[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-gorse-io--gorse":3,"tool-gorse-io--gorse":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",150037,2,"2026-04-10T23:33:47",[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":76,"owner_twitter":77,"owner_website":78,"owner_url":79,"languages":80,"stars":115,"forks":116,"last_commit_at":117,"license":118,"difficulty_score":10,"env_os":119,"env_gpu":120,"env_ram":121,"env_deps":122,"category_tags":129,"github_topics":130,"view_count":32,"oss_zip_url":76,"oss_zip_packed_at":76,"status":17,"created_at":136,"updated_at":137,"faqs":138,"releases":139},6441,"gorse-io\u002Fgorse","gorse","AI powered open source recommender system engine supports classical\u002FLLM rankers and multimodal content via embedding","Gorse 是一款基于 Go 语言开发的高性能开源推荐系统引擎，旨在帮助各类在线服务快速构建个性化的内容推荐功能。它解决了传统推荐系统搭建门槛高、定制复杂以及难以融合多模态数据等痛点，让用户只需导入物品、用户及交互数据，系统即可自动训练模型并生成精准推荐。\n\n这款工具特别适合后端开发者、数据工程师以及需要为应用添加推荐功能的初创团队使用。无论是电商商品、新闻资讯还是像示例中展示的 GitHub 仓库，Gorse 都能轻松应对。其独特亮点在于强大的兼容性与前瞻性：不仅支持协同过滤等经典推荐算法，还率先整合了基于大语言模型（LLM）的排序器；同时具备多模态处理能力，可通过嵌入技术理解文本、图片和视频内容。此外，Gorse 提供了直观的图形化仪表盘用于监控和管理推荐流程，并暴露标准的 RESTful API 方便系统集成。通过单节点训练与分布式预测的架构设计，Gorse 在保证易用性的同时也兼顾了扩展性，是构建现代化推荐服务的理想选择。","# Gorse Open-source Recommender System Engine\n\n\u003Cimg width=160 src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgorse-io_gorse_readme_cb870a69d331.png\"\u002F>\n\n![](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fgo-mod\u002Fgo-version\u002Fzhenghaoz\u002Fgorse)\n[![test](https:\u002F\u002Fgithub.com\u002Fgorse-io\u002Fgorse\u002Factions\u002Fworkflows\u002Fbuild_test.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fgorse-io\u002Fgorse\u002Factions\u002Fworkflows\u002Fbuild_test.yml)\n[![codecov](https:\u002F\u002Fcodecov.io\u002Fgh\u002Fgorse-io\u002Fgorse\u002Fbranch\u002Fmaster\u002Fgraph\u002Fbadge.svg)](https:\u002F\u002Fcodecov.io\u002Fgh\u002Fgorse-io\u002Fgorse)\n[![Discord](https:\u002F\u002Fimg.shields.io\u002Fdiscord\u002F830635934210588743)](https:\u002F\u002Fdiscord.gg\u002Fx6gAtNNkAE)\n[![Twitter Follow](https:\u002F\u002Fimg.shields.io\u002Ftwitter\u002Ffollow\u002Fgorse_io?label=Follow&style=social)](https:\u002F\u002Ftwitter.com\u002Fgorse_io)\n[![Gurubase](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FGurubase-Ask%20Gorse%20Guru-006BFF)](https:\u002F\u002Fgurubase.io\u002Fg\u002Fgorse)\n\nGorse is an AI powered open-source recommender system written in Go. Gorse aims to be a universal open-source recommender system that can be quickly integrated into a wide variety of online services. By importing items, users, and interaction data into Gorse, the system will automatically train models to generate recommendations for each user. Project features are as follows.\n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgorse-io_gorse_readme_3ea49cd526b9.png)\n\n- **Multi-source:** Recommend items from latest, user-to-user, item-to-item, collaborative filtering and etc.\n- **Multimodal:** Support multimodal content (text, image, videos, etc.) via embedding.\n- **AI-powered:** Support both classical recommenders and LLM-based recommenders.\n- **GUI Dashboard:** Provide GUI dashboard for recommendation pipeline editing, system monitoring, and data management.\n- **RESTful APIs:** Expose RESTful APIs for data CRUD and recommendation requests.