[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-software-mansion--react-native-executorch":3,"tool-software-mansion--react-native-executorch":61},[4,18,26,36,44,53],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":17},4358,"openclaw","openclaw\u002Fopenclaw","OpenClaw 是一款专为个人打造的本地化 AI 助手，旨在让你在自己的设备上拥有完全可控的智能伙伴。它打破了传统 AI 助手局限于特定网页或应用的束缚，能够直接接入你日常使用的各类通讯渠道，包括微信、WhatsApp、Telegram、Discord、iMessage 等数十种平台。无论你在哪个聊天软件中发送消息，OpenClaw 都能即时响应，甚至支持在 macOS、iOS 和 Android 设备上进行语音交互，并提供实时的画布渲染功能供你操控。\n\n这款工具主要解决了用户对数据隐私、响应速度以及“始终在线”体验的需求。通过将 AI 部署在本地，用户无需依赖云端服务即可享受快速、私密的智能辅助，真正实现了“你的数据，你做主”。其独特的技术亮点在于强大的网关架构，将控制平面与核心助手分离，确保跨平台通信的流畅性与扩展性。\n\nOpenClaw 非常适合希望构建个性化工作流的技术爱好者、开发者，以及注重隐私保护且不愿被单一生态绑定的普通用户。只要具备基础的终端操作能力（支持 macOS、Linux 及 Windows WSL2），即可通过简单的命令行引导完成部署。如果你渴望拥有一个懂你",349277,3,"2026-04-06T06:32:30",[13,14,15,16],"Agent","开发框架","图像","数据工具","ready",{"id":19,"name":20,"github_repo":21,"description_zh":22,"stars":23,"difficulty_score":10,"last_commit_at":24,"category_tags":25,"status":17},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,"2026-04-05T11:01:52",[14,15,13],{"id":27,"name":28,"github_repo":29,"description_zh":30,"stars":31,"difficulty_score":32,"last_commit_at":33,"category_tags":34,"status":17},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",140436,2,"2026-04-05T23:32:43",[14,13,35],"语言模型",{"id":37,"name":38,"github_repo":39,"description_zh":40,"stars":41,"difficulty_score":32,"last_commit_at":42,"category_tags":43,"status":17},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",107662,"2026-04-03T11:11:01",[14,15,13],{"id":45,"name":46,"github_repo":47,"description_zh":48,"stars":49,"difficulty_score":10,"last_commit_at":50,"category_tags":51,"status":17},4292,"Deep-Live-Cam","hacksider\u002FDeep-Live-Cam","Deep-Live-Cam 是一款专注于实时换脸与视频生成的开源工具，用户仅需一张静态照片，即可通过“一键操作”实现摄像头画面的即时变脸或制作深度伪造视频。它有效解决了传统换脸技术流程繁琐、对硬件配置要求极高以及难以实时预览的痛点，让高质量的数字内容创作变得触手可及。\n\n这款工具不仅适合开发者和技术研究人员探索算法边界，更因其极简的操作逻辑（仅需三步：选脸、选摄像头、启动），广泛适用于普通用户、内容创作者、设计师及直播主播。无论是为了动画角色定制、服装展示模特替换，还是制作趣味短视频和直播互动，Deep-Live-Cam 都能提供流畅的支持。\n\n其核心技术亮点在于强大的实时处理能力，支持口型遮罩（Mouth Mask）以保留使用者原始的嘴部动作，确保表情自然精准；同时具备“人脸映射”功能，可同时对画面中的多个主体应用不同面孔。此外，项目内置了严格的内容安全过滤机制，自动拦截涉及裸露、暴力等不当素材，并倡导用户在获得授权及明确标注的前提下合规使用，体现了技术发展与伦理责任的平衡。",88924,"2026-04-06T03:28:53",[14,15,13,52],"视频",{"id":54,"name":55,"github_repo":56,"description_zh":57,"stars":58,"difficulty_score":32,"last_commit_at":59,"category_tags":60,"status":17},3704,"NextChat","ChatGPTNextWeb\u002FNextChat","NextChat 是一款轻量且极速的 AI 助手，旨在为用户提供流畅、跨平台的大模型交互体验。它完美解决了用户在多设备间切换时难以保持对话连续性，以及面对众多 AI 模型不知如何统一管理的痛点。无论是日常办公、学习辅助还是创意激发，NextChat 都能让用户随时随地通过网页、iOS、Android、Windows、MacOS 或 Linux 端无缝接入智能服务。\n\n这款工具非常适合普通用户、学生、职场人士以及需要私有化部署的企业团队使用。对于开发者而言，它也提供了便捷的自托管方案，支持一键部署到 Vercel 或 Zeabur 等平台。\n\nNextChat 的核心亮点在于其广泛的模型兼容性，原生支持 Claude、DeepSeek、GPT-4 及 Gemini Pro 等主流大模型，让用户在一个界面即可自由切换不同 AI 能力。此外，它还率先支持 MCP（Model Context Protocol）协议，增强了上下文处理能力。针对企业用户，NextChat 提供专业版解决方案，具备品牌定制、细粒度权限控制、内部知识库整合及安全审计等功能，满足公司对数据隐私和个性化管理的高标准要求。",87618,"2026-04-05T07:20:52",[14,35],{"id":62,"github_repo":63,"name":64,"description_en":65,"description_zh":66,"ai_summary_zh":67,"readme_en":68,"readme_zh":69,"quickstart_zh":70,"use_case_zh":71,"hero_image_url":72,"owner_login":73,"owner_name":74,"owner_avatar_url":75,"owner_bio":76,"owner_company":77,"owner_location":77,"owner_email":77,"owner_twitter":78,"owner_website":79,"owner_url":80,"languages":81,"stars":120,"forks":121,"last_commit_at":122,"license":123,"difficulty_score":124,"env_os":125,"env_gpu":126,"env_ram":127,"env_deps":128,"category_tags":137,"github_topics":139,"view_count":32,"oss_zip_url":77,"oss_zip_packed_at":77,"status":17,"created_at":153,"updated_at":154,"faqs":155,"releases":184},4254,"software-mansion\u002Freact-native-executorch","react-native-executorch","Declarative way to run AI models in React Native on device, powered by ExecuTorch.","React Native ExecuTorch 是一款专为移动端开发打造的开源工具，它让开发者能够以声明式的方式，在 React Native 应用中直接运行本地 AI 模型。该工具基于 Meta 推出的 ExecuTorch 框架，有效解决了以往在移动设备上部署大语言模型（LLM）或计算机视觉模型时流程复杂、性能优化困难以及过度依赖云端算力的痛点。\n\n通过桥接 React Native 与原生平台能力，React Native ExecuTorch 实现了高效的端侧推理，不仅降低了网络延迟和数据隐私风险，还提供了开箱即用的支持，涵盖多种主流 AI 模型。其独特的技术亮点在于将复杂的底层模型执行逻辑封装为简洁的接口，使前端开发者无需深入掌握底层原生代码或复杂的机器学习运维知识，即可轻松集成先进的 AI 功能。\n\n这款工具非常适合使用 React Native 进行移动应用开发的工程师、希望快速验证端侧 AI 原型的算法研究人员，以及致力于构建离线可用、高隐私保护智能应用的创业团队。借助 React Native ExecuTorch，开发者可以更专注于业务逻辑与创新体验，轻松为手机等移","React Native ExecuTorch 是一款专为移动端开发打造的开源工具，它让开发者能够以声明式的方式，在 React Native 应用中直接运行本地 AI 模型。该工具基于 Meta 推出的 ExecuTorch 框架，有效解决了以往在移动设备上部署大语言模型（LLM）或计算机视觉模型时流程复杂、性能优化困难以及过度依赖云端算力的痛点。\n\n通过桥接 React Native 与原生平台能力，React Native ExecuTorch 实现了高效的端侧推理，不仅降低了网络延迟和数据隐私风险，还提供了开箱即用的支持，涵盖多种主流 AI 模型。其独特的技术亮点在于将复杂的底层模型执行逻辑封装为简洁的接口，使前端开发者无需深入掌握底层原生代码或复杂的机器学习运维知识，即可轻松集成先进的 AI 功能。\n\n这款工具非常适合使用 React Native 进行移动应用开发的工程师、希望快速验证端侧 AI 原型的算法研究人员，以及致力于构建离线可用、高隐私保护智能应用的创业团队。