[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-explosion--spaCy":3,"tool-explosion--spaCy":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",152630,2,"2026-04-12T23:33:54",[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。",108322,"2026-04-10T11:39:34",[14,15,13],{"id":45,"name":46,"github_repo":47,"description_zh":48,"stars":49,"difficulty_score":32,"last_commit_at":50,"category_tags":51,"status":17},6121,"gemini-cli","google-gemini\u002Fgemini-cli","gemini-cli 是一款由谷歌推出的开源 AI 命令行工具，它将强大的 Gemini 大模型能力直接集成到用户的终端环境中。对于习惯在命令行工作的开发者而言，它提供了一条从输入提示词到获取模型响应的最短路径，无需切换窗口即可享受智能辅助。\n\n这款工具主要解决了开发过程中频繁上下文切换的痛点，让用户能在熟悉的终端界面内直接完成代码理解、生成、调试以及自动化运维任务。无论是查询大型代码库、根据草图生成应用，还是执行复杂的 Git 操作，gemini-cli 都能通过自然语言指令高效处理。\n\n它特别适合广大软件工程师、DevOps 人员及技术研究人员使用。其核心亮点包括支持高达 100 万 token 的超长上下文窗口，具备出色的逻辑推理能力；内置 Google 搜索、文件操作及 Shell 命令执行等实用工具；更独特的是，它支持 MCP（模型上下文协议），允许用户灵活扩展自定义集成，连接如图像生成等外部能力。此外，个人谷歌账号即可享受免费的额度支持，且项目基于 Apache 2.0 协议完全开源，是提升终端工作效率的理想助手。",100752,"2026-04-10T01:20:03",[52,13,15,14],"插件",{"id":54,"name":55,"github_repo":56,"description_zh":57,"stars":58,"difficulty_score":32,"last_commit_at":59,"category_tags":60,"status":17},4721,"markitdown","microsoft\u002Fmarkitdown","MarkItDown 是一款由微软 AutoGen 团队打造的轻量级 Python 工具，专为将各类文件高效转换为 Markdown 格式而设计。它支持 PDF、Word、Excel、PPT、图片（含 OCR）、音频（含语音转录）、HTML 乃至 YouTube 链接等多种格式的解析，能够精准提取文档中的标题、列表、表格和链接等关键结构信息。\n\n在人工智能应用日益普及的今天，大语言模型（LLM）虽擅长处理文本，却难以直接读取复杂的二进制办公文档。MarkItDown 恰好解决了这一痛点，它将非结构化或半结构化的文件转化为模型“原生理解”且 Token 效率极高的 Markdown 格式，成为连接本地文件与 AI 分析 pipeline 的理想桥梁。此外，它还提供了 MCP（模型上下文协议）服务器，可无缝集成到 Claude Desktop 等 LLM 应用中。\n\n这款工具特别适合开发者、数据科学家及 AI 研究人员使用，尤其是那些需要构建文档检索增强生成（RAG）系统、进行批量文本分析或希望让 AI 助手直接“阅读”本地文件的用户。虽然生成的内容也具备一定可读性，但其核心优势在于为机器",93400,"2026-04-06T19:52:38",[52,14],{"id":62,"github_repo":63,"name":64,"description_en":65,"description_zh":66,"ai_summary_zh":66,"readme_en":67,"readme_zh":68,"quickstart_zh":69,"use_case_zh":70,"hero_image_url":71,"owner_login":72,"owner_name":73,"owner_avatar_url":74,"owner_bio":75,"owner_company":76,"owner_location":76,"owner_email":77,"owner_twitter":76,"owner_website":78,"owner_url":79,"languages":80,"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":142,"github_topics":144,"view_count":32,"oss_zip_url":76,"oss_zip_packed_at":76,"status":17,"created_at":162,"updated_at":163,"faqs":164,"releases":192},7018,"explosion\u002FspaCy","spaCy","💫 Industrial-strength Natural Language Processing (NLP) in Python","spaCy 是一款专为 Python 打造的高性能自然语言处理（NLP）库，旨在将前沿的学术研究转化为可直接应用于实际产品的工业级解决方案。它主要解决了开发者在处理文本数据时面临的效率瓶颈与技术落地难题，让用户无需从零构建复杂模型，即可快速实现分词、词性标注、句法分析、命名实体识别及文本分类等核心任务。\n\n这款工具非常适合软件工程师、数据科学家以及需要构建生产级 NLP 应用的研究人员使用。无论是初创团队还是大型企业，都能利用 spaCy 稳健地处理多语言文本数据。其独特的技术亮点在于极致的运行速度与现代化的架构设计：底层采用 Cython 优化以确保高效执行，同时深度融合了 BERT 等预训练 Transformer 模型，支持多任务学习。此外，spaCy 提供了覆盖 70 多种语言的预训练管道，并配备了完整的生产就绪型训练系统与便捷的模型部署方案，让从实验到上线的全流程管理变得简单顺畅。作为商业友好的开源项目，它在保持 MIT 许可开放性的同时，确保了企业级应用所需的稳定性与维护支持。","\u003Ca href=\"https:\u002F\u002Fexplosion.ai\">\u003Cimg src=\"https:\u002F\u002Fexplosion.ai\u002Fassets\u002Fimg\u002Flogo.svg\" width=\"125\" height=\"125\" align=\"right\" \u002F>\u003C\u002Fa>\n\n# spaCy: Industrial-strength NLP\n\nspaCy is a library for **advanced Natural Language Processing** in Python and\nCython. It's built on the very latest research, and was designed from day one to\nbe used in real products.\n\nspaCy comes with [pretrained pipelines](https:\u002F\u002Fspacy.io\u002Fmodels) and currently\nsupports tokenization and training for **70+ languages**. It features\nstate-of-the-art speed and **neural network models** for tagging, parsing,\n**named entity recognition**, **text classification** and more, multi-task\nlearning with pretrained **transformers** like BERT, as well as a\nproduction-ready [**training system**](https:\u002F\u002Fspacy.io\u002Fusage\u002Ftraining) and easy\nmodel packaging, deployment and workflow management. spaCy is commercial\nopen-source software, released under the\n[MIT license](https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\u002Fblob\u002Fmaster\u002FLICENSE).\n\n💫 **Version 3.8 out now!**\n[Check out the release notes here.](https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\u002Freleases)\n\n[![tests](https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\u002Factions\u002Fworkflows\u002Ftests.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\u002Factions\u002Fworkflows\u002Ftests.yml)\n[![Current Release Version](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Frelease\u002Fexplosion\u002Fspacy.svg?style=flat-square&logo=github)](https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\u002Freleases)\n[![pypi Version](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fspacy.svg?style=flat-square&logo=pypi&logoColor=white)](https:\u002F\u002Fpypi.org\u002Fproject\u002Fspacy\u002F)\n[![conda Version](https:\u002F\u002Fimg.shields.io\u002Fconda\u002Fvn\u002Fconda-forge\u002Fspacy.svg?style=flat-square&logo=conda-forge&logoColor=white)](https:\u002F\u002Fanaconda.org\u002Fconda-forge\u002Fspacy)\n[![Python wheels](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fwheels-%E2%9C%93-4c1.svg?longCache=true&style=flat-square&logo=python&logoColor=white)](https:\u002F\u002Fgithub.com\u002Fexplosion\u002Fwheelwright\u002Freleases)\n[![Code style: black](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fcode%20style-black-000000.svg?style=flat-square)](https:\u002F\u002Fgithub.com\u002Fambv\u002Fblack)\n\u003Cbr \u002F>\n[![PyPi downloads](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fexplosion_spaCy_readme_ded789c1672a.png)](https:\u002F\u002Fpypi.org\u002Fproject\u002Fspacy\u002F)\n[![Conda downloads](https:\u002F\u002Fimg.shields.io\u002Fconda\u002Fdn\u002Fconda-forge\u002Fspacy?label=conda%20downloads)](https:\u002F\u002Fanaconda.org\u002Fconda-forge\u002Fspacy)\n\n## 📖 Documentation\n\n| Documentation                                                                                                                                                                                                             |                                                                                                                                                                                                                                                                                                                                              |\n| ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n| ⭐️ **[spaCy 101]**                                                                                                                                                                                                       | New to spaCy? Here's everything you need to know!                                                                                                                                                                                                                                                                                            |\n| 📚 **[Usage Guides]**                                                                                                                                                                                                     | How to use spaCy and its features.                                                                                                                                                                                                                                                                                                           |\n| 🚀 **[New in v3.0]**                                                                                                                                                                                                      | New features, backwards incompatibilities and migration guide.                                                                                                                                                                                                                                                                               |\n| 🪐 **[Project Templates]**                                                                                                                                                                                                | End-to-end workflows you can clone, modify and run.                                                                                                                                                                                                                                                                                          |\n| 🎛 **[API Reference]**                                                                                                                                                                                                     | The detailed reference for spaCy's API.                                                                                                                                                                                                                                                                                                      |\n| ⏩ **[GPU Processing]**                                                                                                                                                                                                    | Use spaCy with CUDA-compatible GPU processing.                                                                                                                                                                                                                                                                                               |\n| 📦 **[Models]**                                                                                                                                                                                                           | Download trained pipelines for spaCy.                                                                                                                                                                                                                                                                                                        |\n| 🦙 **[Large Language Models]**                                                                                                                                                                                            | Integrate LLMs into spaCy pipelines.                                                                                                                                                                                                                                                                                                        |\n| 🌌 **[Universe]**                                                                                                                                                                                                         | Plugins, extensions, demos and books from the spaCy ecosystem.                                                                                                                                                                                                                                                                               |\n| ⚙️ **[spaCy VS Code Extension]**                                                                                                                                                                                          | Additional tooling and features for working with spaCy's config files.                                                                                                                                                                                                                                                                       |\n| 👩‍🏫 **[Online Course]**                                                                                                                                                                                                    | Learn spaCy in this free and interactive online course.                                                                                                                                                                                                                                                                                      |\n| 📰 **[Blog]**                                                                                                                                                                                                             | Read about current spaCy and Prodigy development, releases, talks and more from Explosion.                                                                                                                                                                                                                 |\n| 📺 **[Videos]**                                                                                                                                                                                                           | Our YouTube channel with video tutorials, talks and more.                                                                                                                                                                                                                                                                                    |\n| 🔴 **[Live Stream]**                                                                                                                                                                                                       | Join Matt as he works on spaCy and chat about NLP, live every week.                                                                                                                                                                                                                                                                         |\n| 🛠 **[Changelog]**                                                                                                                                                                                                         | Changes and version history.                                                                                                                                                                                                                                                                                                                 |\n| 💝 **[Contribute]**                                                                                                                                                                                                       | How to contribute to the spaCy project and code base.                                                                                                                                                                                                                                                                                        |\n| 👕 **[Swag]**                                                                                                                                                                                                             | Support us and our work with unique, custom-designed swag!                                                                                                                                                                                                                                                                                   |\n| \u003Ca href=\"https:\u002F\u002Fexplosion.ai\u002Ftailored-solutions\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fexplosion_spaCy_readme_f2343422205d.png\" width=\"150\" alt=\"Tailored Solutions\"\u002F>\u003C\u002Fa> | Custom NLP consulting, implementation and strategic advice by spaCy’s core development team. Streamlined, production-ready, predictable and maintainable. Send us an email or take our 5-minute questionnaire, and well'be in touch! **[Learn more →](https:\u002F\u002Fexplosion.ai\u002Ftailored-solutions)**                 |\n\n[spacy 101]: https:\u002F\u002Fspacy.io\u002Fusage\u002Fspacy-101\n[new in v3.0]: https:\u002F\u002Fspacy.io\u002Fusage\u002Fv3\n[usage guides]: https:\u002F\u002Fspacy.io\u002Fusage\u002F\n[api reference]: https:\u002F\u002Fspacy.io\u002Fapi\u002F\n[gpu processing]: https:\u002F\u002Fspacy.io\u002Fusage#gpu\n[models]: https:\u002F\u002Fspacy.io\u002Fmodels\n[large language models]: https:\u002F\u002Fspacy.io\u002Fusage\u002Flarge-language-models\n[universe]: https:\u002F\u002Fspacy.io\u002Funiverse\n[spacy vs code extension]: https:\u002F\u002Fgithub.com\u002Fexplosion\u002Fspacy-vscode\n[videos]: https:\u002F\u002Fwww.youtube.com\u002Fc\u002FExplosionAI\n[live stream]: https:\u002F\u002Fwww.youtube.com\u002Fplaylist?list=PLBmcuObd5An5_iAxNYLJa_xWmNzsYce8c\n[online course]: https:\u002F\u002Fcourse.spacy.io\n[blog]: https:\u002F\u002Fexplosion.ai\n[project templates]: https:\u002F\u002Fgithub.com\u002Fexplosion\u002Fprojects\n[changelog]: https:\u002F\u002Fspacy.io\u002Fusage#changelog\n[contribute]: https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\u002Fblob\u002Fmaster\u002FCONTRIBUTING.md\n[swag]: https:\u002F\u002Fexplosion.ai\u002Fmerch\n\n## 💬 Where to ask questions\n\nThe spaCy project is maintained by the [spaCy team](https:\u002F\u002Fexplosion.ai\u002Fabout).\nPlease understand that we won't be able to provide individual support via email.\nWe also believe that help is much more valuable if it's shared publicly, so that\nmore people can benefit from it.\n\n| Type                            | Platforms                               |\n| ------------------------------- | --------------------------------------- |\n| 🚨 **Bug Reports**              | [GitHub Issue Tracker]                  |\n| 🎁 **Feature Requests & Ideas** | [GitHub Discussions] · [Live Stream]    |\n| 👩‍💻 **Usage Questions**          | [GitHub Discussions] · [Stack Overflow] |\n| 🗯 **General Discussion**        | [GitHub Discussions] · [Live Stream]   |\n\n[github issue tracker]: https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\u002Fissues\n[github discussions]: https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\u002Fdiscussions\n[stack overflow]: https:\u002F\u002Fstackoverflow.com\u002Fquestions\u002Ftagged\u002Fspacy\n[live stream]: https:\u002F\u002Fwww.youtube.com\u002Fplaylist?list=PLBmcuObd5An5_iAxNYLJa_xWmNzsYce8c\n\n## Features\n\n- Support for **70+ languages**\n- **Trained pipelines** for different languages and tasks\n- Multi-task learning with pretrained **transformers** like BERT\n- Support for pretrained **word vectors** and embeddings\n- State-of-the-art speed\n- Production-ready **training system**\n- Linguistically-motivated **tokenization**\n- Components for named **entity recognition**, part-of-speech-tagging,\n  dependency parsing, sentence segmentation, **text classification**,\n  lemmatization, morphological analysis, entity linking and more\n- Easily extensible with **custom components** and attributes\n- Support for custom models in **PyTorch**, **TensorFlow** and other frameworks\n- Built in **visualizers** for syntax and NER\n- Easy **model packaging**, deployment and workflow management\n- Robust, rigorously evaluated accuracy\n\n📖 **For more details, see the\n[facts, figures and benchmarks](https:\u002F\u002Fspacy.io\u002Fusage\u002Ffacts-figures).