\n\n## Quick Start\n\nThe playground mode has been prepared for beginners. Just set up a recommender system for GitHub repositories by the following commands.\n\n```bash\ndocker run -p 8088:8088 zhenghaoz\u002Fgorse-in-one --playground\n```\n\nThe playground mode will download data from [GitRec](https:\u002F\u002Fgitrec.gorse.io\u002F) and import it into Gorse. The dashboard is available at `http:\u002F\u002Flocalhost:8088`.\n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgorse-io_gorse_readme_1d96b9406b43.png)\n\nAfter the \"Generate item-to-item recommendation\" task is completed on the \"Tasks\" page, try to insert several feedbacks into Gorse. Suppose Bob is a developer who interested in LLM related repositories. We insert his star feedback to Gorse.\n\n```bash\nread -d '' JSON \u003C\u003C EOF\n[\n    { \\\"FeedbackType\\\": \\\"star\\\", \\\"UserId\\\": \\\"bob\\\", \\\"ItemId\\\": \\\"ollama:ollama\\\", \\\"Value\\\": 1.0, \\\"Timestamp\\\": \\\"2022-02-24\\\" },\n    { \\\"FeedbackType\\\": \\\"star\\\", \\\"UserId\\\": \\\"bob\\\", \\\"ItemId\\\": \\\"huggingface:transformers\\\", \\\"Value\\\": 1.0, \\\"Timestamp\\\": \\\"2022-02-25\\\" },\n    { \\\"FeedbackType\\\": \\\"star\\\", \\\"UserId\\\": \\\"bob\\\", \\\"ItemId\\\": \\\"rasbt:llms-from-scratch\\\", \\\"Value\\\": 1.0, \\\"Timestamp\\\": \\\"2022-02-26\\\" },\n    { \\\"FeedbackType\\\": \\\"star\\\", \\\"UserId\\\": \\\"bob\\\", \\\"ItemId\\\": \\\"vllm-project:vllm\\\", \\\"Value\\\": 1.0, \\\"Timestamp\\\": \\\"2022-02-27\\\" },\n    { \\\"FeedbackType\\\": \\\"star\\\", \\\"UserId\\\": \\\"bob\\\", \\\"ItemId\\\": \\\"hiyouga:llama-factory\\\", \\\"Value\\\": 1.0, \\\"Timestamp\\\": \\\"2022-02-28\\\" }\n]\nEOF\n\ncurl -X POST http:\u002F\u002F127.0.0.1:8088\u002Fapi\u002Ffeedback \\\n   -H 'Content-Type: application\u002Fjson' \\\n   -d \"$JSON\"\n```\n\nThen, fetch 10 recommended items from Gorse. We can find that LLM-related repositories are recommended for Bob.\n\n```bash\ncurl http:\u002F\u002F127.0.0.1:8088\u002Fapi\u002Frecommend\u002Fbob?n=10\n```\n\nFor more information：\n\n- Read [official documents](https:\u002F\u002Fgorse.io\u002Fdocs\u002F)\n- Visit [playground](https:\u002F\u002Fplay.gorse.io\u002F) of Gorse dashboard\n- Explore [live demo](https:\u002F\u002Fgitrec.gorse.io\u002F), a recommender system for GitHub repositories\n- Discuss on [Discord](https:\u002F\u002Fdiscord.gg\u002Fx6gAtNNkAE) or [GitHub Discussion](https:\u002F\u002Fgithub.com\u002Fgorse-io\u002Fgorse\u002Fdiscussions)\n\n## Architecture\n\nGorse is a single-node training and distributed prediction recommender system. Gorse stores data in MySQL, MongoDB, Postgres, or ClickHouse, with intermediate results cached in Redis, MySQL, MongoDB and Postgres.\n\n1. The cluster consists of a master node, multiple worker nodes, and server nodes.\n1. The master node is responsible for model training, non-personalized recommendation, configuration management, and membership management.\n1. The server node is responsible for exposing the RESTful APIs and online real-time recommendations.\n1. Worker nodes are responsible for offline recommendations for each user.\n\nIn addition, the administrator can perform system monitoring, data import and export, and system status checking via the dashboard on the master node.