借助 React Native ExecuTorch，开发者可以更专注于业务逻辑与创新体验，轻松为手机等移动设备赋予强大的本地智能处理能力，推动 AI 应用在边缘端的普及。","\u003Cdiv align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fblob\u002Fmain\u002Fdocs\u002Fstatic\u002Fimg\u002Flogo-hero.svg\" alt=\"RNE Logo\" width=\"25%\">\n\u003C\u002Fdiv>\n\n\n\u003Cdiv align=\"center\">\n  \u003Ch1 align=\"center\" style=\"display:inline-block\">React Native ExecuTorch\n  \u003C\u002Fh1>\n\u003C\u002Fdiv>\n\n[![Ad](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsoftware-mansion_react-native-executorch_readme_79816051f8e4.png)](https:\u002F\u002Fswm-delivery.com\u002Fwww\u002Fdelivery\u002Fck.php?zoneid=zone-gh-react-native-executorch-1&n=1)\n[![Ad](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsoftware-mansion_react-native-executorch_readme_79816051f8e4.png)](https:\u002F\u002Fswm-delivery.com\u002Fwww\u002Fdelivery\u002Fck.php?zoneid=zone-gh-react-native-executorch-2&n=1)\n[![Ad](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsoftware-mansion_react-native-executorch_readme_79816051f8e4.png)](https:\u002F\u002Fswm-delivery.com\u002Fwww\u002Fdelivery\u002Fck.php?zoneid=zone-gh-react-native-executorch-3&n=1)\n\n\u003Cdiv align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fgraphs\u002Fcontributors\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fcontributors\u002Fsoftware-mansion\u002Freact-native-executorch?style=for-the-badge&color=00008B\" alt=\"GitHub - Contributors\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fstargazers\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002Fsoftware-mansion\u002Freact-native-executorch?style=for-the-badge&color=00008B\" alt=\"GitHub - Stars\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fdiscord.gg\u002FZGqqY55qkP\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDiscord-Join%20Us-00008B?logo=discord&logoColor=white&style=for-the-badge\" alt=\"Join our Discord community\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002F\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDocumentation-00008B?logo=googledocs&logoColor=white&style=for-the-badge\" alt=\"Documentation\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fswmansion.com\u002Fcontact\">\n    \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FHire%20Us-00008B?logo=react&logoColor=white&color=darkgreen&style=for-the-badge\" alt=\"Hire Us\">\n  \u003C\u002Fa>\n\u003C\u002Fdiv>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fblob\u002Fmain\u002FREADME.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FEN-00008B?logo=&logoColor=white&color=00008B&style=for-the-badge\" alt=\"README\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fblob\u002Fmain\u002Freadmes\u002FREADME_es.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FES-00008B?logo=&logoColor=white&color=00008B&style=for-the-badge\" alt=\"README ES\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fblob\u002Fmain\u002Freadmes\u002FREADME_fr.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FFR-00008B?logo=&logoColor=white&color=00008B&style=for-the-badge\" alt=\"README FR\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fblob\u002Fmain\u002Freadmes\u002FREADME_cn.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FCN-00008B?logo=&logoColor=white&color=00008B&style=for-the-badge\" alt=\"README CN\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fblob\u002Fmain\u002Freadmes\u002FREADME_pt.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPT-00008B?logo=&logoColor=white&color=00008B&style=for-the-badge\" alt=\"README PT\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fblob\u002Fmain\u002Freadmes\u002FREADME_in.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FIN-00008B?logo=&logoColor=white&color=00008B&style=for-the-badge\" alt=\"README IN\">\u003C\u002Fa>\n\u003C\u002Fp>\n\n**React Native ExecuTorch** provides a declarative way to run AI models on-device using React Native, powered by **ExecuTorch** :rocket:. It offers out-of-the-box support for a wide range of LLMs, computer vision models, and more. Visit our [HuggingFace](https:\u002F\u002Fhuggingface.co\u002Fsoftware-mansion) page to explore these models.\n\n[**ExecuTorch**](https:\u002F\u002Fexecutorch.ai), developed by Meta, is a novel framework allowing AI model execution on devices like mobile phones or microcontrollers.\n\nReact Native ExecuTorch bridges the gap between React Native and native platform capabilities, enabling developers to efficiently run local AI models on mobile devices. This can be achieved without the need for extensive expertise in native programming or machine learning.\n\n[![npm version](https:\u002F\u002Fimg.shields.io\u002Fnpm\u002Fv\u002Freact-native-executorch?color=00008B)](https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002Freact-native-executorch)\n[![npm nightly](https:\u002F\u002Fimg.shields.io\u002Fnpm\u002Fv\u002Freact-native-executorch\u002Fexecutorch-nightly?label=nightly&color=00008B)](https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002Freact-native-executorch)\n[![CI](https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Factions\u002Fworkflows\u002Fci.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Factions\u002Fworkflows\u002Fci.yml)\n\n\u003Cdetails>\n\u003Csummary>\u003Cstrong>Table of Contents\u003C\u002Fstrong>\u003C\u002Fsummary>\n\n- [Supported Versions](#supported-versions)\n- [Real-world Example](#real-world-example)\n- [Quickstart - Running Llama](#quickstart---running-llama)\n  - [:one: Installation](#one-installation)\n  - [:two: Setup \\& Initialization](#two-setup--initialization)\n  - [:three: Run the Model!](#three-run-the-model)\n- [Demo Apps](#demo-apps)\n- [Ready-made Models](#ready-made-models)\n- [Documentation](#documentation)\n- [License](#license)\n- [What's Next?](#whats-next)\n- [React Native ExecuTorch is created by Software Mansion](#react-native-executorch-is-created-by-software-mansion)\n\n\u003C\u002Fdetails>\n\n## Supported Versions\n\nThe minimal supported version are:\n* iOS 17.0\n* Android 13\n* React Native - see [compatibility table](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fnext\u002Fother\u002Fcompatibility)\n\n> [!IMPORTANT]\n> React Native ExecuTorch supports only the [New React Native architecture](https:\u002F\u002Freactnative.dev\u002Farchitecture\u002Flanding-page).