**\n\n## ⏳ Install spaCy\n\nFor detailed installation instructions, see the\n[documentation](https:\u002F\u002Fspacy.io\u002Fusage).\n\n- **Operating system**: macOS \u002F OS X · Linux · Windows (Cygwin, MinGW, Visual\n  Studio)\n- **Python version**: Python >=3.7, \u003C3.13 (only 64 bit)\n- **Package managers**: [pip] · [conda] (via `conda-forge`)\n\n[pip]: https:\u002F\u002Fpypi.org\u002Fproject\u002Fspacy\u002F\n[conda]: https:\u002F\u002Fanaconda.org\u002Fconda-forge\u002Fspacy\n\n### pip\n\nUsing pip, spaCy releases are available as source packages and binary wheels.\nBefore you install spaCy and its dependencies, make sure that your `pip`,\n`setuptools` and `wheel` are up to date.\n\n```bash\npip install -U pip setuptools wheel\npip install spacy\n```\n\nTo install additional data tables for lemmatization and normalization you can\nrun `pip install spacy[lookups]` or install\n[`spacy-lookups-data`](https:\u002F\u002Fgithub.com\u002Fexplosion\u002Fspacy-lookups-data)\nseparately. The lookups package is needed to create blank models with\nlemmatization data, and to lemmatize in languages that don't yet come with\npretrained models and aren't powered by third-party libraries.\n\nWhen using pip it is generally recommended to install packages in a virtual\nenvironment to avoid modifying system state:\n\n```bash\npython -m venv .env\nsource .env\u002Fbin\u002Factivate\npip install -U pip setuptools wheel\npip install spacy\n```\n\n### conda\n\nYou can also install spaCy from `conda` via the `conda-forge` channel. For the\nfeedstock including the build recipe and configuration, check out\n[this repository](https:\u002F\u002Fgithub.com\u002Fconda-forge\u002Fspacy-feedstock).\n\n```bash\nconda install -c conda-forge spacy\n```\n\n### Updating spaCy\n\nSome updates to spaCy may require downloading new statistical models. If you're\nrunning spaCy v2.0 or higher, you can use the `validate` command to check if\nyour installed models are compatible and if not, print details on how to update\nthem:\n\n```bash\npip install -U spacy\npython -m spacy validate\n```\n\nIf you've trained your own models, keep in mind that your training and runtime\ninputs must match. After updating spaCy, we recommend **retraining your models**\nwith the new version.\n\n📖 **For details on upgrading from spaCy 2.x to spaCy 3.x, see the\n[migration guide](https:\u002F\u002Fspacy.io\u002Fusage\u002Fv3#migrating).**\n\n## 📦 Download model packages\n\nTrained pipelines for spaCy can be installed as **Python packages**. This means\nthat they're a component of your application, just like any other module. Models\ncan be installed using spaCy's [`download`](https:\u002F\u002Fspacy.io\u002Fapi\u002Fcli#download)\ncommand, or manually by pointing pip to a path or URL.\n\n| Documentation              |                                                                  |\n| -------------------------- | ---------------------------------------------------------------- |\n| **[Available Pipelines]**  | Detailed pipeline descriptions, accuracy figures and benchmarks. |\n| **[Models Documentation]** | Detailed usage and installation instructions.                    |\n| **[Training]**             | How to train your own pipelines on your data.                    |\n\n[available pipelines]: https:\u002F\u002Fspacy.io\u002Fmodels\n[models documentation]: https:\u002F\u002Fspacy.io\u002Fusage\u002Fmodels\n[training]: https:\u002F\u002Fspacy.io\u002Fusage\u002Ftraining\n\n```bash\n# Download best-matching version of specific model for your spaCy installation\npython -m spacy download en_core_web_sm\n\n# pip install .tar.gz archive or .whl from path or URL\npip install \u002FUsers\u002Fyou\u002Fen_core_web_sm-3.0.0.tar.gz\npip install \u002FUsers\u002Fyou\u002Fen_core_web_sm-3.0.0-py3-none-any.whl\npip install https:\u002F\u002Fgithub.com\u002Fexplosion\u002Fspacy-models\u002Freleases\u002Fdownload\u002Fen_core_web_sm-3.0.0\u002Fen_core_web_sm-3.0.0.tar.gz\n```\n\n### Loading and using models\n\nTo load a model, use [`spacy.load()`](https:\u002F\u002Fspacy.io\u002Fapi\u002Ftop-level#spacy.load)\nwith the model name or a path to the model data directory.\n\n```python\nimport spacy\nnlp = spacy.load(\"en_core_web_sm\")\ndoc = nlp(\"This is a sentence.\")\n```\n\nYou can also `import` a model directly via its full name and then call its\n`load()` method with no arguments.\n\n```python\nimport spacy\nimport en_core_web_sm\n\nnlp = en_core_web_sm.load()\ndoc = nlp(\"This is a sentence.\")\n```\n\n📖 **For more info and examples, check out the\n[models documentation](https:\u002F\u002Fspacy.io\u002Fdocs\u002Fusage\u002Fmodels).**\n\n## ⚒ Compile from source\n\nThe other way to install spaCy is to clone its\n[GitHub repository](https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy) and build it from\nsource. That is the common way if you want to make changes to the code base.\nYou'll need to make sure that you have a development environment consisting of a\nPython distribution including header files, a compiler,\n[pip](https:\u002F\u002Fpip.pypa.io\u002Fen\u002Flatest\u002Finstalling\u002F),\n[virtualenv](https:\u002F\u002Fvirtualenv.pypa.io\u002Fen\u002Flatest\u002F) and\n[git](https:\u002F\u002Fgit-scm.com) installed. The compiler part is the trickiest. How to\ndo that depends on your system.\n\n| Platform    |                                                                                                                                                                                                                                                                     |\n| ----------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n| **Ubuntu**  | Install system-level dependencies via `apt-get`: `sudo apt-get install build-essential python-dev git` .                                                                                                                                                            |\n| **Mac**     | Install a recent version of [XCode](https:\u002F\u002Fdeveloper.apple.com\u002Fxcode\u002F), including the so-called \"Command Line Tools\". macOS and OS X ship with Python and git preinstalled.                                                                                        |\n| **Windows** | Install a version of the [Visual C++ Build Tools](https:\u002F\u002Fvisualstudio.microsoft.com\u002Fvisual-cpp-build-tools\u002F) or [Visual Studio Express](https:\u002F\u002Fvisualstudio.microsoft.com\u002Fvs\u002Fexpress\u002F) that matches the version that was used to compile your Python interpreter. |\n\nFor more details and instructions, see the documentation on\n[compiling spaCy from source](https:\u002F\u002Fspacy.io\u002Fusage#source) and the\n[quickstart widget](https:\u002F\u002Fspacy.io\u002Fusage#section-quickstart) to get the right\ncommands for your platform and Python version.\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\ncd spaCy\n\npython -m venv .env\nsource .env\u002Fbin\u002Factivate\n\n# make sure you are using the latest pip\npython -m pip install -U pip setuptools wheel\n\npip install -r requirements.txt\npip install --no-build-isolation --editable .\n```\n\nTo install with extras:\n\n```bash\npip install --no-build-isolation --editable .[lookups,cuda102]\n```\n\n## 🚦 Run tests\n\nspaCy comes with an [extensive test suite](spacy\u002Ftests). In order to run the\ntests, you'll usually want to clone the repository and build spaCy from source.\nThis will also install the required development dependencies and test utilities\ndefined in the [`requirements.txt`](requirements.txt).\n\nAlternatively, you can run `pytest` on the tests from within the installed\n`spacy` package. Don't forget to also install the test utilities via spaCy's\n[`requirements.txt`](requirements.txt):\n\n```bash\npip install -r requirements.txt\npython -m pytest --pyargs spacy\n```\n","\u003Ca href=\"https:\u002F\u002Fexplosion.ai\">\u003Cimg src=\"https:\u002F\u002Fexplosion.ai\u002Fassets\u002Fimg\u002Flogo.svg\" width=\"125\" height=\"125\" align=\"right\" \u002F>\u003C\u002Fa>\n\n# spaCy：工业级自然语言处理\n\nspaCy 是一个用于 Python 和 Cython 的 **高级自然语言处理** 库。它基于最新的研究成果，并从一开始就专为实际产品开发而设计。\n\nspaCy 提供了 [预训练的管道](https:\u002F\u002Fspacy.io\u002Fmodels)，目前支持 **70 多种语言** 的分词和模型训练。它拥有最先进的速度和用于标注、句法分析、**命名实体识别**、**文本分类** 等任务的 **神经网络模型**，并支持与 BERT 等预训练 **Transformer 模型** 结合的多任务学习。此外，spaCy 还配备了一个生产就绪的 [**训练系统**](https:\u002F\u002Fspacy.io\u002Fusage\u002Ftraining)，以及简便的模型打包、部署和工作流管理功能。spaCy 是商业开源软件，采用 **MIT 许可证** 发布。\n\n💫 **版本 3.8 已发布！