\n\n\u003Cimg width=520 src=\"https:\u002F\u002Fgithub.com\u002Fgorse-io\u002Fdocs\u002Fblob\u002Fmain\u002Fsrc\u002Fimg\u002Fcluster.drawio.svg?raw=true\"\u002F>\n\n## Contributors\n\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fgorse-io\u002Fgorse\u002Fgraphs\u002Fcontributors\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgorse-io_gorse_readme_1eda042c47c8.png\" \u002F>\n\u003C\u002Fa>\n\nAny contribution is appreciated: report a bug, give advice or create a pull request. Read [CONTRIBUTING.md](CONTRIBUTING.md) for more information.\n\n## Acknowledgments\n\n`gorse` is inspired by the following projects:\n\n- [Guibing Guo's librec](https:\u002F\u002Fgithub.com\u002Fguoguibing\u002Flibrec)\n- [Nicolas Hug's Surprise](https:\u002F\u002Fgithub.com\u002FNicolasHug\u002FSurprise)\n- [Golang Samples's gopher-vector](https:\u002F\u002Fgithub.com\u002Fgolang-samples\u002Fgopher-vector)\n","# Gorse 开源推荐系统引擎\n\n\u003Cimg width=160 src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgorse-io_gorse_readme_cb870a69d331.png\"\u002F>\n\n![](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fgo-mod\u002Fgo-version\u002Fzhenghaoz\u002Fgorse)\n[![test](https:\u002F\u002Fgithub.com\u002Fgorse-io\u002Fgorse\u002Factions\u002Fworkflows\u002Fbuild_test.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fgorse-io\u002Fgorse\u002Factions\u002Fworkflows\u002Fbuild_test.yml)\n[![codecov](https:\u002F\u002Fcodecov.io\u002Fgh\u002Fgorse-io\u002Fgorse\u002Fbranch\u002Fmaster\u002Fgraph\u002Fbadge.svg)](https:\u002F\u002Fcodecov.io\u002Fgh\u002Fgorse-io\u002Fgorse)\n[![Discord](https:\u002F\u002Fimg.shields.io\u002Fdiscord\u002F830635934210588743)](https:\u002F\u002Fdiscord.gg\u002Fx6gAtNNkAE)\n[![Twitter Follow](https:\u002F\u002Fimg.shields.io\u002Ftwitter\u002Ffollow\u002Fgorse_io?label=Follow&style=social)](https:\u002F\u002Ftwitter.com\u002Fgorse_io)\n[![Gurubase](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FGurubase-Ask%20Gorse%20Guru-006BFF)](https:\u002F\u002Fgurubase.io\u002Fg\u002Fgorse)\n\nGorse 是一个基于 Go 语言编写的、由 AI 驱动的开源推荐系统。Gorse 的目标是成为一个通用的开源推荐系统，能够快速集成到各种在线服务中。通过将物品、用户和交互数据导入 Gorse，系统会自动训练模型，为每个用户生成个性化推荐。项目的主要特性如下。\n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgorse-io_gorse_readme_3ea49cd526b9.png)\n\n- **多源推荐：** 支持从最新内容、用户间、物品间以及协同过滤等多种来源进行推荐。\n- **多模态支持：** 通过嵌入技术支持多模态内容（文本、图片、视频等）。\n- **AI 驱动：** 同时支持传统推荐算法和基于大语言模型的推荐算法。\n- **GUI 控制台：** 提供图形化界面，用于推荐流程编辑、系统监控和数据管理。\n- **RESTful API：** 暴露 RESTful API，用于数据的增删改查及推荐请求。\n\n## 快速开始\n\n我们为初学者准备了试用模式。只需运行以下命令即可为 GitHub 仓库搭建一个推荐系统。\n\n```bash\ndocker run -p 8088:8088 zhenghaoz\u002Fgorse-in-one --playground\n```\n\n试用模式会从 [GitRec](https:\u002F\u002Fgitrec.gorse.io\u002F) 下载数据并导入到 Gorse 中。控制台地址为 `http:\u002F\u002Flocalhost:8088`。\n\n![](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgorse-io_gorse_readme_1d96b9406b43.png)\n\n在“任务”页面完成“生成物品间推荐”任务后，尝试向 Gorse 插入几条反馈。假设 Bob 是一位对大语言模型相关仓库感兴趣的开发者，我们可以将他的点赞反馈插入到 Gorse 中。