\n\n## Real-world Example\n\nReact Native ExecuTorch is powering [Private Mind](https:\u002F\u002Fprivatemind.swmansion.com\u002F), a privacy-first mobile AI app available on [App Store](https:\u002F\u002Fapps.apple.com\u002Fgb\u002Fapp\u002Fprivate-mind\u002Fid6746713439) and [Google Play](https:\u002F\u002Fplay.google.com\u002Fstore\u002Fapps\u002Fdetails?id=com.swmansion.privatemind).\n\n\u003Cimg width=\"2720\" height=\"1085\" alt=\"Private Mind promo\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsoftware-mansion_react-native-executorch_readme_09215c4c5327.png\" \u002F>\n\n## Quickstart - Running Llama\n\n**Get started with AI-powered text generation in 3 easy steps!**\n\n### :one: Installation\n\n```bash\n# Install the package\nyarn add react-native-executorch\n\n# If you use expo, please add these packages for resource fetching:\nyarn add react-native-executorch-expo-resource-fetcher\nyarn add expo-file-system expo-asset\n\n#if you use bare React Native project use these packages:\nyarn add react-native-executorch-bare-resource-fetcher\nyarn add @dr.pogodin\u002Freact-native-fs @kesha-antonov\u002Freact-native-background-downloader\n\n# Depending on the platform, choose either iOS or Android\nyarn \u003C ios | android >\n```\n\n### :two: Setup & Initialization\n\nAdd this to your component file:\n\n```tsx\nimport {\n  useLLM,\n  LLAMA3_2_1B,\n  Message,\n  initExecutorch,\n} from 'react-native-executorch';\nimport { ExpoResourceFetcher } from 'react-native-executorch-expo-resource-fetcher';\n\ninitExecutorch({\n  resourceFetcher: ExpoResourceFetcher,\n});\n\nfunction MyComponent() {\n  \u002F\u002F Initialize the model 🚀\n  const llm = useLLM({ model: LLAMA3_2_1B });\n  \u002F\u002F ... rest of your component\n}\n```\n\n### :three: Run the Model!\n\n```tsx\nconst handleGenerate = async () => {\n  const chat: Message[] = [\n    { role: 'system', content: 'You are a helpful assistant' },\n    { role: 'user', content: 'What is the meaning of life?' }\n  ];\n\n  \u002F\u002F Chat completion\n  await llm.generate(chat);\n  console.log('Llama says:', llm.response);\n};\n```\n\n## Demo Apps\n\nWe currently host a few example [apps](https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Ftree\u002Fmain\u002Fapps) demonstrating use cases of our library:\n\n- `llm` - Chat application showcasing use of LLMs\n- `speech` - Speech to Text & Text to Speech task implementations\n- `computer-vision` - Computer vision related tasks\n- `text-embeddings` - Computing text representations for semantic search\n- `bare_rn` - LLM chat example for bare React Native (without Expo)\n\nIf you would like to run a demo app, navigate to its project directory. Then install dependencies and run app with:\n\n```bash\nyarn && yarn \u003C ios | android >\n```\n\n> [!WARNING]\n> Running LLMs requires a significant amount of RAM. If you are encountering unexpected app crashes, try to increase the amount of RAM allocated to the emulator.\n\n## Ready-made Models\n\nOur library has a number of ready-to-use AI models; a complete list is available in the documentation. If you're interested in running your own AI model, you need to first export it to the `.pte` format. Instructions on how to do this are available in the [Python API](https:\u002F\u002Fdocs.pytorch.org\u002Fexecutorch\u002Fstable\u002Fusing-executorch-export.html) and [optimum-executorch README](https:\u002F\u002Fgithub.com\u002Fhuggingface\u002Foptimum-executorch?tab=readme-ov-file#option-2-export-and-load-separately).\n\n## Documentation\n\nCheck out how our library can help you build your React Native AI features by visiting our docs:\nhttps:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\n\n## License\n\nThis library is licensed under [The MIT License](.\u002FLICENSE).\n\n## What's Next?\n\nTo learn about our upcoming plans and developments, please visit our [milestones](https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fmilestones).\n\n## React Native ExecuTorch is created by Software Mansion\n\nSince 2012, [Software Mansion](https:\u002F\u002Fswmansion.com) is a software agency with experience in building web and mobile apps. We are Core React Native Contributors and experts in dealing with all kinds of React Native issues. We can help you build your next dream product – [Hire us](https:\u002F\u002Fswmansion.com\u002Fcontact?utm_source=react-native-executorch&utm_medium=readme).