**\n[请在此处查看发行说明。](https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\u002Freleases)\n\n[![测试](https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\u002Factions\u002Fworkflows\u002Ftests.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\u002Factions\u002Fworkflows\u002Ftests.yml)\n[![当前发布版本](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Frelease\u002Fexplosion\u002Fspacy.svg?style=flat-square&logo=github)](https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\u002Freleases)\n[![PyPI 版本](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fspacy.svg?style=flat-square&logo=pypi&logoColor=white)](https:\u002F\u002Fpypi.org\u002Fproject\u002Fspacy\u002F)\n[![Conda 版本](https:\u002F\u002Fimg.shields.io\u002Fconda\u002Fvn\u002Fconda-forge\u002Fspacy.svg?style=flat-square&logo=conda-forge&logoColor=white)](https:\u002F\u002Fanaconda.org\u002Fconda-forge\u002Fspacy)\n[![Python 轮子包](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fwheels-%E2%9C%93-4c1.svg?longCache=true&style=flat-square&logo=python&logoColor=white)](https:\u002F\u002Fgithub.com\u002Fexplosion\u002Fwheelwright\u002Freleases)\n[![代码风格：black](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fcode%20style-black-000000.svg?style=flat-square)](https:\u002F\u002Fgithub.com\u002Fambv\u002Fblack)\n\u003Cbr \u002F>\n[![PyPI 下载量](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fexplosion_spaCy_readme_ded789c1672a.png)](https:\u002F\u002Fpypi.org\u002Fproject\u002Fspacy\u002F)\n[![Conda 下载量](https:\u002F\u002Fimg.shields.io\u002Fconda\u002Fdn\u002Fconda-forge\u002Fspacy?label=conda%20downloads)](https:\u002F\u002Fanaconda.org\u002Fconda-forge\u002Fspacy)\n\n## 📖 文档\n\n| 文档                                                                                                                                                                                                             |                                                                                                                                                                                                                                                                                                                                              |\n| ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n| ⭐️ **[spaCy 101]**                                                                                                                                                                                                       | 刚接触 spaCy？这里有一切你需要了解的内容！                                                                                                                                                                                                                                                                                            |\n| 📚 **[使用指南]**                                                                                                                                                                                                         | 如何使用 spaCy 及其各项功能。                                                                                                                                                                                                                                                                                                           |\n| 🚀 **[v3.0 新特性]**                                                                                                                                                                                                      | 新功能、向后不兼容变更及迁移指南。                                                                                                                                                                                                                                                                                                       |\n| 🪐 **[项目模板]**                                                                                                                                                                                                        | 你可以克隆、修改并运行的端到端工作流。                                                                                                                                                                                                                                                                                          |\n| 🎛 **[API 参考]**                                                                                                                                                                                                         | spaCy API 的详细参考文档。                                                                                                                                                                                                                                                                                                                 |\n| ⏩ **[GPU 处理]**                                                                                                                                                                                                         | 使用支持 CUDA 的 GPU 来加速 spaCy 的处理。                                                                                                                                                                                                                                                                                               |\n| 📦 **[模型]**                                                                                                                                                                                                             | 下载适用于 spaCy 的预训练管道。                                                                                                                                                                                                                                                                                                        |\n| 🦙 **[大型语言模型]**                                                                                                                                                                                                    | 将 LLM 集成到 spaCy 管道中。                                                                                                                                                                                                                                                                                                        |\n| 🌌 **[生态社区]**                                                                                                                                                                                                         | spaCy 生态系统中的插件、扩展、演示和书籍。                                                                                                                                                                                                                                                                               |\n| ⚙️ **[spaCy VS Code 扩展]**                                                                                                                                                                                              | 用于处理 spaCy 配置文件的额外工具和功能。                                                                                                                                                                                                                                                                                    |\n| 👩‍🏫 **[在线课程]**                                                                                                                                                                                                      | 在这门免费且互动式的在线课程中学习 spaCy。                                                                                                                                                                                                                                                                                      |\n| 📰 **[博客]**                                                                                                                                                                                                             | 阅读 Explosion 公司关于 spaCy 和 Prodigy 的最新开发进展、版本发布、技术分享等内容。                                                                                                                                                                                                                 |\n| 📺 **[视频]**                                                                                                                                                                                                             | 我们的 YouTube 频道，提供视频教程、技术演讲等丰富内容。                                                                                                                                                                                                                                                                                    |\n| 🔴 **[直播]**                                                                                                                                                                                                             | 每周与 Matt 一起实时开发 spaCy，并畅聊自然语言处理相关话题。                                                                                                                                                                                                                                                                         |\n| 🛠 **[更新日志]**                                                                                                                                                                                                         | 变更记录与版本历史。                                                                                                                                                                                                                                                                                                                         |\n| 💝 **[贡献]**                                                                                                                                                                                                             | 如何为 spaCy 项目及其代码库做出贡献。                                                                                                                                                                                                                                                                                        |\n| 👕 **[周边商品]**                                                                                                                                                                                                         | 通过独特、定制设计的周边商品来支持我们及我们的工作！                                                                                                                                                                                                                                                                                   |\n| \u003Ca href=\"https:\u002F\u002Fexplosion.ai\u002Ftailored-solutions\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fexplosion_spaCy_readme_f2343422205d.png\" width=\"150\" alt=\"定制解决方案\"\u002F>\u003C\u002Fa> | 由 spaCy 核心开发团队提供的定制化 NLP 咨询、实施及战略建议。流程精简、可直接投入生产、结果可预测且易于维护。发送邮件给我们或填写我们的 5 分钟问卷，我们将尽快与你联系！ **[了解更多 →](https:\u002F\u002Fexplosion.ai\u002Ftailored-solutions)**                 |\n\n[spaCy 101]: https:\u002F\u002Fspacy.io\u002Fusage\u002Fspacy-101\n[v3.0 新特性]: https:\u002F\u002Fspacy.io\u002Fusage\u002Fv3\n[使用指南]: https:\u002F\u002Fspacy.io\u002Fusage\u002F\n[API 参考文档]: https:\u002F\u002Fspacy.io\u002Fapi\u002F\n[GPU 处理]: https:\u002F\u002Fspacy.io\u002Fusage#gpu\n[模型]: https:\u002F\u002Fspacy.io\u002Fmodels\n[大型语言模型]: https:\u002F\u002Fspacy.io\u002Fusage\u002Flarge-language-models\n[宇宙]: https:\u002F\u002Fspacy.io\u002Funiverse\n[spaCy VS Code 扩展]: https:\u002F\u002Fgithub.com\u002Fexplosion\u002Fspacy-vscode\n[视频]: https:\u002F\u002Fwww.youtube.com\u002Fc\u002FExplosionAI\n[直播]: https:\u002F\u002Fwww.youtube.com\u002Fplaylist?list=PLBmcuObd5An5_iAxNYLJa_xWmNzsYce8c\n[在线课程]: https:\u002F\u002Fcourse.spacy.io\n[博客]: https:\u002F\u002Fexplosion.ai\n[项目模板]: https:\u002F\u002Fgithub.com\u002Fexplosion\u002Fprojects\n[变更日志]: https:\u002F\u002Fspacy.io\u002Fusage#changelog\n[贡献指南]: https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\u002Fblob\u002Fmaster\u002FCONTRIBUTING.md\n[周边商品]: https:\u002F\u002Fexplosion.ai\u002Fmerch\n\n\n\n## 💬 提问渠道\n\nspaCy 项目由 [spaCy 团队](https:\u002F\u002Fexplosion.ai\u002Fabout) 维护。请理解，我们无法通过电子邮件提供一对一的支持。同时，我们认为公开分享的帮助更有价值，这样更多人可以从中受益。\n\n| 类型                            | 平台                               |\n| ------------------------------- | --------------------------------------- |\n| 🚨 **Bug 报告**              | [GitHub 问题追踪器]                  |\n| 🎁 **功能请求与建议**         | [GitHub 讨论区] · [直播]    |\n| 👩‍💻 **使用问题**          | [GitHub 讨论区] · [Stack Overflow] |\n| 🗯 **一般讨论**        | [GitHub 讨论区] · [直播]   |\n\n[github 问题追踪器]: https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\u002Fissues\n[github 讨论区]: https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\u002Fdiscussions\n[stack overflow]: https:\u002F\u002Fstackoverflow.com\u002Fquestions\u002Ftagged\u002Fspacy\n[live stream]: https:\u002F\u002Fwww.youtube.com\u002Fplaylist?list=PLBmcuObd5An5_iAxNYLJa_xWmNzsYce8c\n\n## 特性\n\n- 支持 **70+ 种语言**\n- 针对不同语言和任务的 **预训练管道**\n- 使用 BERT 等预训练 **Transformer** 模型进行多任务学习\n- 支持预训练的 **词向量** 和嵌入\n- 行业领先的运行速度\n- 生产就绪的 **训练系统**\n- 基于语言学原理的 **分词**\n- 用于命名实体识别、词性标注、依存句法分析、句子切分、**文本分类**、词形还原、形态分析、实体链接等功能的组件\n- 可通过 **自定义组件** 和属性轻松扩展\n- 支持在 PyTorch、TensorFlow 等框架中使用自定义模型\n- 内置语法和 NER 的 **可视化工具**\n- 简单的 **模型打包**、部署和工作流管理\n- 稳健且经过严格评估的准确性\n\n📖 **更多详情，请参阅\n[事实、数据与基准测试](https:\u002F\u002Fspacy.io\u002Fusage\u002Ffacts-figures)。**\n\n## ⏳ 安装 spaCy\n\n详细的安装说明请参阅\n[文档](https:\u002F\u002Fspacy.io\u002Fusage)。\n\n- **操作系统**: macOS \u002F OS X · Linux · Windows (Cygwin, MinGW, Visual Studio)\n- **Python 版本**: Python >=3.7, \u003C3.13（仅 64 位）\n- **包管理器**: [pip] · [conda]（通过 `conda-forge`）\n\n[pip]: https:\u002F\u002Fpypi.org\u002Fproject\u002Fspacy\u002F\n[conda]: https:\u002F\u002Fanaconda.org\u002Fconda-forge\u002Fspacy\n\n### pip\n\n通过 pip，spaCy 发布了源码包和二进制 wheel 文件。在安装 spaCy 及其依赖项之前，请确保您的 `pip`、`setuptools` 和 `wheel` 均为最新版本。\n\n```bash\npip install -U pip setuptools wheel\npip install spacy\n```\n\n若需安装用于词形还原和归一化的额外数据表，可运行 `pip install spacy[lookups]`，或单独安装\n[`spacy-lookups-data`](https:\u002F\u002Fgithub.com\u002Fexplosion\u002Fspacy-lookups-data)。该 lookups 包用于创建包含词形还原数据的空白模型，以及对尚未提供预训练模型且未借助第三方库的语言进行词形还原。\n\n使用 pip 时，通常建议在虚拟环境中安装软件包，以避免修改系统状态：\n\n```bash\npython -m venv .env\nsource .env\u002Fbin\u002Factivate\npip install -U pip setuptools wheel\npip install spacy\n```\n\n### conda\n\n您也可以通过 `conda-forge` 通道从 conda 安装 spaCy。有关包含构建配方和配置的 feedstock，请查看\n[此仓库](https:\u002F\u002Fgithub.com\u002Fconda-forge\u002Fspacy-feedstock)。\n\n```bash\nconda install -c conda-forge spacy\n```\n\n### 更新 spaCy\n\n某些 spaCy 更新可能需要下载新的统计模型。如果您正在使用 spaCy v2.0 或更高版本，可以使用 `validate` 命令检查已安装的模型是否兼容；如果不兼容，则会显示更新方法的详细信息：\n\n```bash\npip install -U spacy\npython -m spacy validate\n```\n\n如果您曾训练过自己的模型，请注意训练输入和运行时输入必须一致。更新 spaCy 后，我们建议您使用新版本 **重新训练模型**。\n\n📖 **关于从 spaCy 2.x 升级到 spaCy 3.x 的详细信息，请参阅\n[迁移指南](https:\u002F\u002Fspacy.io\u002Fusage\u002Fv3#migrating)。**\n\n## 📦 下载模型包\n\nspaCy 的预训练管道可以作为 **Python 包** 安装。这意味着它们是您应用程序的一部分，与其他模块无异。模型可以通过 spaCy 的 [`download`](https:\u002F\u002Fspacy.io\u002Fapi\u002Fcli#download) 命令安装，也可以手动通过指定路径或 URL 进行安装。\n\n| 文档              |                                                                  |\n| -------------------------- | ---------------------------------------------------------------- |\n| **[可用管道]**  | 详细的管道描述、准确率数据和基准测试。                         |\n| **[模型文档]** | 详细的使用和安装说明。                                         |\n| **[训练]**             | 如何基于您的数据训练自定义管道。                    |\n\n[available pipelines]: https:\u002F\u002Fspacy.io\u002Fmodels\n[models documentation]: https:\u002F\u002Fspacy.io\u002Fusage\u002Fmodels\n[training]: https:\u002F\u002Fspacy.io\u002Fusage\u002Ftraining\n\n```bash\n# 为您的 spaCy 安装下载最匹配的特定模型版本\npython -m spacy download en_core_web_sm\n\n# 使用 pip 安装 .tar.gz 归档或来自路径或 URL 的 .whl 文件\npip install \u002FUsers\u002Fyou\u002Fen_core_web_sm-3.0.0.tar.gz\npip install \u002FUsers\u002Fyou\u002Fen_core_web_sm-3.0.0-py3-none-any.whl\npip install https:\u002F\u002Fgithub.com\u002Fexplosion\u002Fspacy-models\u002Freleases\u002Fdownload\u002Fen_core_web_sm-3.0.0\u002Fen_core_web_sm-3.0.0.tar.gz\n```\n\n### 加载和使用模型\n\n要加载一个模型，可以使用 [`spacy.load()`](https:\u002F\u002Fspacy.io\u002Fapi\u002Ftop-level#spacy.load)，传入模型名称或指向模型数据目录的路径。\n\n```python\nimport spacy\nnlp = spacy.load(\"en_core_web_sm\")\ndoc = nlp(\"This is a sentence.\")\n```\n\n你也可以直接通过模型的完整名称导入它，然后调用其 `load()` 方法，无需传递任何参数。\n\n```python\nimport spacy\nimport en_core_web_sm\n\nnlp = en_core_web_sm.load()\ndoc = nlp(\"This is a sentence.\")\n```\n\n📖 **更多详细信息和示例，请参阅\n[模型文档](https:\u002F\u002Fspacy.io\u002Fdocs\u002Fusage\u002Fmodels)。**\n\n## ⚒ 从源代码编译\n\n安装 spaCy 的另一种方式是克隆其\n[GitHub 仓库](https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy)，然后从源代码构建。如果你希望对代码库进行修改，这种方式通常更为常见。你需要确保已搭建好开发环境，包括带有头文件的 Python 发行版、编译器、\n[pip](https:\u002F\u002Fpip.pypa.io\u002Fen\u002Flatest\u002Finstalling\u002F)、\n[virtualenv](https:\u002F\u002Fvirtualenv.pypa.io\u002Fen\u002Flatest\u002F) 和\n[git](https:\u002F\u002Fgit-scm.com)。其中编译器的配置最为复杂，具体操作取决于你的操作系统。\n\n| 平台    |                                                                                                                                                                                                                                                                     |\n| ----------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n| **Ubuntu**  | 通过 `apt-get` 安装系统级依赖：`sudo apt-get install build-essential python-dev git` 。                                                                                                                                                            |\n| **Mac**     | 安装最新版本的 [XCode](https:\u002F\u002Fdeveloper.apple.com\u002Fxcode\u002F)及其“命令行工具”。macOS 和 OS X 系统默认预装了 Python 和 git。                                                                                        |\n| **Windows** | 安装与你的 Python 解释器编译时所使用的版本相匹配的 [Visual C++ Build Tools](https:\u002F\u002Fvisualstudio.microsoft.com\u002Fvisual-cpp-build-tools\u002F) 或 [Visual Studio Express](https:\u002F\u002Fvisualstudio.microsoft.com\u002Fvs\u002Fexpress\u002F)。 |\n\n有关更详细的说明，请参阅关于\n[从源代码编译 spaCy](https:\u002F\u002Fspacy.io\u002Fusage#source) 的文档以及\n[快速入门小工具](https:\u002F\u002Fspacy.io\u002Fusage#section-quickstart)，以获取适用于你所在平台和 Python 版本的正确命令。