\n\n```bash\nread -d '' JSON \u003C\u003C EOF\n[\n    { \\\"FeedbackType\\\": \\\"star\\\", \\\"UserId\\\": \\\"bob\\\", \\\"ItemId\\\": \\\"ollama:ollama\\\", \\\"Value\\\": 1.0, \\\"Timestamp\\\": \\\"2022-02-24\\\" },\n    { \\\"FeedbackType\\\": \\\"star\\\", \\\"UserId\\\": \\\"bob\\\", \\\"ItemId\\\": \\\"huggingface:transformers\\\", \\\"Value\\\": 1.0, \\\"Timestamp\\\": \\\"2022-02-25\\\" },\n    { \\\"FeedbackType\\\": \\\"star\\\", \\\"UserId\\\": \\\"bob\\\", \\\"ItemId\\\": \\\"rasbt:llms-from-scratch\\\", \\\"Value\\\": 1.0, \\\"Timestamp\\\": \\\"2022-02-26\\\" },\n    { \\\"FeedbackType\\\": \\\"star\\\", \\\"UserId\\\": \\\"bob\\\", \\\"ItemId\\\": \\\"vllm-project:vllm\\\", \\\"Value\\\": 1.0, \\\"Timestamp\\\": \\\"2022-02-27\\\" },\n    { \\\"FeedbackType\\\": \\\"star\\\", \\\"UserId\\\": \\\"bob\\\", \\\"ItemId\\\": \\\"hiyouga:llama-factory\\\", \\\"Value\\\": 1.0, \\\"Timestamp\\\": \\\"2022-02-28\\\" }\n]\nEOF\n\ncurl -X POST http:\u002F\u002F127.0.0.1:8088\u002Fapi\u002Ffeedback \\\n   -H 'Content-Type: application\u002Fjson' \\\n   -d \"$JSON\"\n```\n\n然后，从 Gorse 获取 10 条推荐结果。你会发现，Bob 获得的推荐主要都是与大语言模型相关的仓库。\n\n```bash\ncurl http:\u002F\u002F127.0.0.1:8088\u002Fapi\u002Frecommend\u002Fbob?n=10\n```\n\n更多信息：\n\n- 阅读 [官方文档](https:\u002F\u002Fgorse.io\u002Fdocs\u002F)\n- 访问 Gorse 控制台的 [试用版](https:\u002F\u002Fplay.gorse.io\u002F)\n- 探索 [实时演示](https:\u002F\u002Fgitrec.gorse.io\u002F)——一个针对 GitHub 仓库的推荐系统\n- 在 [Discord](https:\u002F\u002Fdiscord.gg\u002Fx6gAtNNkAE) 或 [GitHub 讨论区](https:\u002F\u002Fgithub.com\u002Fgorse-io\u002Fgorse\u002Fdiscussions) 中交流\n\n## 架构\n\nGorse 是一个单节点训练、分布式预测的推荐系统。它使用 MySQL、MongoDB、Postgres 或 ClickHouse 存储数据，并将中间结果缓存在 Redis、MySQL、MongoDB 和 Postgres 中。\n\n1. 集群由一个主节点、多个工作节点和服务器节点组成。\n1. 主节点负责模型训练、非个性化推荐、配置管理和用户管理。\n1. 服务器节点负责暴露 RESTful API 和在线实时推荐。\n1. 工作节点负责为每个用户生成离线推荐。\n\n此外，管理员可以通过主节点上的控制台进行系统监控、数据导入导出以及系统状态检查。\n\n\u003Cimg width=520 src=\"https:\u002F\u002Fgithub.com\u002Fgorse-io\u002Fdocs\u002Fblob\u002Fmain\u002Fsrc\u002Fimg\u002Fcluster.drawio.svg?raw=true\"\u002F>\n\n## 贡献者\n\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fgorse-io\u002Fgorse\u002Fgraphs\u002Fcontributors\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgorse-io_gorse_readme_1eda042c47c8.png\" \u002F>\n\u003C\u002Fa>\n\n无论您是报告 bug、提出建议，还是创建 pull request，您的贡献都将受到欢迎。更多详情请参阅 [CONTRIBUTING.md](CONTRIBUTING.md)。\n\n## 致谢\n\n`gorse` 的灵感来源于以下项目：\n\n- [Guibing Guo 的 librec](https:\u002F\u002Fgithub.com\u002Fguoguibing\u002Flibrec)\n- [Nicolas Hug 的 Surprise](https:\u002F\u002Fgithub.com\u002FNicolasHug\u002FSurprise)\n- [Golang Samples 的 gopher-vector](https:\u002F\u002Fgithub.com\u002Fgolang-samples\u002Fgopher-vector)","# Gorse 开源推荐系统快速上手指南\n\nGorse 是一个基于 Go 语言开发的高性能开源推荐系统引擎。它支持多源数据、多模态内容（文本、图像等）以及传统算法与大模型（LLM）结合的推荐策略，并提供可视化的管理仪表盘和 RESTful API。\n\n## 环境准备\n\n*   **操作系统**：Linux, macOS, Windows (WSL 推荐)\n*   **核心依赖**：\n    *   **Docker**：推荐使用 Docker 进行一键部署（本指南采用此方式）。\n    *   若需源码编译，需安装 Go 1.20+。\n*   **网络要求**：初次运行 Playground 模式时，系统会自动从互联网下载示例数据集（GitRec），请确保网络畅通。国内用户若遇到拉取镜像或数据缓慢，建议配置 Docker 镜像加速器。\n\n## 安装步骤\n\nGorse 提供了 `in-one` 镜像，集成了所有组件，最适合快速体验。只需执行以下命令即可启动包含示例数据的游乐场模式（Playground Mode）：\n\n```bash\ndocker run -p 8088:8088 zhenghaoz\u002Fgorse-in-one --playground\n```\n\n*   该命令会自动下载并导入 GitHub 仓库推荐数据集。\n*   启动完成后，访问 `http:\u002F\u002Flocalhost:8088` 即可打开可视化仪表盘。\n\n## 基本使用\n\n以下演示如何通过 API 插入用户行为数据并获取个性化推荐结果。\n\n### 1. 等待任务完成\n在仪表盘的 **Tasks** 页面，等待 \"Generate item-to-item recommendation\" 任务状态变为完成。\n\n### 2. 插入用户反馈\n假设用户 \"bob\" 对 LLM 相关的仓库感兴趣，我们通过 API 插入他的点赞（star）记录。