\n\n[![swm](https:\u002F\u002Flogo.swmansion.com\u002Flogo?color=white&variant=desktop&width=150&tag=react-native-executorch-github 'Software Mansion')](https:\u002F\u002Fswmansion.com)\n","\u003Cdiv align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fblob\u002Fmain\u002Fdocs\u002Fstatic\u002Fimg\u002Flogo-hero.svg\" alt=\"RNE Logo\" width=\"25%\">\n\u003C\u002Fdiv>\n\n\n\u003Cdiv align=\"center\">\n  \u003Ch1 align=\"center\" style=\"display:inline-block\">React Native ExecuTorch\n  \u003C\u002Fh1>\n\u003C\u002Fdiv>\n\n[![广告](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsoftware-mansion_react-native-executorch_readme_79816051f8e4.png)](https:\u002F\u002Fswm-delivery.com\u002Fwww\u002Fdelivery\u002Fck.php?zoneid=zone-gh-react-native-executorch-1&n=1)\n[![广告](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsoftware-mansion_react-native-executorch_readme_79816051f8e4.png)](https:\u002F\u002Fswm-delivery.com\u002Fwww\u002Fdelivery\u002Fck.php?zoneid=zone-gh-react-native-executorch-2&n=1)\n[![广告](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsoftware-mansion_react-native-executorch_readme_79816051f8e4.png)](https:\u002F\u002Fswm-delivery.com\u002Fwww\u002Fdelivery\u002Fck.php?zoneid=zone-gh-react-native-executorch-3&n=1)\n\n\u003Cdiv align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fgraphs\u002Fcontributors\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fcontributors\u002Fsoftware-mansion\u002Freact-native-executorch?style=for-the-badge&color=00008B\" alt=\"GitHub - Contributors\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fstargazers\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002Fsoftware-mansion\u002Freact-native-executorch?style=for-the-badge&color=00008B\" alt=\"GitHub - Stars\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fdiscord.gg\u002FZGqqY55qkP\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDiscord-Join%20Us-00008B?logo=discord&logoColor=white&style=for-the-badge\" alt=\"加入我们的 Discord 社区\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002F\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDocumentation-00008B?logo=googledocs&logoColor=white&style=for-the-badge\" alt=\"文档\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fswmansion.com\u002Fcontact\">\n    \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FHire%20Us-00008B?logo=react&logoColor=white&color=darkgreen&style=for-the-badge\" alt=\"聘请我们\">\n  \u003C\u002Fa>\n\u003C\u002Fdiv>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fblob\u002Fmain\u002FREADME.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FEN-00008B?logo=&logoColor=white&color=00008B&style=for-the-badge\" alt=\"README 英文版\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fblob\u002Fmain\u002Freadmes\u002FREADME_es.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FES-00008B?logo=&logoColor=white&color=00008B&style=for-the-badge\" alt=\"README 西班牙语版\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fblob\u002Fmain\u002Freadmes\u002FREADME_fr.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FFR-00008B?logo=&logoColor=white&color=00008B&style=for-the-badge\" alt=\"README 法语版\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fblob\u002Fmain\u002Freadmes\u002FREADME_cn.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FCN-00008B?logo=&logoColor=white&color=00008B&style=for-the-badge\" alt=\"README 中文版\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fblob\u002Fmain\u002Freadmes\u002FREADME_pt.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPT-00008B?logo=&logoColor=white&color=00008B&style=for-the-badge\" alt=\"README 葡萄牙语版\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fblob\u002Fmain\u002Freadmes\u002FREADME_in.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FIN-00008B?logo=&logoColor=white&color=00008B&style=for-the-badge\" alt=\"README 印地语版\">\u003C\u002Fa>\n\u003C\u002Fp>\n\n**React Native ExecuTorch** 提供了一种使用 React Native 在设备端运行 AI 模型的声明式方法，由 **ExecuTorch** 驱动 :rocket:。它开箱即用地支持广泛的 LLM、计算机视觉模型等。请访问我们的 [HuggingFace](https:\u002F\u002Fhuggingface.co\u002Fsoftware-mansion) 页面以探索这些模型。\n\n由 Meta 开发的 **ExecuTorch** 是一种创新框架，允许在手机或微控制器等设备上执行 AI 模型。\n\nReact Native ExecuTorch 弥合了 React Native 与原生平台能力之间的差距，使开发者能够高效地在移动设备上运行本地 AI 模型。这一切无需深厚的原生编程或机器学习专业知识。\n\n[![npm 版本](https:\u002F\u002Fimg.shields.io\u002Fnpm\u002Fv\u002Freact-native-executorch?color=00008B)](https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002Freact-native-executorch)\n[![npm 夜间版本](https:\u002F\u002Fimg.shields.io\u002Fnpm\u002Fv\u002Freact-native-executorch\u002Fexecutorch-nightly?label=nightly&color=00008B)](https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002Freact-native-executorch)\n[![CI](https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Factions\u002Fworkflows\u002Fci.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Factions\u002Fworkflows\u002Fci.yml)\n\n\u003Cdetails>\n\u003Csummary>\u003Cstrong>目录\u003C\u002Fstrong>\u003C\u002Fsummary>\n\n- [支持的版本](#supported-versions)\n- [实际案例](#real-world-example)\n- [快速入门 - 运行 Llama](#quickstart---running-llama)\n  - [:one: 安装](#one-installation)\n  - [:two: 设置与初始化](#two-setup--initialization)\n  - [:three: 运行模型！](#three-run-the-model)\n- [演示应用](#demo-apps)\n- [现成模型](#ready-made-models)\n- [文档](#documentation)\n- [许可证](#license)\n- [下一步是什么？](#whats-next)\n- [React Native ExecuTorch 由 Software Mansion 创建](#react-native-executorch-is-created-by-software-mansion)\n\n\u003C\u002Fdetails>\n\n## 支持的版本\n\n最低支持版本如下：\n* iOS 17.0\n* Android 13\n* React Native - 请参阅 [兼容性表](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fnext\u002Fother\u002Fcompatibility)\n\n> [!重要提示]\n> React Native ExecuTorch 仅支持 [新的 React Native 架构](https:\u002F\u002Freactnative.dev\u002Farchitecture\u002Flanding-page)。\n\n## 实际案例\n\nReact Native ExecuTorch 正在为 [Private Mind](https:\u002F\u002Fprivatemind.swmansion.com\u002F) 提供支持，这是一款注重隐私的移动 AI 应用程序，现已在 [App Store](https:\u002F\u002Fapps.apple.com\u002Fgb\u002Fapp\u002Fprivate-mind\u002Fid6746713439) 和 [Google Play](https:\u002F\u002Fplay.google.com\u002Fstore\u002Fapps\u002Fdetails?id=com.swmansion.privatemind) 上架。