\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\ncd spaCy\n\npython -m venv .env\nsource .env\u002Fbin\u002Factivate\n\n# 确保使用最新版本的 pip\npython -m pip install -U pip setuptools wheel\n\npip install -r requirements.txt\npip install --no-build-isolation --editable .\n```\n\n若需安装附加组件：\n\n```bash\npip install --no-build-isolation --editable .[lookups,cuda102]\n```\n\n## 🚦 运行测试\n\nspaCy 自带一套\n[全面的测试套件](spacy\u002Ftests)。为了运行这些测试，通常需要先克隆仓库并从源代码构建 spaCy。这样做还会安装在 [`requirements.txt`](requirements.txt) 中定义的必要开发依赖项和测试工具。\n\n另外，你也可以在已安装的 `spacy` 包中直接运行 `pytest` 来执行测试。别忘了先通过 spaCy 的\n[`requirements.txt`](requirements.txt) 安装测试所需的工具：\n\n```bash\npip install -r requirements.txt\npython -m pytest --pyargs spacy\n```","# spaCy 快速上手指南\n\nspaCy 是一个用于 Python 和 Cython 的**工业级自然语言处理（NLP）**库。它基于最新的研究成果构建，专为实际产品设计，支持 70 多种语言的标记化、训练、命名实体识别、文本分类等功能，并提供高性能的神经网络模型。\n\n## 环境准备\n\n*   **操作系统**：支持 Linux、macOS 和 Windows。\n*   **Python 版本**：需要 Python 3.8+。\n*   **编译器**：在 Windows 上安装可能需要 Microsoft C++ Build Tools；Linux\u002FmacOS 通常需要 gcc\u002Fclang。\n*   **依赖管理**：建议使用 `venv` 或 `conda` 创建独立的虚拟环境。\n\n## 安装步骤\n\n### 1. 安装 spaCy 核心库\n\n推荐使用国内镜像源（如清华源）以加速下载：\n\n```bash\npip install -i https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple spacy\n```\n\n或者使用 conda 安装：\n\n```bash\nconda install -c conda-forge spacy\n```\n\n### 2. 下载预训练模型\n\nspaCy 本身不包含模型数据，需根据目标语言单独下载。以下以英文小模型（`en_core_web_sm`）和中文小模型（`zh_core_web_sm`）为例：\n\n**直接下载（推荐）：**\n```bash\npython -m spacy download en_core_web_sm\npython -m spacy download zh_core_web_sm\n```\n\n**若网络受限，可先下载 wheel 文件后离线安装：**\n访问 [https:\u002F\u002Fgithub.com\u002Fexplosion\u002Fspacy-models\u002Freleases](https:\u002F\u002Fgithub.com\u002Fexplosion\u002Fspacy-models\u002Freleases) 下载对应 `.whl` 文件，然后执行：\n```bash\npip install en_core_web_sm-3.8.0-py3-none-any.whl\n```\n*(请将文件名替换为实际下载的版本)*\n\n## 基本使用\n\n以下示例展示如何加载模型并进行分词、词性标注和命名实体识别。\n\n### Python 代码示例\n\n```python\nimport spacy\n\n# 加载中文预训练模型 (确保已运行: python -m spacy download zh_core_web_sm)\nnlp = spacy.load(\"zh_core_web_sm\")\n\n# 待处理的文本\ntext = \"埃隆·马斯克于 2002 年创立了 SpaceX，总部位于美国加利福尼亚州。\"\n\n# 处理文本\ndoc = nlp(text)\n\n# 遍历分词结果\nprint(\"分词与词性标注：\")\nfor token in doc:\n    print(f\"{token.text} \\t {token.pos_}\")\n\n# 提取命名实体\nprint(\"\\n命名实体识别：\")\nfor ent in doc.ents:\n    print(f\"{ent.text} \\t {ent.label_}\")\n```\n\n### 命令行快速测试\n\n你也可以直接在终端使用 `spacy` 命令快速测试模型效果：\n\n```bash\npython -m spacy predict zh_core_web_sm --text \"苹果公司计划在加州发布新产品\"\n```","某电商公司的数据团队需要每天从全球各地的用户评论中提取产品缺陷和提及的品牌，以生成质量监控报告。\n\n### 没有 spaCy 时\n- 开发人员不得不手动编写复杂的正则表达式来提取人名、地名和品牌名，一旦遇到拼写变体或非标准格式，提取准确率极低。\n- 处理多语言评论（如英语、德语、日语混合）时，需要为每种语言单独寻找并集成不同的轻量级库，导致代码库臃肿且维护困难。\n- 由于缺乏高效的词性标注和依存句法分析，难以判断用户是在“赞扬电池续航”还是“抱怨电池发热”，情感倾向分析经常出错。\n- 面对百万级的日增评论数据，基于纯 Python 编写的传统 NLP 脚本运行缓慢，往往需要数小时才能完成当日数据的预处理。\n\n### 使用 spaCy 后\n- 直接调用 spaCy 预训练的命名实体识别（NER）模型，无需编写规则即可精准识别全球主流品牌及特定产品部件，大幅降低漏检率。\n- 利用其支持的 70 多种语言管道，同一套代码架构即可无缝切换处理不同语种的评论，显著简化了工程部署流程。\n- 借助高精度的依存句法分析，系统能准确捕捉形容词与名词的修饰关系，从而正确区分用户对同一功能点的正反两面评价。\n- 凭借底层 Cython 优化和工业级设计，数据处理速度提升数十倍，原本需数小时的任务现在仅需几分钟即可完成，实现了近实时监控。\n\nspaCy 将原本繁琐脆弱的文本清洗工作转化为高效、精准的标准化流程，让团队能将精力真正聚焦于业务洞察而非底层算法调试。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fexplosion_spaCy_551af06f.png","explosion","Explosion","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fexplosion_6bc9ac84.png","Software company specializing in developer tools and tailored solutions for AI and Natural Language Processing",null,"contact@explosion.ai","https:\u002F\u002Fexplosion.ai","https:\u002F\u002Fgithub.com\u002Fexplosion",[81,85,89,93,97,101,105,109,113,116],{"name":82,"color":83,"percentage":84},"Python","#3572A5",54.1,{"name":86,"color":87,"percentage":88},"MDX","#fcb32c",31.2,{"name":90,"color":91,"percentage":92},"Cython","#fedf5b",10.5,{"name":94,"color":95,"percentage":96},"JavaScript","#f1e05a",2.6,{"name":98,"color":99,"percentage":100},"Sass","#a53b70",0.8,{"name":102,"color":103,"percentage":104},"TypeScript","#3178c6",0.4,{"name":106,"color":107,"percentage":108},"Jinja","#a52a22",0.2,{"name":110,"color":111,"percentage":112},"C","#555555",0.1,{"name":114,"color":115,"percentage":112},"HTML","#e34c26",{"name":117,"color":118,"percentage":119},"Makefile","#427819",0,33455,4672,"2026-04-12T23:51:00","MIT",1,"Linux, macOS, Windows","可选（非必需）。如需加速需 NVIDIA GPU 及 CUDA 兼容环境，具体型号和显存取决于所选模型大小。","未说明（取决于模型大小，小型模型仅需几百 MB，大型 Transformer 模型推荐 16GB+）",{"notes":129,"python":130,"dependencies":131},"spaCy 支持 70 多种语言，提供预训练管道。若需使用 BERT 等 Transformer 模型进行多任务学习，需额外安装 PyTorch 或 TensorFlow 以及 transformers 库。建议使用虚拟环境（如 venv 或 conda）进行安装。生产环境部署可使用其内置的训练系统和模型打包功能。","3.8+",[132,133,134,135,136,137,138,139,140,141],"numpy>=1.15.0","murmurhash>=0.28.0","cymem>=2.0.2","preshed>=3.0.2","thinc>=8.2.0","wasabi>=0.9.1","srsly>=2.4.0","jinja2","setuptools","packaging>=20.0",[143,16,14,35,15,13],"其他",[145,146,147,148,149,150,151,152,153,154,155,156,157,158,159,160,161],"natural-language-processing","data-science","machine-learning","python","cython","nlp","artificial-intelligence","ai","spacy","nlp-library","neural-network","neural-networks","deep-learning","named-entity-recognition","entity-linking","text-classification","tokenization","2026-03-27T02:49:30.150509","2026-04-13T13:48:22.121208",[165,170,175,180,184,188],{"id":166,"question_zh":167,"answer_zh":168,"source_url":169},31587,"如何在 spaCy v2.0 中合并名词短语（noun chunks）？","在 spaCy v2.0 中，你需要遍历 `doc.noun_chunks` 中的每个名词短语并调用 `merge` 方法。关键是要重新分配标签（tag）、词形（lemma）和实体类型（ent_type）。代码示例如下：\n\n```python\nfor np in list(doc.noun_chunks):\n    np.merge(tag=np.root.tag_, lemma=np.root.lemma_, ent_type=np.root.ent_type_)\n```\n\n或者使用关键字参数形式：\n```python\nspan.merge(tag=tag, lemma=lemma, ent_type=ent_type)\n```","https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\u002Fissues\u002F1105",{"id":171,"question_zh":172,"answer_zh":173,"source_url":174},31588,"spaCy 是否支持日语模型？如何获取？","是的，spaCy 已经支持日语模型。首个官方日语模型随 spaCy 2.3.2 版本发布。如果你在使用日语模型时遇到准确率下降的问题，请确保安装了正确版本的依赖库 `sudachipy`。建议指定版本为 `sudachipy>=0.4.8`，因为该版本修复了之前版本（0.4.6-0.4.7）中报告的准确性问题。","https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\u002Fissues\u002F3756",{"id":176,"question_zh":177,"answer_zh":178,"source_url":179},31589,"如何在 Windows (MinGW) 上从源码构建 spaCy？","早期版本在 MinGW 下编译可能会遇到 Cython 相对导入错误（如 `'vocab.pxd' not found`）。虽然可以通过手动修改 `.pxd` 文件将相对导入改为绝对导入（例如将 `from ..vocab cimport Vocab` 改为 `from spacy.vocab cimport Vocab`）来暂时解决，但官方已在 v0.100 版本中提供了针对 32 位和 64 位 Windows 的修复构建。建议直接升级到 v0.100 或更高版本，以避免手动修补源码。此外，可能还需要对依赖库 `cymem` 和 `preshed` 进行小幅修改才能顺利编译。","https:\u002F\u002Fgithub.com\u002Fexplosion\u002FspaCy\u002Fissues\u002F132",{"id":181,"question_zh":182,"answer_zh":183,"source_url":169},31590,"spaCy v2.0 相比 v1.x 有哪些主要新特性？","spaCy v2.0 引入了多项重大更新：\n1. **神经网络模型**：包含新的英语和多语言命名实体识别（NER）模型，并支持通过 Chainer\u002FCuPy 进行 GPU 加速。\n2. **字符串哈希**：字符串映射到哈希值而非整数 ID，确保跨模型匹配一致性。\n3. **序列化改进**：统一的保存\u002F加载 API，支持 Pickle。\n4. **可视化**：内置 displaCy 可视化工具，支持 Jupyter Notebook。\n5. **多语言支持**：增加了挪威语、日语、丹麦语和波兰语的 Tokenization 支持，以及多种语言的查找式词形还原（lemmatization）。\n6. **API 修订**：改进了 `Matcher` 和处理管道的 API。\n7. **文档向量**：支持可训练的文档向量和基于卷积神经网络的上下文相似度计算。",{"id":185,"question_zh":186,"answer_zh":187,"source_url":174},31591,"日语 NER（命名实体识别）数据集从哪里获取？","目前缺乏标准的日语 NER 数据集。常见的替代方案包括：\n1. 基于 Wikipedia 数据，利用类似 Nothman et al. 的方法自动生成训练数据（该方法曾用于其他 spaCy 模型，但因分词问题跳过了日语，需自行解决分词）。\n2. 利用现有的分词器字典（如 Unidic），这些字典通常包含类似实体的信息，允许用户轻松添加自定义条目，这是日语社区中较常用的方法。\n3. 参考 UD Japanese GSD 用于依存句法分析，但其不包含 NER 标注。",{"id":189,"question_zh":190,"answer_zh":191,"source_url":179},31592,"在 Windows 上编译 spaCy 时遇到 'operator=' 错误或链接问题怎么办？","这通常是旧版本 spaCy 在 Windows 环境下与编译器兼容性有关的问题。维护者已在 v0.100 版本中修复了针对 32 位和 64 位 Windows 的构建问题。如果遇到此类底层编译错误（特别是涉及 `ner.cpp` 或操作符重载的错误），最直接的解决方案是升级到已修复构建的版本（v0.100+），而不是尝试手动修复生成的 C++ 代码。同时确保依赖项 `cymem` 和 `preshed` 也是最新且兼容的版本。",[193,198,203,208,213,217,222,227,232,237,242,247,252,256,261,266,271,276,281,286],{"id":194,"version":195,"summary_zh":196,"released_at":197},238822,"release-v3.8.14","- 修复在 `pip` 不在 PATH 中但作为 Python 模块可用的环境中（例如某些虚拟环境和容器）`spacy download` 失败的问题。","2026-03-29T07:47:14",{"id":199,"version":200,"summary_zh":201,"released_at":202},238823,"release-v3.8.13","v3.8.12 版本未更新配置文件中的密钥，导致执行升级安装时模型无法加载。","2026-03-23T17:40:51",{"id":204,"version":205,"summary_zh":206,"released_at":207},238824,"release-v3.8.12","我们使用 confection v1.3 和 Thinc v8.3.13，它们通过自定义验证逻辑替代了 Pydantic，从而让我们能够顺利迁移到 Pydantic v2，并全面支持 Python 3.14。\n\n我们的依赖树以非传统方式使用了 Pydantic v1，并且依赖于 Pydantic v2 已经重构的行为。自 Pydantic v2 发布以来，曾有几次尝试迁移到新版本，但这一任务一直颇具挑战，原因在于 confection 库的实现相当复杂，而我在 2024 年和 2025 年用于开源工作的精力又较为有限。\n\n具体来说，confection 是我们在 spaCy 和 Thinc 中使用的可扩展配置系统。该配置系统允许引用由任意函数提供的值，例如用于定义神经网络模型或其子层的函数。confection 的功能之所以复杂，是因为我们在配置规范的设计上极度注重用户体验，即便这意味着实现层面的复杂度会相应增加。\n\nconfection 的原始实现为函数提供的动态值（“承诺”）构建了一个动态的 Pydantic v1 模式。我们会先对模式进行验证，然后再调用所有承诺并用其实际值替换占位符，最后再对更新后的模式做一次验证。变量插值机制进一步增加了实现难度，而且我们不得不基于 Python 内置的 configparser 进行子类化，这使得我们受限于一些当初的选择，如果现在从头开始，我会采取不同的做法。\n\n以下是对 Pydantic v1 特有行为的一个总结，这些行为使得迁移到 v2 对我们而言尤为困难。这份总结是在与 Claude Code Opus 4.6 的对话中生成的，因此其中可能存在不准确之处。实际上，围绕这一迁移所做的尝试经历了多次重构，每次间隔数月，所以我并没有完整记录下所有遇到的难题。因此，这份总结中的某些细节很可能并不准确。\n\n我们始终面临的核心问题在于：Pydantic v2 会在初始化时就编译好验证模式，并且对不可变性有着更为严格的要求。整个迁移过程实际上就是一系列针对这些问题的变通方案：\n\n``` \n1. 模式变更——v1 允许直接修改 __fields__；而 v2 则需要调用 model_rebuild()，但这会导致前向引用命名空间丢失，或者必须使用 create_model 子类，而这种方式又无法将更改传播到父级模式。\n2. model_dump 与 dict——v2 会将数据类转换为字典，从而破坏已解析的对象。为此我们不得不编写一个自定义的 _model_to_dict 辅助函数。\n3. model_construct 会丢弃额外字段——v2 会静默地丢弃那些设置了 extra=\"forbid\" 的字段，因此必须手动处理。\n4. 严格的类型转换——v2 会通过迭代将 ndarray 转换为 List[Floats1d]，这就要求启用 strict=True 参数。\n5. 前向引用——凡是引入了 TYPE_CHECKING 导入的模式，都需要在正确的命名空间下调用 model_rebuild()；然而，当 confection 后续再次重建模式时，这种处理方式又会失效。