执行以下命令：\n\n```bash\nread -d '' JSON \u003C\u003C EOF\n[\n    { \\\"FeedbackType\\\": \\\"star\\\", \\\"UserId\\\": \\\"bob\\\", \\\"ItemId\\\": \\\"ollama:ollama\\\", \\\"Value\\\": 1.0, \\\"Timestamp\\\": \\\"2022-02-24\\\" },\n    { \\\"FeedbackType\\\": \\\"star\\\", \\\"UserId\\\": \\\"bob\\\", \\\"ItemId\\\": \\\"huggingface:transformers\\\", \\\"Value\\\": 1.0, \\\"Timestamp\\\": \\\"2022-02-25\\\" },\n    { \\\"FeedbackType\\\": \\\"star\\\", \\\"UserId\\\": \\\"bob\\\", \\\"ItemId\\\": \\\"rasbt:llms-from-scratch\\\", \\\"Value\\\": 1.0, \\\"Timestamp\\\": \\\"2022-02-26\\\" },\n    { \\\"FeedbackType\\\": \\\"star\\\", \\\"UserId\\\": \\\"bob\\\", \\\"ItemId\\\": \\\"vllm-project:vllm\\\", \\\"Value\\\": 1.0, \\\"Timestamp\\\": \\\"2022-02-27\\\" },\n    { \\\"FeedbackType\\\": \\\"star\\\", \\\"UserId\\\": \\\"bob\\\", \\\"ItemId\\\": \\\"hiyouga:llama-factory\\\", \\\"Value\\\": 1.0, \\\"Timestamp\\\": \\\"2022-02-28\\\" }\n]\nEOF\n\ncurl -X POST http:\u002F\u002F127.0.0.1:8088\u002Fapi\u002Ffeedback \\\n   -H 'Content-Type: application\u002Fjson' \\\n   -d \"$JSON\"\n```\n\n### 3. 获取推荐结果\n请求 Gorse 为 \"bob\" 生成 10 个推荐项。系统将基于他刚才的反馈，返回相关的 LLM 仓库列表：\n\n```bash\ncurl http:\u002F\u002F127.0.0.1:8088\u002Fapi\u002Frecommend\u002Fbob?n=10\n```\n\n---\n**更多资源：**\n*   [官方文档](https:\u002F\u002Fgorse.io\u002Fdocs\u002F)\n*   [在线演示 (GitRec)](https:\u002F\u002Fgitrec.gorse.io\u002F)\n*   [Discord 社区](https:\u002F\u002Fdiscord.gg\u002Fx6gAtNNkAE)","某中型技术博客平台希望提升用户阅读留存率，计划为每位开发者定制个性化的文章推荐流。\n\n### 没有 gorse 时\n- **冷启动困难**：新发布的优质技术文章因缺乏历史点击数据，很难被目标读者发现，长期沉底。\n- **开发成本高昂**：团队需从零搭建推荐算法模型，耗费数月时间处理数据清洗、特征工程和模型训练，且难以支持多模态内容（如文章封面图、代码块）。\n- **策略单一僵化**：仅能基于简单的“热门文章”或“同类标签”进行粗糙推荐，无法实现精准的“协同过滤”或“看了又看”功能。\n- **运维监控缺失**：缺乏可视化界面监控推荐效果，调整策略需修改代码并重新部署，响应市场变化极慢。\n\n### 使用 gorse 后\n- **即时冷启动破解**：利用 gorse 的多模态嵌入能力，自动提取新文章的文本和图像特征，即使无交互数据也能立即推荐给兴趣匹配的用户。\n- **开箱即用加速上线**：通过 Docker 一键部署，导入用户行为和文章内容后，系统自动训练经典算法与 LLM 排序模型，将研发周期从数月缩短至几天。\n- **混合策略精准触达**：灵活组合“最新”、“协同过滤”及“物品关联”等多种推荐源，成功让喜欢 LLM 教程的用户精准获取相关前沿资讯。\n- **可视化敏捷运营**：借助 GUI 仪表盘实时监控推荐流水线状态，运营人员可动态调整参数而无需重启服务，大幅提升迭代效率。\n\ngorse 将复杂的推荐系统工程化为标准化的基础设施，让中小型团队也能以极低门槛拥有大厂级别的个性化推荐能力。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fgorse-io_gorse_1d96b940.png","gorse-io","Gorse","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fgorse-io_df6990bd.png","AI powered open source recommender system.",null,"gorse_io","https:\u002F\u002Fgorse.io","https:\u002F\u002Fgithub.com\u002Fgorse-io",[81,85,89,93,97,101,104,108,111],{"name":82,"color":83,"percentage":84},"Go","#00ADD8",73.3,{"name":86,"color":87,"percentage":88},"Assembly","#6E4C13",17.8,{"name":90,"color":91,"percentage":92},"C","#555555",7.8,{"name":94,"color":95,"percentage":96},"Python","#3572A5",0.4,{"name":98,"color":99,"percentage":100},"Jinja","#a52a22",0.2,{"name":102,"color":103,"percentage":100},"Dockerfile","#384d54",{"name":105,"color":106,"percentage":107},"Shell","#89e051",0.1,{"name":109,"color":110,"percentage":107},"HCL","#844FBA",{"name":112,"color":113,"percentage":114},"Makefile","#427819",0,9598,887,"2026-04-10T03:07:49","Apache-2.0","Linux, macOS, Windows","未说明 (基于 Go 语言开发，主要依赖 CPU 进行经典推荐算法训练；若使用 LLM 功能可能需额外配置，但 README 未明确具体 GPU 需求)","未说明 (架构包含主节点、工作节点和服务节点，内存需求取决于数据量和并发量)",{"notes":123,"python":124,"dependencies":125},"1. 该工具主要使用 Go 语言编写，无需 Python 环境即可运行核心功能。\n2. 支持多种数据库作为后端存储（MySQL, MongoDB, Postgres, ClickHouse）和缓存层（Redis 等）。\n3. 提供 Docker 一键启动模式（playground）用于快速体验。\n4. 架构为单节点训练、分布式预测，包含主节点（负责训练和管理）、工作节点（负责离线推荐）和服务节点（负责 API 和实时推荐）。\n5. 