\n\n\u003Cimg width=\"2720\" height=\"1085\" alt=\"Private Mind 宣传图\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsoftware-mansion_react-native-executorch_readme_09215c4c5327.png\" \u002F>\n\n## 快速入门 - 运行 Llama\n\n**只需 3 个简单步骤，即可开始使用 AI 驱动的文本生成！**\n\n### :one: 安装\n\n```bash\n# 安装包\nyarn add react-native-executorch\n\n# 如果您使用 Expo，请添加以下用于资源获取的包：\nyarn add react-native-executorch-expo-resource-fetcher\nyarn add expo-file-system expo-asset\n\n# 如果您使用原生 React Native 项目，请使用以下包：\nyarn add react-native-executorch-bare-resource-fetcher\nyarn add @dr.pogodin\u002Freact-native-fs @kesha-antonov\u002Freact-native-background-downloader\n\n# 根据平台选择 iOS 或 Android\nyarn \u003C ios | android >\n```\n\n### :two: 设置与初始化\n\n将以下内容添加到你的组件文件中：\n\n```tsx\nimport {\n  useLLM,\n  LLAMA3_2_1B,\n  Message,\n  initExecutorch,\n} from 'react-native-executorch';\nimport { ExpoResourceFetcher } from 'react-native-executorch-expo-resource-fetcher';\n\ninitExecutorch({\n  resourceFetcher: ExpoResourceFetcher,\n});\n\nfunction MyComponent() {\n  \u002F\u002F 初始化模型 🚀\n  const llm = useLLM({ model: LLAMA3_2_1B });\n  \u002F\u002F ... 组件的其余部分\n}\n```\n\n### :three: 运行模型！\n\n```tsx\nconst handleGenerate = async () => {\n  const chat: Message[] = [\n    { role: 'system', content: '你是一个乐于助人的助手' },\n    { role: 'user', content: '生命的意义是什么？' }\n  ];\n\n  \u002F\u002F 聊天完成\n  await llm.generate(chat);\n  console.log('Llama 说：', llm.response);\n};\n```\n\n## 演示应用\n\n我们目前托管了一些示例[应用](https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Ftree\u002Fmain\u002Fapps)，展示了我们库的各种使用场景：\n\n- `llm` - 展示 LLM 使用的聊天应用\n- `speech` - 语音转文本和文本转语音任务的实现\n- `computer-vision` - 计算机视觉相关任务\n- `text-embeddings` - 用于语义搜索的文本表示计算\n- `bare_rn` - 适用于原生 React Native（不使用 Expo）的 LLM 聊天示例\n\n如果你想运行某个演示应用，请进入其项目目录。然后安装依赖并运行应用：\n\n```bash\nyarn && yarn \u003C ios | android >\n```\n\n> [!警告]\n> 运行 LLM 需要大量的内存。如果你遇到意外的应用崩溃，请尝试增加模拟器分配的内存。\n\n## 现成模型\n\n我们的库包含许多开箱即用的 AI 模型；完整列表可在文档中找到。如果你有兴趣运行自己的 AI 模型，首先需要将其导出为 `.pte` 格式。有关如何操作的说明，请参阅 [Python API](https:\u002F\u002Fdocs.pytorch.org\u002Fexecutorch\u002Fstable\u002Fusing-executorch-export.html) 和 [optimum-executorch README](https:\u002F\u002Fgithub.com\u002Fhuggingface\u002Foptimum-executorch?tab=readme-ov-file#option-2-export-and-load-separately)。\n\n## 文档\n\n请访问我们的文档，了解我们的库如何帮助你构建 React Native AI 功能：\nhttps:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\n\n## 许可证\n\n本库采用 [MIT 许可证](.\u002FLICENSE) 许可。\n\n## 下一步？\n\n如需了解我们的未来计划和发展，请访问我们的[里程碑](https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fmilestones)。\n\n## React Native ExecuTorch 由 Software Mansion 打造\n\n自 2012 年以来，[Software Mansion](https:\u002F\u002Fswmansion.com) 是一家拥有丰富经验的软件公司，专注于构建 Web 和移动应用。我们是 React Native 的核心贡献者，也是处理各种 React Native 问题的专家。我们可以帮助你打造下一个梦想产品——[联系我们](https:\u002F\u002Fswmansion.com\u002Fcontact?utm_source=react-native-executorch&utm_medium=readme)。\n\n[![swm](https:\u002F\u002Flogo.swmansion.com\u002Flogo?color=white&variant=desktop&width=150&tag=react-native-executorch-github 'Software Mansion')](https:\u002F\u002Fswmansion.com)","# React Native ExecuTorch 快速上手指南\n\nReact Native ExecuTorch 是一个基于 Meta **ExecuTorch** 框架的声明式库，允许你在 React Native 应用中直接在设备端运行 AI 模型（如 LLM、计算机视觉模型等），无需深厚的原生开发或机器学习背景。\n\n## 环境准备\n\n在开始之前，请确保你的开发环境满足以下最低要求：\n\n*   **操作系统**:\n    *   iOS: 17.0 及以上\n    *   Android: 13 (API Level 33) 及以上\n*   **React Native**: 必须使用 **新架构 (New Architecture)**。\n    *   具体版本兼容性请参考官方文档。\n*   **依赖管理**: Yarn 或 npm。\n\n> **注意**：本项目不支持旧版 React Native 架构。\n\n## 安装步骤\n\n### 1. 安装核心包\n\n在项目根目录下运行以下命令安装主包：\n\n```bash\nyarn add react-native-executorch\n```\n\n### 2. 安装资源获取插件\n\n根据你是否使用 Expo，选择安装对应的资源获取插件：\n\n**如果你使用 Expo:**\n\n```bash\nyarn add react-native-executorch-expo-resource-fetcher\nyarn add expo-file-system expo-asset\n```\n\n**如果你使用 bare React Native (非 Expo) 项目:**\n\n```bash\nyarn add react-native-executorch-bare-resource-fetcher\nyarn add @dr.pogodin\u002Freact-native-fs @kesha-antonov\u002Freact-native-background-downloader\n```\n\n### 3. 安装原生依赖并构建\n\n完成 JS 包安装后，执行以下命令安装 iOS\u002FAndroid 原生依赖：\n\n```bash\n# 对于 iOS\ncd ios && pod install && cd ..\nyarn ios\n\n# 对于 Android\nyarn android\n```\n\n## 基本使用\n\n以下是最简单的使用示例，展示如何在组件中初始化并运行 Llama 3.2 模型进行文本生成。\n\n### 1. 初始化与模型加载\n\n在你的组件文件中引入必要的钩子和配置：\n\n```tsx\nimport {\n  useLLM,\n  LLAMA3_2_1B,\n  Message,\n  initExecutorch,\n} from 'react-native-executorch';\n\u002F\u002F 如果使用 Expo，引入 ExpoResourceFetcher\nimport { ExpoResourceFetcher } from 'react-native-executorch-expo-resource-fetcher';\n\u002F\u002F 如果是 Bare RN 项目，请改为引入 BareResourceFetcher\n\n\u002F\u002F 全局初始化 Executorch\ninitExecutorch({\n  resourceFetcher: ExpoResourceFetcher, \u002F\u002F 或 BareResourceFetcher\n});\n\nfunction MyComponent() {\n  \u002F\u002F 初始化模型 🚀\n  \u002F\u002F 首次加载时会自动下载模型文件\n  const llm = useLLM({ model: LLAMA3_2_1B });\n  \n  \u002F\u002F ... 其余组件逻辑\n}\n```\n\n### 2. 运行模型生成文本\n\n调用 `generate` 方法发送消息并获取回复：\n\n```tsx\nconst handleGenerate = async () => {\n  const chat: Message[] = [\n    { role: 'system', content: 'You are a helpful assistant' },\n    { role: 'user', content: 'What is the meaning of life?' }\n  ];\n\n  try {\n    \u002F\u002F 执行对话补全\n    await llm.generate(chat);\n    \n    \u002F\u002F 获取生成的回复\n    console.log('Llama says:', llm.response);\n  } catch (error) {\n    console.error('Generation failed:', error);\n  }\n};\n```\n\n> **性能提示**: 运行大型语言模型 (LLM) 需要大量内存。