[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-mlrun--mlrun":3,"tool-mlrun--mlrun":64},[4,17,27,35,43,56],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":16},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。",162132,3,"2026-04-05T11:01:52",[13,14,15],"开发框架","图像","Agent","ready",{"id":18,"name":19,"github_repo":20,"description_zh":21,"stars":22,"difficulty_score":23,"last_commit_at":24,"category_tags":25,"status":16},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",138956,2,"2026-04-05T11:33:21",[13,15,26],"语言模型",{"id":28,"name":29,"github_repo":30,"description_zh":31,"stars":32,"difficulty_score":23,"last_commit_at":33,"category_tags":34,"status":16},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",107662,"2026-04-03T11:11:01",[13,14,15],{"id":36,"name":37,"github_repo":38,"description_zh":39,"stars":40,"difficulty_score":23,"last_commit_at":41,"category_tags":42,"status":16},3704,"NextChat","ChatGPTNextWeb\u002FNextChat","NextChat 是一款轻量且极速的 AI 助手，旨在为用户提供流畅、跨平台的大模型交互体验。它完美解决了用户在多设备间切换时难以保持对话连续性，以及面对众多 AI 模型不知如何统一管理的痛点。无论是日常办公、学习辅助还是创意激发，NextChat 都能让用户随时随地通过网页、iOS、Android、Windows、MacOS 或 Linux 端无缝接入智能服务。\n\n这款工具非常适合普通用户、学生、职场人士以及需要私有化部署的企业团队使用。对于开发者而言，它也提供了便捷的自托管方案，支持一键部署到 Vercel 或 Zeabur 等平台。\n\nNextChat 的核心亮点在于其广泛的模型兼容性，原生支持 Claude、DeepSeek、GPT-4 及 Gemini Pro 等主流大模型，让用户在一个界面即可自由切换不同 AI 能力。此外，它还率先支持 MCP（Model Context Protocol）协议，增强了上下文处理能力。针对企业用户，NextChat 提供专业版解决方案，具备品牌定制、细粒度权限控制、内部知识库整合及安全审计等功能，满足公司对数据隐私和个性化管理的高标准要求。",87618,"2026-04-05T07:20:52",[13,26],{"id":44,"name":45,"github_repo":46,"description_zh":47,"stars":48,"difficulty_score":23,"last_commit_at":49,"category_tags":50,"status":16},2268,"ML-For-Beginners","microsoft\u002FML-For-Beginners","ML-For-Beginners 是由微软推出的一套系统化机器学习入门课程，旨在帮助零基础用户轻松掌握经典机器学习知识。这套课程将学习路径规划为 12 周，包含 26 节精炼课程和 52 道配套测验，内容涵盖从基础概念到实际应用的完整流程，有效解决了初学者面对庞大知识体系时无从下手、缺乏结构化指导的痛点。\n\n无论是希望转型的开发者、需要补充算法背景的研究人员，还是对人工智能充满好奇的普通爱好者，都能从中受益。课程不仅提供了清晰的理论讲解，还强调动手实践，让用户在循序渐进中建立扎实的技能基础。其独特的亮点在于强大的多语言支持，通过自动化机制提供了包括简体中文在内的 50 多种语言版本，极大地降低了全球不同背景用户的学习门槛。此外，项目采用开源协作模式，社区活跃且内容持续更新，确保学习者能获取前沿且准确的技术资讯。如果你正寻找一条清晰、友好且专业的机器学习入门之路，ML-For-Beginners 将是理想的起点。",84991,"2026-04-05T10:45:23",[14,51,52,53,15,54,26,13,55],"数据工具","视频","插件","其他","音频",{"id":57,"name":58,"github_repo":59,"description_zh":60,"stars":61,"difficulty_score":10,"last_commit_at":62,"category_tags":63,"status":16},3128,"ragflow","infiniflow\u002Fragflow","RAGFlow 是一款领先的开源检索增强生成（RAG）引擎，旨在为大语言模型构建更精准、可靠的上下文层。它巧妙地将前沿的 RAG 技术与智能体（Agent）能力相结合，不仅支持从各类文档中高效提取知识，还能让模型基于这些知识进行逻辑推理和任务执行。\n\n在大模型应用中，幻觉问题和知识滞后是常见痛点。RAGFlow 通过深度解析复杂文档结构（如表格、图表及混合排版），显著提升了信息检索的准确度，从而有效减少模型“胡编乱造”的现象，确保回答既有据可依又具备时效性。其内置的智能体机制更进一步，使系统不仅能回答问题，还能自主规划步骤解决复杂问题。\n\n这款工具特别适合开发者、企业技术团队以及 AI 研究人员使用。无论是希望快速搭建私有知识库问答系统，还是致力于探索大模型在垂直领域落地的创新者，都能从中受益。RAGFlow 提供了可视化的工作流编排界面和灵活的 API 接口，既降低了非算法背景用户的上手门槛，也满足了专业开发者对系统深度定制的需求。作为基于 Apache 2.0 协议开源的项目，它正成为连接通用大模型与行业专有知识之间的重要桥梁。",77062,"2026-04-04T04:44:48",[15,14,13,26,54],{"id":65,"github_repo":66,"name":67,"description_en":68,"description_zh":69,"ai_summary_zh":70,"readme_en":71,"readme_zh":72,"quickstart_zh":73,"use_case_zh":74,"hero_image_url":75,"owner_login":67,"owner_name":76,"owner_avatar_url":77,"owner_bio":78,"owner_company":79,"owner_location":79,"owner_email":79,"owner_twitter":79,"owner_website":80,"owner_url":81,"languages":82,"stars":112,"forks":113,"last_commit_at":114,"license":115,"difficulty_score":116,"env_os":117,"env_gpu":118,"env_ram":117,"env_deps":119,"category_tags":124,"github_topics":125,"view_count":23,"oss_zip_url":79,"oss_zip_packed_at":79,"status":16,"created_at":136,"updated_at":137,"faqs":138,"releases":167},2093,"mlrun\u002Fmlrun","mlrun","MLRun is an open source MLOps platform for quickly building and managing continuous ML applications across their lifecycle. MLRun integrates into your development and CI\u002FCD environment and automates the delivery of production data, ML pipelines, and online applications.","MLRun 是一款开源的 MLOps 平台，旨在帮助团队快速构建并管理贯穿全生命周期的连续机器学习与生成式 AI 应用。它无缝集成到现有的开发流程及 CI\u002FCD 环境中，自动化处理生产数据交付、机器学习流水线编排以及在线应用的部署监控。