支持通过嵌入方式处理多模态内容（文本、图像、视频），并支持基于 LLM 的推荐器，但具体 LLM 依赖未在 README 中详细列出。","不需要 (核心引擎由 Go 语言编写)",[126,127,128],"Go (版本见徽章指示)","MySQL \u002F MongoDB \u002F Postgres \u002F ClickHouse (数据存储)","Redis \u002F MySQL \u002F MongoDB \u002F Postgres (缓存)",[14],[131,132,133,134,135],"recommender-system","collaborative-filtering","go","knn","machine-learning","2026-03-27T02:49:30.150509","2026-04-11T08:11:30.724344",[],[140,145,150,155,160,165,170,175,180,185,190,195,200,205,210,215,220,225,230,235],{"id":141,"version":142,"summary_zh":143,"released_at":144},197952,"v0.5.6","## 功能\r\n\r\n- 为隐藏商品添加商品到商品的推荐功能 (#1197)\r\n\r\n## 修复\r\n- 通过精确到毫秒的方式，修复协同过滤模型和点击模型中的物料ID冲突问题 (#1196)","2026-03-20T14:48:10",{"id":146,"version":147,"summary_zh":148,"released_at":149},197953,"v0.5.5","## 功能\n- 使用重排序器替换 LLM 排序器 (#1170)\n## 修复\n- 修复 Redis 在大规模集合上更新分数的问题 (#1173)\n- 修复模型元数据更新时的锁问题 (#1185)，由 @sandesh-bhandari-dev 完成\n- 修复 PostgreSQL `COUNT()` 返回负值的问题 (#1186)，由 @legenduck 完成\n- 修复 Postgres 和 MySQL 中的排序规则问题 (#1191)\n","2026-03-07T03:27:45",{"id":151,"version":152,"summary_zh":153,"released_at":154},197954,"v0.5.4","## 功能\n\n- 添加 OpenAI 配置环境变量 (#1159)。\n\n## 修复\n\n- 如果没有排序器，则从推荐器获取结果 (#1156)。\n- 修复在 ClickHouse 中使用 HAVING 子句时 GetUserFeedback 的问题 (#1157)。\n- 修复 Docker 镜像中证书验证失败的问题 (#1161)。","2026-02-03T12:44:13",{"id":156,"version":157,"summary_zh":158,"released_at":159},197955,"v0.5.3","## 功能特性\n\n- 支持基于大语言模型的排序器 (#1129)。\n- 支持在 RecFlow 中不使用排序节点 (#1148)。\n- 支持在 RecFlow 中集成外部推荐系统 (#1146)。\n- 支持无需配置文件即可启动主节点 (#1150)。\n- 增加对 Azure Blob 存储的支持 (#1152)。\n\n## 修复\n\n- 在设置新推荐之前，从缓存中删除过时的推荐 (#1145)。\n\n## 重大变更\n\n- 需要指定协同过滤的类型。\n\n```toml\n[recommend.collaborative]\n\n# 协同过滤的类型。支持的值：\n#   none：禁用协同过滤。\n#   mf：矩阵分解。\ntype = \"mf\"\n```","2026-01-31T06:47:21",{"id":161,"version":162,"summary_zh":163,"released_at":164},197956,"v0.5.2","## 功能特性\n\n- 数据导入后自动重启任务 (#1130)。\n- 支持通过 LogicFlow 编辑推荐管道 (#1131)。\n- 最新物品 API 支持偏移量和用户 ID 参数 (#1136)。\n- 支持通过 `X-API-Version: 2` 头部返回推荐得分 (#1140)。\n- 添加 MySQL 连接池配置 (#1141)，由 @slaout 实现。\n- 添加 Postgres 连接池配置 (#1142)，由 @slaout 实现。\n- 支持基于嵌入的多模态点击率预测 (#1134)。\n- 支持通过设置 `optimize_period = \"0\"` 来禁用优化 (#1144)。\n\n## 重大变更\n\n- 配置文件中必须指定排序器类型，否则将不使用任何排序器。\n\n```toml\n[recommend.ranker]\n\n# 排序器的类型。有两种类型：\n#   none：无排序（默认）。\n#   fm：因子分解机。\ntype = \"fm\"\n```","2026-01-17T12:50:43",{"id":166,"version":167,"summary_zh":168,"released_at":169},197957,"v0.5.1","## 修复\r\n\r\n- 修复 Blob 客户端初始化问题 (#1128)。","2025-12-23T04:00:55",{"id":171,"version":172,"summary_zh":173,"released_at":174},197958,"v0.5.0","## 功能特性\n\n- 支持 Google Cloud Storage (#1123)。\n\n## 修复\n\n- 修复 API 文档中返回类型的拼写错误 (#1108)。\n- 修复获取用户邻居的 API (#1110)。\n\n## 破坏性变更\n\n- 在反馈中添加最后更新时间戳 (#1114)。\n- 向 `[recommend.ranker]` 配置中添加拟合和优化配置 (#1124)。\n\n升级指南：[Gorse v0.5 已发布](https:\u002F\u002Fgorse.io\u002Fposts\u002Frelease-0.5.html)","2025-12-20T05:21:22",{"id":176,"version":177,"summary_zh":178,"released_at":179},197959,"v0.5.0-rc","> **警告** 这是 Gorse 推荐系统的候选发布版本。这意味着：\n> 1. 请勿在生产环境中使用此预发布版本，除非您能够承担相关风险。\n> 2. 预发布版本之间可能会对 API、配置和数据存储进行更改。\n> 3. 升级后请清除本地缓存和缓存数据库。\n\n## 修复\n\n- 移除 ClickHouse 中的反馈覆盖支持 (#1103)。\n- 修复从 Redis 解码空类别的问题 (#1104)。\n- 在工作节点上添加就绪检查端点 (#1105)。\n- 修复 MySQL 中的零日期时间值问题 (#1106)。\n","2025-11-23T06:02:10",{"id":181,"version":182,"summary_zh":183,"released_at":184},197960,"v0.5.0-alpha.8","> **警告** 这是 Gorse 推荐系统的预发布版本。这意味着：\n> 1. 请勿在生产环境中使用预发布版本，除非您能够承担相关风险。\n> 2. 预发布版本之间可能会对 API、配置和数据存储进行更改。\n> 3. 升级后请清除本地缓存和缓存数据库。\n## 性能优化\n\n- 修复 gorse-in-one 的 CPU 竞争问题 (#1092)。\n- 按需为工作进程缓存物品 (#1093)。\n- 在离线推荐中跳过冷启动用户 (#1095)。