如果在模拟器中遇到应用崩溃，请尝试增加分配给模拟器的 RAM 大小，或在真机上测试。","一家初创团队正在开发一款面向户外爱好者的 React Native 徒步助手 App，核心功能是通过手机摄像头实时识别沿途的植物种类，并要求在无网络信号的山区也能流畅运行。\n\n### 没有 react-native-executorch 时\n- **依赖云端导致不可用**：必须将图片上传至服务器进行推理，一旦进入无信号的深山，核心识别功能直接瘫痪。\n- **原生开发门槛极高**：若想在本地运行模型，团队需分别招募 iOS (Swift\u002FCoreML) 和 Android (Kotlin\u002FNNAPI) 专家编写底层桥接代码，人力成本翻倍。\n- **维护与同步困难**：两套原生代码逻辑难以保持一致，每次更新 AI 模型都需要重新编译发布两个平台的原生包，迭代周期长达数周。\n- **性能优化复杂**：缺乏统一的声明式接口，开发者需手动管理内存和线程，极易导致 App 在低端手机上崩溃或卡顿。\n\n### 使用 react-native-executorch 后\n- **真正的离线可用**：利用 ExecuTorch 引擎直接在设备端运行量化后的视觉模型，无需联网即可毫秒级返回植物识别结果。\n- **跨平台一套代码**：通过声明式 API，前端团队仅用 JavaScript\u002FTypeScript 即可调用原生 AI 能力，无需编写任何 Objective-C 或 Java 桥接代码。\n- **高效迭代部署**：模型更新只需替换资源文件并热更新 JS 包，显著缩短从训练到上线的验证周期，实现天级迭代。\n- **开箱即用的性能**：内置针对移动端的优化策略，自动处理线程调度与内存管理，确保在各类中低端机型上均能稳定运行。\n\nreact-native-executorch 让 React Native 开发者能以纯前端的工作流，低成本构建出高性能、完全离线的原生级 AI 应用。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsoftware-mansion_react-native-executorch_09215c4c.png","software-mansion","Software Mansion","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fsoftware-mansion_ec79379b.png","We are Software Mansion - we build open-source developer tools and provide expert software consulting.",null,"swmansion","https:\u002F\u002Fswmansion.com","https:\u002F\u002Fgithub.com\u002Fsoftware-mansion",[82,86,90,94,98,102,106,109,113,116],{"name":83,"color":84,"percentage":85},"C++","#f34b7d",68.4,{"name":87,"color":88,"percentage":89},"TypeScript","#3178c6",27.3,{"name":91,"color":92,"percentage":93},"C","#555555",2.7,{"name":95,"color":96,"percentage":97},"CMake","#DA3434",0.6,{"name":99,"color":100,"percentage":101},"Shell","#89e051",0.4,{"name":103,"color":104,"percentage":105},"JavaScript","#f1e05a",0.2,{"name":107,"color":108,"percentage":105},"Kotlin","#A97BFF",{"name":110,"color":111,"percentage":112},"Ruby","#701516",0.1,{"name":114,"color":115,"percentage":112},"Objective-C++","#6866fb",{"name":117,"color":118,"percentage":119},"Swift","#F05138",0,1384,67,"2026-04-05T21:15:57","NOASSERTION",4,"macOS, Windows, Linux","未说明 (模型在移动设备端运行，依赖设备本地 NPU\u002FGPU\u002FCPU)","开发环境未说明；运行 LLM 需要大量 RAM，若模拟器崩溃需增加分配内存",{"notes":129,"python":130,"dependencies":131},"1. 必须使用 React Native 新架构 (New Architecture)。\n2. 最低支持 iOS 17.0 和 Android 13。\n3. 运行大型语言模型 (LLM) 对内存要求较高，建议在真机测试或在模拟器中增加内存分配以防崩溃。\n4. 若需运行自定义模型，需先将其导出为 .pte 格式。","未说明 (主要作为 React Native 库使用，模型导出需参考 PyTorch\u002FExecuTorch Python API)",[64,132,133,134,135,136],"react-native (需启用新架构 New Architecture)","expo-file-system (仅 Expo 项目)","expo-asset (仅 Expo 项目)","@dr.pogodin\u002Freact-native-fs (仅 Bare RN 项目)","@kesha-antonov\u002Freact-native-background-downloader (仅 Bare RN 项目)",[35,14,138,15],"音频",[140,141,142,143,144,145,146,147,148,149,150,151,152],"computer-vision","executorch","llm-inference","object-detection","ocr","on-device-ai","react-native-ai","speech-to-text","text-to-speech","image-embeddings","segmentation","text-embeddings","vlm","2026-03-27T02:49:30.150509","2026-04-06T15:00:22.695828",[156,161,166,171,176,180],{"id":157,"question_zh":158,"answer_zh":159,"source_url":160},19379,"为什么使用 yarn add 安装 react-native-executorch 失败，而 npm install 可以成功？","该库目前必须依赖 Expo Modules 才能工作，因为内部使用了 expo-file-system 来加载模型（无论是 bundled 还是下载的）。如果不使用 Expo Modules 架构，仅使用纯 React Native CLI 项目可能会导致兼容性问题。建议确保项目中集成了 Expo Modules，或者参考文档确认是否支持非 Expo 环境。","https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fissues\u002F121",{"id":162,"question_zh":163,"answer_zh":164,"source_url":165},19380,"遇到 'UndefinedErrorModuleNotLoaded' 或 'package doesn't seem to be linked' 错误怎么办？","这通常意味着原生模块未正确链接。请确保执行了以下步骤：1. 运行 'pod install' (iOS)；2. 在安装包后重新构建整个应用；3. 确认没有在使用 Expo Go（需要使用开发客户端 Dev Client 或 bare workflow）。此外，文档和语音转文字算法实现近期已更新，如果问题依旧，请检查是否使用了最新的文档配置。","https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fissues\u002F462",{"id":167,"question_zh":168,"answer_zh":169,"source_url":170},19381,"Android 构建失败，提示 Jitpack 依赖无法解析 (Could not resolve com.github.software-mansion:react-native-executorch) 如何解决？","旧版本（如 0.1.3 及之前的快照版本）的 Jitpack 依赖问题官方已不再修复。解决方案是将 'react-native-executorch' 升级到最新版本（例如 0.5 或更高），新版本已解决构建问题并移除了对不稳定 Jitpack 快照的依赖。注意：升级到 0.5+ 可能会引入 SQLite 依赖，需确保项目中没有其他冲突的 SQLite 版本。","https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fissues\u002F400",{"id":172,"question_zh":173,"answer_zh":174,"source_url":175},19382,"在 iOS 新架构 (Fabric) 下运行时报错 'Cannot read property NativeModule of undefined' 如何处理？","此错误通常与环境配置特定问题有关，而非库本身的普遍 Bug。维护者在干净的项目中无法复现该问题。建议步骤：1. 检查是否严格遵循了安装文档中的所有步骤；2. 尝试运行官方提供的示例应用 (apps\u002Fllm) 以验证环境是否正常；3. 确认 React Native 版本与库版本的兼容性；4. 如果是特定设备或特定 RN 小版本导致，尝试清理缓存并重新执行 pod install。","https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fissues\u002F358",{"id":177,"question_zh":178,"answer_zh":179,"source_url":170},19383,"升级库版本后 iOS 构建出现 SQLite 符号冲突 (clash in symbols) 怎么办？","从 0.5 版本开始，库引入了 SQLite 依赖。如果你的项目中其他依赖也捆绑了特定版本的 SQLite 静态库，会导致符号冲突。解决方法是统一项目中所有依赖所使用的 SQLite 版本，或者等待库维护者提供解决符号冲突的补丁。在干净的项目中测试通常能正常工作，问题主要出现在复杂依赖树的项目中。",{"id":181,"question_zh":182,"answer_zh":183,"source_url":160},19384,"该库是否支持不使用 Expo 的纯 React Native CLI 项目？","不完全支持。虽然不需要使用 Expo 的其他工具（如 Expo Go），但必须使用 Expo Modules 架构，因为库依赖 'expo-file-system' 来处理模型文件的加载。如果是纯 CLI 项目且未集成 Expo Modules，将无法正常使用该库。