\n\n在人工智能项目从实验走向生产的过程中，团队常面临数据孤岛、工程实施复杂、资源消耗大以及协作效率低等挑战。MLRun 通过统一的编排能力有效解决了这些痛点，显著减少了重复的工程工作量，缩短了模型上线时间，并优化了计算资源的使用。它打破了数据科学家、算法工程师、软件开发人员及运维团队之间的壁垒，让不同角色的成员能在同一平台上高效协作，实现模型的持续迭代与改进。\n\nMLRun 特别适合从事机器学习工程化、大模型应用开发的技术团队使用，包括算法研究员、后端开发者及 MLOps 工程师。其独特的技术亮点在于支持任意本地或云端 IDE 进行开发，具备强大的数据血缘追踪与版本管理能力，并能灵活适配多种数据存储、向量数据库及主流 AI 框架。无论是处理结构化数据还是利用大语言模型加工非结构化信息，MLRun 都能提供从数据预处理、模型训练评估到大规模部署的一站式解决","MLRun 是一款开源的 MLOps 平台，旨在帮助团队快速构建并管理贯穿全生命周期的连续机器学习与生成式 AI 应用。它无缝集成到现有的开发流程及 CI\u002FCD 环境中，自动化处理生产数据交付、机器学习流水线编排以及在线应用的部署监控。\n\n在人工智能项目从实验走向生产的过程中，团队常面临数据孤岛、工程实施复杂、资源消耗大以及协作效率低等挑战。MLRun 通过统一的编排能力有效解决了这些痛点，显著减少了重复的工程工作量，缩短了模型上线时间，并优化了计算资源的使用。它打破了数据科学家、算法工程师、软件开发人员及运维团队之间的壁垒，让不同角色的成员能在同一平台上高效协作，实现模型的持续迭代与改进。\n\nMLRun 特别适合从事机器学习工程化、大模型应用开发的技术团队使用，包括算法研究员、后端开发者及 MLOps 工程师。其独特的技术亮点在于支持任意本地或云端 IDE 进行开发，具备强大的数据血缘追踪与版本管理能力，并能灵活适配多种数据存储、向量数据库及主流 AI 框架。无论是处理结构化数据还是利用大语言模型加工非结构化信息，MLRun 都能提供从数据预处理、模型训练评估到大规模部署的一站式解决方案，让 AI 应用的落地变得更加简单可控。","\u003Ca id=\"top\">\u003C\u002Fa>\n[![Build Status](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fmlrun\u002Factions\u002Fworkflows\u002Fbuild.yaml\u002Fbadge.svg?branch=development)](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fmlrun\u002Factions\u002Fworkflows\u002Fbuild.yaml?query=branch%3Adevelopment)\n[![License](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-Apache%202.0-blue.svg)](https:\u002F\u002Fopensource.org\u002Flicenses\u002FApache-2.0)\n[![PyPI version fury.io](https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Fmlrun.svg)](https:\u002F\u002Fpypi.python.org\u002Fpypi\u002Fmlrun\u002F)\n[![Documentation](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlrun_mlrun_readme_13d664e1afd7.png)](https:\u002F\u002Fmlrun.readthedocs.io\u002Fen\u002Flatest\u002F?badge=latest)\n[![Ruff](https:\u002F\u002Fimg.shields.io\u002Fendpoint?url=https:\u002F\u002Fraw.githubusercontent.com\u002Fastral-sh\u002Fruff\u002Fmain\u002Fassets\u002Fbadge\u002Fv2.json)](https:\u002F\u002Fgithub.com\u002Fastral-sh\u002Fruff)\n[![GitHub commit activity](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fcommit-activity\u002Fw\u002Fmlrun\u002Fmlrun)](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fmlrun\u002Fcommits\u002Fmain)\n[![GitHub release (latest SemVer)](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fv\u002Frelease\u002Fmlrun\u002Fmlrun?sort=semver)](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fmlrun\u002Freleases)\n[![Join MLOps Live](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fslack-join_chat-white.svg?logo=slack&style=social)](https:\u002F\u002Fmlopslive.slack.com)\n\n\u003Cdiv>\n  \u003Cspan>\n    \u003Cpicture>\n      \u003Cimg img align=\"left\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlrun_mlrun_readme_48179110f2aa.png\" alt=\"MLRun logo\" width=\"150\"\u002F>\n    \u003C\u002Fpicture>\n  \u003C\u002Fspan>\n  \u003Cspan>\n    \u003Cpicture>\n      \u003Cimg img align=\"right\" src=\".\u002Fdocs\u002F_static\u002Fimages\u002Fmaintenance_logo.svg\" alt=\"Maintenance logo\" width=\"250\"\u002F>\n    \u003C\u002Fpicture>\n  \u003C\u002Fspan>\n  \u003Cbr clear=\"all\"\u002F>\n\u003C\u002Fdiv>\n\n# Using MLRun \n\nMLRun is an open source AI orchestration platform for quickly building and managing continuous (gen) AI applications across their lifecycle. MLRun integrates into your development and CI\u002FCD environment and automates the delivery of production data, ML pipelines, and online applications. \nMLRun significantly reduces engineering efforts, time to production, and computation resources. \nWith MLRun, you can choose any IDE on your local machine or on the cloud. MLRun breaks the silos between data, ML, software, and DevOps\u002FMLOps teams, enabling collaboration and fast continuous improvements.\n\nGet started with the MLRun [**Tutorials and Examples**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002Findex.html) and the [**Set up your client environment**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fsetup-guide.md), or read about the [**MLRun Architecture**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Farchitecture.html).