\n\n## 修复\n\n- 添加非个性化推荐摘要 (#1086)。\n- 当缓存值不存在时，返回零值而非错误 (#1088)。\n- 修复带有数值的反馈导入问题 (#1090)。\n\n## 重大变更\n\n- 实现新的多路排序器 (#1077)。\n- 将欧氏距离转换为相似度 (#1091)。\n- 从二进制转储文件加载 Playground 数据 (#1100)。","2025-11-15T19:12:33",{"id":186,"version":187,"summary_zh":188,"released_at":189},197961,"v0.5.0-alpha.7","> **警告** 这是 Gorse 推荐系统的预发布版本。这意味着：\n> 1. 请勿在生产环境中使用预发布版本，除非您能够承担相关风险。\n> 2. 预发布版本之间可能会对 API、配置和数据存储进行更改。\n> 3. 升级后请清除本地缓存和缓存数据库。\n## 功能\n\n- 实现外部推荐器 (#1068)。\n- 支持提前停止训练 (#1074)。\n- 在仪表板上按时间戳对用户反馈进行排序 (#1076)。\n\n## 性能\n\n- 默认支持 OpenBLAS (#1066)。\n- 训练结束后释放数据集 (#1075)。\n- 从数据库而非缓存获取最新项目 (#1078)。\n\n## 修复\n\n- 修复加载空数据集时的宕机问题 (#1069)。\n- 通过降级 gorm 修复 PostgreSQL 初始化问题 (#1071)。\n\n## 重大变更\n\n- 移除 `User` 结构体中无用的 `Subscribe` 字段 (#1072)。\n- 移除热门推荐器 (#1080)。","2025-11-02T16:06:43",{"id":191,"version":192,"summary_zh":193,"released_at":194},197962,"v0.5.0-alpha.6","> **Warning** This is a pre-release version of the Gorse recommender system. It means:\r\n> 1. Don't use the pre-release version in production unless you can afford the risks.\r\n> 2. APIs, configurations, and data storage might be changed between pre-release versions.\r\n> 3. Clear local caches and cache databases after the upgrade.\r\n## Performance\r\n\r\n- Dump matrix factorization to HNSW index (#1051).\r\n- Optimze hyperparameter via optuna (#1002).\r\n- Release memory of models after dump (#1053).\r\n- Support Intel® OneAPI Math Kernel Library (#1058).\r\n\r\n## Fix\r\n- Fix user-to-user recommendation preview (#1054).","2025-09-21T15:25:35",{"id":196,"version":197,"summary_zh":198,"released_at":199},197963,"v0.5.0-alpha.5","> **Warning** This is a pre-release version of the Gorse recommender system. It means:\r\n> 1. Don't use the pre-release version in production unless you can afford the risks.\r\n> 2. APIs, configurations, and data storage might be changed between pre-release versions.\r\n> 3. Clear local caches and cache databases after the upgrade.\r\n## Features\r\n- Support Apache Kvrocks (#1034).\r\n\r\n## Fix\r\n- Fix broken item neighbors API (#1040).\r\n- Fix user avatar in dashboard (#1048).\r\n\r\n## BREAK CHANGES\r\n- Name workers and servers by hostname and port (#1046).\r\n- Replace `\u002Fapi\u002Fintermediate\u002Frecommend\u002F*` with `\u002Fapi\u002Fcollaborative-filtering\u002F*` (#1047).","2025-08-25T14:16:12",{"id":201,"version":202,"summary_zh":203,"released_at":204},197964,"v0.5.0-alpha.4","> **Warning** This is a pre-release version of the Gorse recommender system. It means:\r\n> 1. Don't use the pre-release version in production unless you can afford the risks.\r\n> 2. APIs, configurations, and data storage might be changed between pre-release versions.\r\n> 3. Clear local caches and cache databases after the upgrade.\r\n## Features\r\n- Implement item to item recommendation (#905).\r\n- Implement LLM-based recommenders (#941).\r\n- Support feedback with value (#1007).\r\n## BREAK CHANGES\r\n- Remove inverted file index for neighbors (#929).\r\n- Enable HNSW index for matrix factorization by default (#934).","2025-08-15T14:00:39",{"id":206,"version":207,"summary_zh":208,"released_at":209},197965,"v0.5.0-alpha.3","> **Warning** This is a pre-release version of the Gorse recommender system. It means:\r\n> 1. Don't use the pre-release version in production unless you can afford the risks.