.

  • 国际化双语模型赋能：CodeRider 提供了国际化双语模型支持，帮助全球团队实现无障碍协作研发，针对所有产品功能实现流畅精准的中英文切换生成，满足全球研发场景需求。

  • International Bilingual Model Enablement: CodeRider offers international bilingual model support, facilitating seamless collaborative development for global teams. It provides smooth and precise Chinese-English switching across all product functions, catering to the needs of global development scenarios.

产品部署条件 Deployment Requirements

服务端CPU Server CPU Configuration

CPU 需求取决于用户数量和预期的工作负载，确切需求更多地取决于您的工作负载。您的工作负载受多重因素影响，不限于您的用户活跃程度等。

以下是针对部分用户数量群体，推荐的最低 CPU 硬件要求。

  • 16 核 是推荐的最小核数，可以支持大约 100 名用户进行较为频繁的 GitLab 文档知识库问答。若您的用户没有太多 GitLab 文档知识库问答的需求，则 CPU 的核数可减少至 8 核。

CPU requirements depend on the number of users and expected workload, with exact needs determined largely by your specific workload. Your workload is influenced by multiple factors, including but not limited to user activity levels.

This is the recommended minimum CPU hardware requirement for a subset of user quantity groups.

  • 16 cores is the recommended minimum number of cores, capable of supporting approximately 100 users for frequent GitLab documentation knowledge base queries. If users do not have a high demand for GitLab documentation knowledge base queries, the number of CPU cores can be reduced to 8.

服务端内存 Server Memory Configuration

内存需求取决于用户数量和预期的工作负载，确切需求更多地取决于您的工作负载。您的工作负载受多重因素影响，不限于您的用户活跃程度等。

以下是针对部分用户数量群体，推荐的最低内存硬件要求。

  • 64GB RAM 是必需的最小内存，可以支持大约 100 名用户进行较为频繁的 GitLab 文档知识库问答。若您的用户没有太多 GitLab 文档知识库问答的需求，则内存可减少至 32GB RAM。

Memory requirements depend on the number of users and expected workload, with exact needs determined largely by your specific workload. Your workload is influenced by multiple factors, including but not limited to user activity levels.

This is the recommended minimum memory hardware requirement for a subset of user quantity groups.

  • 64GB of RAM is the required minimum memory to support approximately 100 users for frequent GitLab documentation knowledge base queries. If users do not have a high demand for GitLab documentation knowledge base queries, memory can be reduced to 32GB of RAM.

服务端存储 Server Storage Configuration

必要的硬盘空间在很大程度上取决于您为您的用户选择的模型的大小和模型数量，通常情况下一个 7B 的大模型的大小为 4GB，一个 2B 的大模型的大小为 1.6GB，请根据实际需求来计算。一期的部署实施方案中，所有大模型的镜像将从 jihulab.com 的镜像库下载，您不必为此准备服务端存储空间。

The required disk space largely depends on the size and quantity of models chosen for your users. Typically, a large model of size 7B is around 4GB, and a large model of size 2B is approximately 1.6GB. Please calculate based on your actual needs. In the initial deployment plan, all large model images will be downloaded from the mirror repository at jihulab.com, so you do not need to prepare server-side storage space for this purpose.

GitLab软件 GitLab

CodeRider 支持 GitLab CE/EE 以及极狐GitLab 的基础版、专业版和旗舰版，建议您的 GitLab 版本不低于 12.0。如果您已经安装了 GitLab 实例，仅需将其配置到驭码 CodeRider 服务端即可（免费注册试用SaaS用户无需配置服务端）。

CodeRider supports GitLab CE/EE as well as the Basic, Professional, and Ultimate editions of JH GitLab. We recommend using GitLab version 12.0 or higher. If you already have a GitLab instance installed, you simply need to configure it with the CodeRider server. (Free trial registration for SaaS users does not require server-side configuration.)

开发者设备 Developer Device Configuration

  1. 本地模型 Local Model

    CodeRider 目前支持模型部署于本地，开发者可使用桌面计算机或笔记本电脑以运行CodeRider桌面端。

    以下是桌面个人电脑或者笔记本电脑的推荐配置：

    • Intel/AMD CPU + 英伟达 GPU 卡（显存8GB及以上）

    • Intel/AMD CPU + AMD GPU 卡（显存8GB及以上）

    • Apple Silicon M1/M2/M3（统一内存16GB及以上）

    • 操作系统：Windows 10/11、Linux（主流发布版）、MacOS

    • 磁盘空间：不低于50GB硬盘空间，固态硬盘性能更佳


    CodeRider currently supports local model deployment, allowing developers to use desktop computers or laptops to run the CodeRider desktop client.

    This is the recommended configuration for desktop or laptop computers.

    • Intel/AMD CPU + NVIDIA GPU (with 8GB VRAM or more)

    • Intel/AMD CPU + AMD GPU (with 8GB VRAM or more)

    • Apple Silicon M1/M2/M3 (Unified memory 16GB or more)

    • Operating System: Windows 10/11, Linux (major distributions), MacOS

    • Disk space: Not less than 50GB of hard disk space, with solid-state drive (SSD) performance recommended

  2. 云端模型 Cloud Model

    CodeRider 目前支持开发者使用云端模型，若您使用云端模型，则可灵活选择开发者设备。

    CodeRider now allows developers to use cloud-based models. By using these cloud models, developers can flexibly select their development devices.

重要声明 Important Statement

驭码CodeRider《AI产品服务协议》、《AI产品法律声明》和《极狐GitLab服务条款》

使用大模型进行辅助编程时，我们需要获取您的代码仓库上下文信息，但上下文信息不会被存储或用于其他任何目的，该数据完全由您所有及控制。驭码（CodeRider）及相关服务的输出是依托大模型技术提供的人工智能服务，不应视为互联网新闻信息，不能代替专业领域从业人员向您解答对应疑问。相关回答基于大模型训练时所使用的有限数据，生成的结果仅供您测试参考，并不代表任何一方观点或意见。

CodeRIder “AI Product Service Agreement”, “AI Product Legal Statement” and “JH GitLab Terms of Service”.

When using CodeRider for assisted programming, we may need access to the context information of your code repository. However, this context information will not be stored or used for any other purposes, and the data remains entirely owned and controlled by you. CodeRider and its associated services leverage AI technology powered by large models. Outputs should not be considered as internet news information and cannot substitute responses from professionals in specific fields to address corresponding queries. Responses are generated based on limited data used during the training of large models, and the results are provided for your testing and reference purposes only, without representing any specific viewpoints or opinions.