建议在项目中安装并配置 expo-modules-core 相关依赖。",[185,190,195,200,205,210,215,220,225,230,235,240,245,250,255,260,265,270,275,280],{"id":186,"version":187,"summary_zh":188,"released_at":189},117405,"v0.8.1","## 变更内容\n新增对 LFM2.5 350M 的支持：为 XNNPACK 提供量化后的 8da4w 模型和基座 fp16 模型。\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fcompare\u002Fv0.8.0...v0.8.1","2026-04-02T08:56:07",{"id":191,"version":192,"summary_zh":193,"released_at":194},117406,"v0.9.0-libs-test","用于验证按需原生库下载流程的测试版本。非正式发布版本。","2026-03-31T20:05:44",{"id":196,"version":197,"summary_zh":198,"released_at":199},117407,"v0.8.0","## React Native ExecuTorch v0.8.0 :eye: :atom_symbol:\n\n 这是我们迄今为止最大、最令人兴奋的一次发布！如果你正在寻找使用 AI 模型处理摄像头输入、运行多模态大模型（VLM），或者需要 Bare RN 支持，那么这次更新正是你所需要的 :point_down:\n\n ### **有哪些新功能？**\n\n **计算机视觉**\n - **VisionCamera 集成:** 所有计算机视觉 Hook 现在都提供一个 `runOnFrame` worklet，用于通过 VisionCamera v5 实时处理摄像头帧 [文档](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fhooks\u002Fcomputer-vision\u002Fvisioncamera-integration)\n - **实例分割:** 新增 `useInstanceSegmentation` Hook，用于生成像素级对象掩码，支持 YOLO（nano\u002Fsmall\u002Fmedium\u002Flarge\u002Fextra-large）和 RF-DETR 模型 [文档](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fhooks\u002Fcomputer-vision\u002FuseInstanceSegmentation)\n - **目标检测:** 新增对 YOLO 和 RF-DETR 模型的支持 [文档](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fhooks\u002Fcomputer-vision\u002FuseObjectDetection)\n - **语义分割:** 新增对 DeepLab V3、LRASPP、FCN 以及 Selfie Segmentation 模型的支持 [文档](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fhooks\u002Fcomputer-vision\u002FuseSemanticSegmentation)\n - **量化 CV 模型:** 新增 CLIP、风格迁移、EfficientNetV2 和 SSDLite 的量化版本\n - **自定义模型支持目标检测及实例\u002F语义分割:** 这两个模块现在都是通用的——只要你的模型符合预期的输入输出规范，就可以通过 `fromCustomModel()` 插入自定义模型（详情请参阅相关文档）\n\n **多模态大模型 (VLM)**\n - **视觉语言模型 (VLM):** `useLLM` 现在支持多模态模型——只需传入 `capabilities: ['vision']`，即可在发送消息的同时传递图像 [文档](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fhooks\u002Fnatural-language-processing\u002FuseLLM#vision-language-models-vlm)\n - 新增对 **Liquid Foundation Models** 视觉语言模型 (`LFM2_VL_1_6B_QUANTIZED`) 的支持。\n\n **大语言模型 (LLM)**\n - **LFM-2.5 支持:** 新增对 Liquid Foundation Model 2.5 Instruct 的支持 [文档](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fhooks\u002Fnatural-language-processing\u002FuseLLM)\n - **LLM 上下文管理:** 在 `ChatConfig` 中新增了 `ContextStrategy`——可配置朴素策略、基于消息数量的策略，或滑动窗口上下文截断策略。目前默认采用的是滑动窗口策略 [文档](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fhooks\u002Fnatural-language-processing\u002FuseLLM)\n\n **音频**\n - **Whisper 现在速度提升至原来的 3 倍**\n - **TTS + LLM 集成:** 文本转语音流式合成现在支持增量扩展的文本输入，从而实现从正在运行的 LLM 直接输出实时语音 [文档](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fhooks\u002Fnatural-language-processing\u002FuseTextToSpeech)\n - **Kokoro TTS 预计算音素:** 新增 `forwardFromPhonemes` \u002F `streamFromPhonemes` API，允许绕过内置的 G2P 转换流程，并","2026-03-25T15:30:29",{"id":201,"version":202,"summary_zh":203,"released_at":204},117408,"v0.7.2","## 变更内容\n* 由 @msluszniak 在 https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fpull\u002F889 中发布了 v0.7.2 版本\n\n\n**完整变更日志**: https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fcompare\u002Fv0.7.1...v0.7.2","2026-02-27T19:15:44",{"id":206,"version":207,"summary_zh":208,"released_at":209},117409,"v0.7.1","## 变更内容\n* 二进制文件由 @benITo47 在 https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fpull\u002F803 中更新\n\n\n**完整变更日志**: https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fcompare\u002Fv0.7.0...v0.7.1","2026-02-12T11:09:29",{"id":211,"version":212,"summary_zh":213,"released_at":214},117410,"v0.7.0","## 宣布 React Native ExecuTorch v0.7.0 🌟 \n\n如果你厌倦了机器人般的语音合成器，那么这款产品正适合你！\n\n### **新功能有哪些？**\n- **文本转语音：** 支持使用 **Kokoro** 生成语音 [文档](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fhooks\u002Fnatural-language-processing\u002FuseTextToSpeech)\n- **包体积缩减：** 我们更换了用于分词的库，使包体积减少了 64%。\n- **OCR 文件大小优化：** 之前，OCR 流程需要四个不同的模型文件。现在我们只需一个文件即可完成，总文件大小因此缩小了三倍（313MB → 101MB）[文档](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fhooks\u002Fcomputer-vision\u002FuseOCR)\n- **错误处理改进：** 现在我们的库抛出的错误会附带 `code` 属性，便于进行精细化的错误处理 [文档](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Futilities\u002Ferror-handling)\n- **大语言模型：** 新增统计 API（[getPromptTokenCount](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fapi-reference\u002Finterfaces\u002FLLMType#getprompttokencount)、[getTotalTokenCount](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fapi-reference\u002Finterfaces\u002FLLMType#gettotaltokencount)）\n- **大语言模型：** `generate()` API 现在返回字符串，而非 `void` [文档](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fapi-reference\u002Finterfaces\u002FLLMType#generate)\n- 文档新增了 [API 参考](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fapi-reference) 部分。\n- 通用 Bug 修复。\n\n### **重大变更 ⚠️**\n- 移除了 LLM API 中的 `responseCallback()`\n- OCR 现在只接受一个 `modelSource` 参数，不再需要三个。\n- 移除了 `VERTICAL_OCR` 常量。今后，无论是普通 OCR 还是竖向 OCR，都应统一使用 `OCR_*` 常量。\n- 在语音转文本功能中，不再支持以 `number[]` 格式传递波形数据，请改用 `Float32Array`。\n- `TokenizerModule` 现在会在 `idToToken` 或 `tokenToIdFails` 时抛出 `RnExecutorchError` 错误（例如，当传入的标记不在词汇表范围内时）。\n- 所有 Hook 中的 `error` 属性现在类型为 [RnExecutorchError](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fapi-reference\u002Fclasses\u002FRnExecutorchError)，而非 `string`。","2026-02-05T13:38:31",{"id":216,"version":217,"summary_zh":218,"released_at":219},117411,"v0.6.0","### 宣布 React Native ExecuTorch v0.6.0 🫶🏻\n\n我们非常高兴地发布 v0.6.0！这个版本带来了显著的性能提升，并引入了强大的生成式功能，包括 Stable Diffusion、VAD 以及改进的 Whisper 支持。