\n为了应对诸如此类的行为差异，我曾对 confection 进行重构，使其能够分别构建不同版本的模式，在多个…","2026-03-23T13:33:37",{"id":209,"version":210,"summary_zh":211,"released_at":212},238825,"release-v3.8.11","为 Windows ARM 平台添加 Python 3.11、3.12、3.13 和 3.14 的轮子包。由于 NumPy 尚未提供 Python 3.10 及更早版本的 Windows ARM 轮子包，因此本次不包含这些版本。","2025-11-17T20:37:38",{"id":214,"version":215,"summary_zh":76,"released_at":216},238826,"release-v3.8.10","2025-11-17T15:54:29",{"id":218,"version":219,"summary_zh":220,"released_at":221},238827,"release-v3.8.9","为 Python 3.14 添加轮子包","2025-11-13T14:57:22",{"id":223,"version":224,"summary_zh":225,"released_at":226},238828,"release-v3.8.8","* 修复来自 click 导入的弃用警告\n* 更新依赖项，包括切换到 `typer-slim` 以减少依赖体积\n* 停止对 Python 3.9 的支持（已停止维护）\n\nspaCy 依赖树中的其他依赖也已更新，放宽了 numpy 的版本兼容性约束，这应能减少部分用户的安装问题。","2025-11-07T09:26:30",{"id":228,"version":229,"summary_zh":230,"released_at":231},238829,"release-v3.8.7","为支持 Python 3.13，spaCy 现在使用 Cython 3 进行编译。这带来了运行时类型处理方式的变化：Cython 3 使用 `from __future__ import annotations` 语义，会在运行时将类型存储为字符串。这一差异导致在 Cython 文件中注册的组件出现问题，因为我们依赖于从工厂函数签名构建 Pydantic 模型来进行验证。\n\n为了支持 Python 3.13，我们因此创建了一个新模块 `spacy.pipeline.factories`，其中包含工厂函数的实现。同时，在这些函数之前的位置添加了 `__getattr__` 导入适配器，以避免向后不兼容的问题。\n\n除了迁移工厂函数之外，新的实现还通过将装饰器的实际调用移至一个函数内部，并在 `Language` 类初始化时仅执行一次，从而避免了导入时的副作用。\n\n目录注册装饰器也进行了相应的改动。我们创建了一个新的模块 `spacy.registrations`，用于执行所有的目录注册操作。将这些注册逻辑从函数中移出，可以防止这些装饰器在导入时被执行。虽然这一改动并非支持 Python 3.13 所必需，但它使我们不再依赖任何导入时的副作用，从而有望改善 spaCy 的导入时间，进而提升 CLI 的执行速度。此外，这一调整也简化了维护工作，因为不同注册函数的实现更容易被找到（这也有助于库的使用者）。","2025-05-23T08:53:14",{"id":233,"version":234,"summary_zh":235,"released_at":236},238830,"release-v3.8.6","恢复了对 ARM 平台滚轮的支持，并正确标注了兼容性范围。","2025-05-19T07:52:28",{"id":238,"version":239,"summary_zh":240,"released_at":241},238831,"release-v3.8.3","修复了在内存区内将非临时字符串添加到 StringStore 时出现的内存区域错误。该错误会导致形态分析器在内存区生效期间抛出“字符串未找到”异常。","2024-12-11T13:11:38",{"id":243,"version":244,"summary_zh":245,"released_at":246},238832,"release-v3.8.2","# 可选的持久化服务内存管理\n\n新增上下文管理器方法 `Language.memory_zone()`，允许长时间运行的服务避免因 `Vocab` 或 `StringStore` 中的缓存条目而导致内存占用不断增加。当内存区域块结束时，spaCy 会清除在该块期间添加的 `Vocab` 和 `StringStore` 条目，从而释放内存。在内存区域块内创建的 `Doc` 对象不应在块外访问。\n\n当前实现会在内存区域内禁用分词器缓存的填充，这会导致一定的性能影响。如果你运行的是完整管道，这种性能差异可能可以忽略；但如果你只运行分词器，则速度会明显变慢。如果遇到这个问题，可以通过预先“预热”缓存来缓解：先处理前几个批次的文本而不创建内存区域。未来更新中将为分词器添加对内存区域的支持。\n\n`Language.memory_zone()` 上下文管理器还会检查管道组件是否具有 `memory_zone()` 方法，以便组件在必要时执行类似的内存管理操作。目前，所有内置组件都不需要此功能。\n\n如果你的组件需要向 `StringStore` 或 `Vocab` 添加非临时条目，可以在调用 `Vocab.add()` 或 `StringStore.add()` 时传递 `allow_transient=False` 标志。\n\n使用示例：\n\n```python\nimport spacy\nimport json\nfrom pathlib import Path\nfrom typing import Iterator\nfrom collections import Counter\nimport typer\nfrom spacy.util import minibatch\n\n\ndef texts(path: Path) -> Iterator[str]:\n    with path.open(\"r\", encoding=\"utf8\") as file_:\n        for line in file_:\n            yield json.loads(line)[\"text\"]\n\ndef main(jsonl_path: Path) -> None:\n    nlp = spacy.load(\"en_core_web_sm\")\n    counts = Counter()\n    batches = minibatch(texts(jsonl_path), 1000)\n    for i, batch in enumerate(batches):\n        print(\"Batch\", i)\n        with nlp.memory_zone():\n            for doc in nlp.pipe(batch):\n                for token in doc:\n                    counts[token.text] += 1\n    for word, count in counts.most_common(100):\n        print(count, word)\n\nif __name__ == \"__main__\":\n    typer.run(main)\n```\n\n# Numpy v2 兼容性\n\nNumpy 2.0 与 Numpy v1 不兼容二进制格式，因此我们需要基于其中一个版本进行构建。本次发布将依赖项变更单独隔离，且没有其他更改，以便在依赖项变更导致问题时更易于处理。\n\n此前，这一依赖项变更已在 3.7.6 版本中尝试过，但由于 3.7 系列模型中的依赖关系存在冲突，且一些依赖 Numpy v1 的包与 3.7.6 不兼容，因此我取消了 3.7.6 版本，并用此次发布的、小版本号递增的版本取而代之。\n\n# 模型包不再将 spaCy 列为依赖项\n\n我还对模型打包方式进行了调整，以使发布过程更加简便。","2024-10-01T18:19:33",{"id":248,"version":249,"summary_zh":250,"released_at":251},238833,"prerelease-v3.8.0.dev0","Support a new context manager method `Language.memory_zone()`, to allow long-running services to avoid growing memory usage from cached entries in the `Vocab` or `StringStore`. Once the memory zone block ends, spaCy will evict `Vocab` and `StringStore` entries that were added during the block, freeing up memory. `Doc` objects created inside a memory zone block should not be accessed outside the block.\r\n\r\nThe current implementation disables population of the tokenizer cache inside the memory zone, resulting in some performance impact. The performance difference will likely be negligible if you're running a full pipeline, but if you're only running the tokenizer, it'll be much slower. If this is a problem, you can mitigate it by warming the cache first, by processing the first few batches of text without creating a memory zone. Support for memory zones in the tokenizer will be added in a future update.\r\n\r\nThe `Language.memory_zone()` context manager also checks for a `memory_zone()` method on pipeline components, so that components can perform similar memory management if necessary. None of the built-in components currently require this.\r\n\r\nIf you component needs to add non-transient entries to the `StringStore` or `Vocab`, you can pass the `allow_transient=False` flag to the `Vocab.add()` or `StringStore.add()` components.\r\n\r\nExample usage:\r\n\r\n```python\r\n\r\nimport spacy\r\nimport json\r\nfrom pathlib import Path\r\nfrom typing import Iterator\r\nfrom collections import Counter\r\nimport typer\r\nfrom spacy.util import minibatch\r\n\r\n\r\ndef texts(path: Path) -> Iterator[str]:\r\n    with path.open(\"r\", encoding=\"utf8\") as file_:\r\n        for line in file_:\r\n            yield json.loads(line)[\"text\"]\r\n\r\ndef main(jsonl_path: Path) -> None:\r\n    nlp = spacy.load(\"en_core_web_sm\")\r\n    counts = Counter()\r\n    batches = minibatch(texts(jsonl_path), 1000)\r\n    for i, batch in enumerate(batches):\r\n        print(\"Batch\", i)\r\n        with nlp.vocab.memory_zone():\r\n            for doc in nlp.pipe(batch):\r\n                for token in doc:\r\n                    counts[token.text] += 1\r\n    for word, count in counts.most_common(100):\r\n        print(count, word)\r\n\r\nif __name__ == \"__main__\":\r\n    typer.run(main)```","2024-09-09T14:19:15",{"id":253,"version":254,"summary_zh":76,"released_at":255},238834,"prerelease-v3.7.6a","2024-08-20T10:09:49",{"id":257,"version":258,"summary_zh":259,"released_at":260},238835,"v3.7.5","# ✨ New features and improvements\r\n\r\n* Sanitize direct download for `spacy download` (#13313).\r\n* Convert Cython properties to decorator syntax (#13390).\r\n* Bump Weasel pin to allow v0.4.x (#13409).\r\n* Improvements to the test suite (#13469, #13470).\r\n* Bump Typer pin to allow v0.10.0 and above (#13471).\r\n* Allow `typing-extensions\u003C5.0.0` for Python \u003C 3.8 (#13516).\r\n\r\n\r\n## 🔴 Bug fixes\r\n\r\n* #13400: Fix `use_gold_ents` behaviour for EntityLinker.\r\n\r\n## 📖 Documentation and examples\r\n\r\n* Make the file name for code listings stick to the top (#13379).\r\n* Update the documentation of `MorphAnalysis` (#13433).\r\n* Typo fixes in the documentation (#13466).\r\n\r\n## 👥 Contributors\r\n\r\n@danieldk, @honnibal, @ines, @JoeSchiff, @nokados, @Paillat-dev, @rmitsch, @schorfma, @strickvl, @svlandeg, @ynx0  ","2024-06-05T07:57:36",{"id":262,"version":263,"summary_zh":264,"released_at":265},238836,"v3.7.4","## ✨ New features and improvements\r\n\r\n* Improve NumPy 2.0 compatibility (#13103).\r\n* Added language extensions for Faroese and Norwegian Nynorsk (#13116).\r\n* Add new [`TextCatReduce.v1`](https:\u002F\u002Fspacy.io\u002Fapi\u002Farchitectures#TextCatReduce) layer for text classification (#13181).\r\n* Add new [`TextCatParametricAttention.v1 `](https:\u002F\u002Fspacy.io\u002Fapi\u002Farchitectures#TextCatParametricAttention) layer for text classification (#13201).\r\n* Use `build` module for creating model packages by default (#13109).\r\n* Add support for code loading to the [`benchmark speed`](https:\u002F\u002Fspacy.io\u002Fapi\u002Fcli#benchmark-speed) command (#13247).\r\n* Extend lexical attributes for English with more numericals (#13106).\r\n* Warn about reloading dependencies after downloading models (#13081).\r\n\r\n## 🔴 Bug fixes\r\n\r\n* #13259, #13304, #13321: Correctness fixes for multiprocessing support in [`Language.pipe`](https:\u002F\u002Fspacy.io\u002Fapi\u002Flanguage#pipe).\r\n* #13187: Typing and documentation fixes for [`Doc`](https:\u002F\u002Fspacy.io\u002Fapi\u002Fdoc).\r\n* #13086: Update [`Tokenizer.explain`](https:\u002F\u002Fspacy.io\u002Fapi\u002Ftokenizer#explain) for special cases with whitespace.\r\n* #13068: Fix displaCy span stacking.\r\n* #13149: Add [spacy.TextCatBOW.v3](https:\u002F\u002Fspacy.io\u002Fapi\u002Farchitectures#TextCatBOW) to use the fixed [`SparseLinear`](https:\u002F\u002Fthinc.ai\u002Fdocs\u002Fapi-layers#sparselinear_v2) layer.\r\n\r\n## 📖 Documentation and examples\r\n\r\n* Many improvements and updates to the [LLM documentation](https:\u002F\u002Fspacy.io\u002Fusage\u002Flarge-language-models).\r\n* Update `trf_data` examples and the [transformer pipeline design](https:\u002F\u002Fspacy.io\u002Fmodels#design-trf) section.\r\n\r\n## 👥 Contributors\r\n\r\n@adrianeboyd, @danieldk, @evornov, @honnibal, @ines, @lise-brinck, @ridge-kimani, @rmitsch, @shadeMe, @svlandeg ","2024-02-15T19:16:59",{"id":267,"version":268,"summary_zh":269,"released_at":270},238837,"v3.7.2","## ✨ New features and improvements\r\n\r\n- Update `__all__` fields (#13063).\r\n\r\n## 🔴 Bug fixes\r\n\r\n- #13035: Remove Pathy requirement.