\r\n> 2. APIs, configurations, and data storage might be changed between pre-release versions.\r\n> 3. Clear local caches and cache databases after the upgrade.\r\n## Features\r\n- Re-enable Redis cluster support (#924).","2025-01-14T13:06:21",{"id":211,"version":212,"summary_zh":213,"released_at":214},197966,"v0.5.0-alpha.2","> **Warning** This is a pre-release version of the Gorse recommender system. It means:\r\n> 1. Don't use the pre-release version in production unless you can afford the risks.\r\n> 2. APIs, configurations, and data storage might be changed between pre-release versions.\r\n> 3. Clear local caches and cache databases after the upgrade.\r\n## Features\r\n- Support dump and restore backup (#886).\r\n- Support dashboard authentication via OIDC (#888).\r\n- Support set isolation level in configuration (#889).\r\n- Support mTLS between nodes (#893).\r\n- Support SQLite in cluster mode (#896).\r\n- Add non-personalized recommender (#884).\r\n## Fix\r\n- Enable multi-categories filtering for all APIs (#902).\r\n## BREAK CHANGES\r\n- Replace the import and export formats with JSON (#885).","2024-12-21T01:56:46",{"id":216,"version":217,"summary_zh":218,"released_at":219},197967,"v0.5.0-alpha.1","> **Warning** This is a pre-release version of the Gorse recommender system. It means:\r\n> 1. Don't use the pre-release version in production unless you can afford risks.\r\n> 2. APIs, configurations, and data storage might be changed between pre-release versions.\r\n> 3. Clear local caches and cache databases after the upgrade.\r\n## Features\r\n- Support ClickHouse back as data storage (#874).\r\n## Fix\r\n- Fix RediSearch query using escape (#770).\r\n- Upgrade the Redis client version for RediSearch (#868).\r\n- Remove unnecessary score check (#873).","2024-11-05T12:01:20",{"id":221,"version":222,"summary_zh":223,"released_at":224},197968,"v0.4.16","## Features\r\n\r\n- Re-enable Redis Cluster since RediSearch works in the cluster with the right plugin (#807) by @martinhoch42.","2025-01-07T13:58:48",{"id":226,"version":227,"summary_zh":228,"released_at":229},197969,"v0.4.15","## Features\r\n\r\n- Get more details from the recommendation (https:\u002F\u002Fgithub.com\u002Fgorse-io\u002Fgorse\u002Fpull\u002F796) by @ubrmnsh","2024-01-10T04:44:45",{"id":231,"version":232,"summary_zh":233,"released_at":234},197970,"v0.5.0-alpha","> **Warning** This is a pre-release version of the Gorse recommender system. It means:\r\n> 1. Don't use the pre-release version in production unless you can afford risks.\r\n> 2. APIs, configurations, and data storage might be changed between pre-release versions.\r\n> 3. Clear local caches and cache databases after the upgrade.\r\n## Features\r\n- Support JSON labels (#647).\r\n- Remove recommendation cache for inactive users (#691).\r\n- Support multiple categories filtering (#704).\r\n## BREAK CHANGES\r\n- Local caches are stored in a folder (#640).\r\n- Remove Oracle database support (#641).\r\n- Remove Redis cluster support (#645).\r\n- Remove ClickHouse support (#648).","2023-06-05T13:50:43",{"id":236,"version":237,"summary_zh":238,"released_at":239},197971,"v0.4.14","## Fix\r\n\r\n- Fix the concurrent access problem of  `rand.Rand` (#694) by @WisperDin.\r\n- Fix concurrent map iteration and map write in the worker (#697).\r\n- Fix index out of range caused by duplicate labels (#700).","2023-06-04T06:05:14"]