\n\n### 新增内容？\n\n- **重大性能优化：** 更多详情请参阅 [基准测试](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fbenchmarks\u002Finference-time)\n- **文本到图像：** 现已支持 Stable Diffusion！[文档](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fhooks\u002Fcomputer-vision\u002FuseTextToImage)\n- **语音活动检测：** 通过基于 FSMN 的 VAD 实现检测语音片段。[文档](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fhooks\u002Fnatural-language-processing\u002FuseVAD)\n- **量化 Whisper：** 我们现在支持量化后的 Whisper 模型，在几乎不损失精度的情况下，使模型文件大小缩小至原来的 **四分之一**。[文档](https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fblob\u002F25342e5a4eacfbe929084ffb84dd5d2fb8238b68\u002Fpackages\u002Freact-native-executorch\u002Fsrc\u002Fconstants\u002FmodelUrls.ts#L344)\n- **LLM 控制：** 现在可以通过可配置的 `temperature` 和 `topp` 参数来调整生成结果。[文档](https:\u002F\u002Fdocs.swmansion.com\u002Freact-native-executorch\u002Fdocs\u002Fhooks\u002Fnatural-language-processing\u002FuseLLM#configuring-the-model)\n- **ExecuTorch 运行时：** 已升级至 **v1.0.0**，以匹配上游版本的稳定性和功能。\n- **Android 支持：** 增加了对 16KB 内存页大小的支持。\n- 一般性错误修复\n\n### 重大变更 ⚠️\n\n**需升级 Expo SDK**\nReact Native ExecuTorch 现在依赖于最新的 `expo-file-system` API 变更。您必须将项目升级至 **Expo SDK 54** 才能使用此版本。","2025-12-11T15:19:11",{"id":221,"version":222,"summary_zh":223,"released_at":224},117412,"v0.5.15","## 变更内容\n修复了 Android 上的一个 bug：线程池中的线程数可能被设置为 0，导致库实际上被冻结。（该问题仅影响模拟器）\n\n\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fcompare\u002Fv0.5.14...v0.5.15","2025-10-29T14:25:54",{"id":226,"version":227,"summary_zh":228,"released_at":229},117413,"v0.4.10","## 变更内容\n* 修复了一个影响大模型响应质量的 bug，该 bug 有时会导致模型持续生成内容，直到达到最大 token 数限制。\n\n\n**完整变更日志**: https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fcompare\u002Fv0.4.9...v0.4.10","2025-10-27T11:07:41",{"id":231,"version":232,"summary_zh":233,"released_at":234},117414,"v0.5.13","## 变更内容\n* 由 @msluszniak 在 https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fpull\u002F654 中移除了 iOS 构建工作流中的 CocoaPods 缓存步骤。\n* 修复了一个影响 LLM 响应质量的 bug，该 bug 有时会导致模型持续生成内容，直到达到最大 token 数量为止。\n\n\n**完整变更日志**: https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fcompare\u002Fv0.5.12...v0.5.13","2025-10-27T11:09:48",{"id":236,"version":237,"summary_zh":238,"released_at":239},117415,"v0.5.12","Fixed bug where special characters resulted in malformed utf-8 encodings when using both usSpeechToText and SpeechToTextModule API.","2025-10-17T14:19:21",{"id":241,"version":242,"summary_zh":243,"released_at":244},117416,"v0.4.9","Changed jitpack target.","2025-10-15T12:43:13",{"id":246,"version":247,"summary_zh":248,"released_at":249},117417,"v0.5.11","## What's Changed\r\n* chore: 0.5.11 by @mkopcins in https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fpull\u002F649\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fcompare\u002Fv0.5.10...v0.5.11","2025-10-15T11:36:58",{"id":251,"version":252,"summary_zh":253,"released_at":254},117418,"v0.5.10","## What's Changed\r\n* chore: readme update by @p3t3rzb in https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fpull\u002F641\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fcompare\u002Fv0.5.9...v0.5.10","2025-10-09T14:39:40",{"id":256,"version":257,"summary_zh":258,"released_at":259},117419,"v0.5.9","## What's Changed\r\n* fix: useExecutorchModule not working by @jakmro in https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fpull\u002F637\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fcompare\u002Fv0.5.8...v0.5.9","2025-10-08T10:12:21",{"id":261,"version":262,"summary_zh":263,"released_at":264},117420,"v0.5.8","## What's Changed\r\n* Added missing getGeneratedTokenCount for LLMModule.ts\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fcompare\u002Fv0.5.7...v0.5.8","2025-10-02T14:27:33",{"id":266,"version":267,"summary_zh":268,"released_at":269},117421,"v0.5.7","## What's Changed\r\n* Added token batching during llm inference to reduce rerenders and improve performance\r\n* Improved threadpool configuration on iOS significantly increasing inference speed and reducing power consumption\r\n* Fix SQLite bug crashing app builds when SQLite is included from other libraries\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fcompare\u002Fv0.5.6...v0.5.7","2025-10-01T15:16:31",{"id":271,"version":272,"summary_zh":273,"released_at":274},117422,"v0.5.6","## What's Changed\r\n* fix: s2t demo app crashes by @jakmro in https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fpull\u002F611\r\n* chore: bump version to v0.5.6 by @jakmro in https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fpull\u002F618\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fcompare\u002Fv0.5.5...v0.5.6","2025-09-23T08:43:37",{"id":276,"version":277,"summary_zh":278,"released_at":279},117423,"v0.5.5","## What's Changed\r\n* bump 0.5.4 by @mkopcins in https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fpull\u002F593\r\n* V0.5.5 by @jakmro in https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fpull\u002F602\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fcompare\u002Fv0.5.4...v0.5.5","2025-09-15T15:09:36",{"id":281,"version":282,"summary_zh":283,"released_at":284},117424,"v0.5.4","## What's Changed\r\n* Improvements to Speech to text pipeline (migrated typescipt code to cpp)\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fsoftware-mansion\u002Freact-native-executorch\u002Fcompare\u002Fv0.5.3...v0.5.4","2025-09-10T13:34:07"]