\r\n- #13053: Restore `spacy.cli.project` API.\r\n- #13057: Support `Any` comparisons for `Token` and `Span`.\r\n\r\n## 📖 Documentation and examples\r\n\r\n- Many updates for `spacy-llm` including Azure OpenAI, PaLM, and Mistral support.\r\n- Various documentation corrections.\r\n\r\n## 👥 Contributors\r\n\r\n@adrianeboyd, @honnibal, @ines, @rmitsch, @svlandeg","2023-10-16T16:11:22",{"id":272,"version":273,"summary_zh":274,"released_at":275},238838,"v3.7.1","## 🔴 Bug fixes\r\n\r\n- Revert lazy loading of CLI module for `spacy.info` to fix availability of `spacy.cli` following `import spacy` (#13040).\r\n\r\n## 👥 Contributors\r\n\r\n@adrianeboyd, @honnibal, @ines, @svlandeg","2023-10-05T06:46:15",{"id":277,"version":278,"summary_zh":279,"released_at":280},238839,"v3.7.0","This release drops support for Python 3.6 and adds support for Python 3.12.\r\n\r\n## ✨ New features and improvements\r\n\r\n- Add support for Python 3.12 (#12979).\r\n- Use the new library [Weasel](https:\u002F\u002Fgithub.com\u002Fexplosion\u002Fweasel) for spaCy projects functionality (#12769).\r\n    - All `spacy project` commands should run as before, just now they're using Weasel under the hood.\r\n    - ⚠️ Remote storage  is not yet supported for Python 3.12. Use Python 3.11 or earlier for remote storage.\r\n- Extend to Thinc v8.2 (#12897).\r\n- Extend `transformers` extra to `spacy-transformers` v1.3 (#13025).\r\n- Support [registered vectors](https:\u002F\u002Fspacy.io\u002Fusage\u002Fembeddings-transformers#custom-vectors) (#12492).\r\n- Add `--spans-key` option for CLI evaluation with `spacy benchmark accuracy` (#12981).\r\n- Load the CLI module lazily for `spacy.info` (#12962).\r\n- Add type stubs for `spacy.training.example` (#12801).\r\n- Warn for unsupported pattern keys in dependency matcher (#12928).\r\n- `Language.replace_listeners`: Pass the replaced listener and the `tok2vec` pipe to the callback in order to support `spacy-curated-transformers` (#12785).\r\n- Always use `tqdm` with `disable=None` to disable output in non-interactive environments (#12979).\r\n- Language updates:\r\n    - Add left and right pointing angle brackets as punctuation to ancient Greek (#12829).\r\n    - Update example sentences for Turkish (#12895).\r\n- Package setup updates:\r\n    - Update NumPy build constraints for NumPy 1.25+ (#12839). For Python 3.9+, it is no longer necessary to set build constraints while building binary wheels.\r\n    - Refactor Cython profiling in order to disable profiling for Python 3.12 in the package setup, since Cython does not currently support profiling for Python 3.12 (#12979).\r\n\r\n## 📦 Trained pipelines updates\r\n\r\nThe transformer-based `trf` pipelines have been updated to use our new [Curated Transformers](https:\u002F\u002Fgithub.com\u002Fexplosion\u002Fcurated-transformers) library through the Thinc model wrappers and pipeline component from [spaCy Curated Transformers](https:\u002F\u002Fgithub.com\u002Fexplosion\u002Fspacy-curated-transformers).\r\n\r\n## ⚠️ Backwards incompatibilities\r\n\r\n- Drop support for Python 3.6.\r\n- Drop mypy checks for Python 3.7.\r\n- Remove `ray` extra.\r\n- `spacy project` has a few backwards incompatibilities due to the transition to the standalone library [Weasel](https:\u002F\u002Fgithub.com\u002Fexplosion\u002Fweasel), which is not as tightly coupled to spaCy. Weasel produces warnings when it detects older spaCy-specific settings in your environment or project config.\r\n    - Support for the `spacy_version` configuration key has been dropped.\r\n    - Support for the `check_requirements` configuration key has been dropped due to the deprecation of `pkg_resources`.\r\n    - The `SPACY_CONFIG_OVERRIDES` environment variable is no longer checked. You can set configuration overrides using `WEASEL_CONFIG_OVERRIDES`.\r\n    - Support for `SPACY_PROJECT_USE_GIT_VERSION` environment variable has been dropped.\r\n    - Error codes are now Weasel-specific and do not follow spaCy error codes.\r\n\r\n## 📖 Documentation and examples\r\n\r\n- New and updated documentation for [large language models](https:\u002F\u002Fspacy.io\u002Fusage\u002Flarge-language-models) and [spaCy Curated Transformers](https:\u002F\u002Fspacy.io\u002Fapi\u002Fcuratedtransformer).\r\n- Various documentation corrections and updates.\r\n- New additions to the [spaCy Universe](https:\u002F\u002Fspacy.io\u002Funiverse):\r\n    - [Hobbit spaCy](https:\u002F\u002Fspacy.io\u002Funiverse\u002Fproject\u002Fhobbit-spacy): NLP for Middle Earth\r\n    - [rolegal](https:\u002F\u002Fspacy.io\u002Funiverse\u002Fproject\u002Frolegal): a spaCy Package for Noisy Romanian Legal Document Processing\r\n\r\n## 👥 Contributors\r\n\r\n@adrianeboyd, @bdura, @connorbrinton, @danieldk, @davidberenstein1957, @denizcodeyaa, @eltociear, @evornov, @honnibal, @ines, @jmyerston, @koaning, @magdaaniol, @pdhall99, @ringohoffman, @rmitsch, @senisioi, @shadeMe, @svlandeg, @vinbo8, @wjbmattingly","2023-10-02T09:10:08",{"id":282,"version":283,"summary_zh":284,"released_at":285},238840,"v3.6.1","## ✨ New features and improvements\r\n\r\n- Allow Pydantic v2 using transitional v1 support (#12888).\r\n- Add `find-function` CLI for finding locations of registered functions (#12757).\r\n- Add extra `spacy[cuda12x]` for `cupy-cuda12x` (#12890).\r\n- Extend tests for `init config` and `train` CLI (#12173).\r\n- Switch from `distutils` to `setuptools`\u002F`sysconfig` (#12853).\r\n\r\n## 🔴 Bug fixes\r\n\r\n- #12817: Escape annotated HTML tags in displaCy span renderer.\r\n- #12857: Display model's full base version string in incompatibility warning.\r\n- #12882: Update `\u003Cbr>` tags in displaCy.\r\n\r\n## 📖 Documentation and examples\r\n\r\n- Various documentation corrections and updates.\r\n- New additions to spaCy Universe:\r\n    - [OdyCy](https:\u002F\u002Fspacy.io\u002Funiverse\u002Fproject\u002Fodycy)\r\n    - [SaysWho](https:\u002F\u002Fspacy.io\u002Funiverse\u002Fproject\u002Fsayswho)\r\n\r\n## 👥 Contributors\r\n\r\n@adrianeboyd, @afriedman412, @arplusman, @bdura, @connorbrinton, @honnibal, @ines, @it176131, @pmbaumgartner, @rmitsch, @shadeMe, @svlandeg, @thomashacker, @victorialslocum, @x-tabdeveloping\r\n","2023-08-08T14:45:58",{"id":287,"version":288,"summary_zh":289,"released_at":290},238841,"v3.6.0","## ✨ New features and improvements\r\n\r\n- **NEW**: [`span_finder` pipeline component](https:\u002F\u002Fspacy.io\u002Fapi\u002Fspanfinder) to identify overlapping, unlabeled spans (#12507).\r\n- Language updates:\r\n    - Add initial support for Malay (#12602).\r\n    - Update Latin defaults to support noun chunks, update lexical\u002Ftokenizer defaults and add example sentences (#12538).\r\n- Add option to return scores separately keyed by component name with `spacy evaluate --per-component`, `Language.evaluate(per_component=True)` and `Scorer.score(per_component=True)` (#12540).\r\n- Support custom token\u002Flexeme attribute for vectors (#12625).\r\n- Support `spancat_singlelabel` in `spacy debug data` CLI (#12749).\r\n- Typing updates for `PhraseMatcher` and `SpanGroup` (#12642, #12714).\r\n\r\n## 🔴 Bug fixes\r\n\r\n- #12569: Require that all `SpanGroup` spans come from the current doc.\r\n\r\n## 📦 Trained pipelines updates\r\n\r\nWe have added new pipelines for Slovenian that use the trainable lemmatizer and floret vectors.\r\n\r\n| Package | UPOS | Parser LAS | NER F |\r\n| --- | --- | --- | --- |\r\n| [`sl_core_news_sm`](https:\u002F\u002Fspacy.io\u002Fmodels\u002Fsl#sl_core_news_sm) | 96.9 | 82.1 | 62.9 |\r\n| [`sl_core_news_md`](https:\u002F\u002Fspacy.io\u002Fmodels\u002Fsl#sl_core_news_md) | 97.6 | 84.3 | 73.5 |\r\n| [`sl_core_news_lg`](https:\u002F\u002Fspacy.io\u002Fmodels\u002Fsl#sl_core_news_lg) | 97.7 | 84.3 | 79.0 |\r\n| [`sl_core_news_trf`](https:\u002F\u002Fspacy.io\u002Fmodels\u002Fsl#sl_core_news_trf) | 99.0 | 91.7 | 90.0 |\r\n\r\n- 🙏 **Special thanks** to @orglce for help with the new pipelines!\r\n\r\nThe English pipelines have been updated to improve handling of contractions with various apostrophes and to lemmatize \"get\" as a passive auxiliary.\r\n\r\nThe Danish pipeline `da_core_news_trf` has been updated to use [`vesteinn\u002FDanskBERT`](https:\u002F\u002Fhuggingface.co\u002Fvesteinn\u002FDanskBERT) with performance improvements across the board.\r\n\r\n## ⚠️ Backwards incompatibilities\r\n\r\n- `SpanGroup` spans are now required to be from the same doc. When initializing a `SpanGroup`, there is a new check to verify that all added spans refer to the current doc. Without this check, it was possible to run into string store or other errors.\r\n\r\n## 📖 Documentation and examples\r\n\r\n- Various documentation corrections and updates.\r\n- New additions to spaCy Universe:\r\n    - [spaCy-SetFit](https:\u002F\u002Fspacy.io\u002Funiverse\u002Fproject\u002Fspacysetfit)\r\n    - [spacy-vscode](https:\u002F\u002Fspacy.io\u002Funiverse\u002Fproject\u002Fspacy-vscode)\r\n    - [SpanMarker](https:\u002F\u002Fspacy.io\u002Funiverse\u002Fproject\u002Fspan_marker)\r\n    - [vetiver](https:\u002F\u002Fspacy.io\u002Funiverse\u002Fproject\u002Fvetiver)\r\n\r\n## 👥 Contributors\r\n\r\n@adrianeboyd, @bdura, @danieldk, @davidberenstein1957, @diyclassics, @essenmitsosse, @honnibal, @ines, @isabelizimm, @jmyerston, @kadarakos, @KennethEnevoldsen, @khursani8, @ljvmiranda921, @rmitsch, @shadeMe, @svlandeg, @tomaarsen, @victorialslocum, @vin-ivar, @ZiadAmerr","2023-07-07T08:32:23"]