\n\nThis page explains how MLRun addresses the [**gen AI tasks**](#genai-tasks), [**MLOps tasks**](#mlops-tasks), and presents the [**MLRun core components**](#core-components).\n\nSee the supported data stores, development tools, services, platforms, etc., supported by MLRun's open architecture in **https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fecosystem.html**.\n\n## Gen AI tasks\n\n\u003Cp align=\"center\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlrun_mlrun_readme_a0e624040ffc.png\" alt=\"ai-tasks\" width=\"800\"\u002F>\u003C\u002Fp>\u003Cbr>\n\nUse MLRun to develop, scale, deploy, and monitor your AI model across your enterprise. The [**gen AI development workflow**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fgenai-flow.html) \nsection describes the different tasks and stages in detail.\n\n### Data management\n\n\nMLRun supports batch or realtime data processing at scale, data lineage and versioning, structured and unstructured data, and more. \nRemoving inappropriate data at an early stage saves resources that would otherwise be required later on.\n\n\n**Docs:**\n[Using LLMs to process unstructured data](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fdata-mgmt\u002Funstructured-data.html),\n[Vector databases](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fdata-mgmt\u002Fvector-databases.html),\n[Guardrails for data management](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fdata-mgmt\u002Fguardrails-data.html)\n**Demo:**\n[Call center demo](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fdemo-call-center)\n**Video:**\n[Call center](https:\u002F\u002Fyoutu.be\u002FYycMbxRgLBA)\n\n### Development\nUse MLRun to build an automated ML pipeline to: collect data, \npreprocess (prepare) the data, run the training pipeline, and evaluate the model.\n\n**Docs:**\n[Working with RAG](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fdevelopment\u002Fworking-with-rag.html), [Evalating LLMs](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fdevelopment\u002Fevaluating-llms.html), [Fine tuning LLMS](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fdevelopment\u002Ffine-tuning-llms.html)\n**Demos:**\n[Call center demo](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fdemo-call-center),\n[Banking agent demo](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fdemo-banking-agent)\n**Video:**\n[Call center](https:\u002F\u002Fyoutu.be\u002FYycMbxRgLBA)\n\n\n### Deployment\nMLRun serving can productize the newly trained LLM as a serverless function using real-time auto-scaling Nuclio serverless functions. \nThe application pipeline includes all the steps from accepting events or data, contextualizing it with a state  preparing the required model features, \ninferring results using one or more models, and driving actions. \n\n\n**Docs:**\n[Serving gen AI models](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fdeployment\u002Fgenai_serving.html), [GPU utilization](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fdeployment\u002Fgpu_utilization.html), [Gen AI realtime serving graph](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fdeployment\u002Fgenai_serving_graph.html)\n**Tutorial:**\n[Deploy LLM using MLRun](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002Fgenai-01-basic-tutorial.html)\n**Demos:**\n[Call center demo](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fdemo-call-center),\n[Banking agent demo](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fdemo-banking-agent)\n**Video:**\n[Call center](https:\u002F\u002Fyoutu.be\u002FYycMbxRgLBA)\n\n\n### Live Ops\nMonitor all resources, data, model and application metrics to ensure performance. Then identify risks, control costs, and measure business KPIs.\nCollect production data, metadata, and metrics to tune the model and application further, and to enable governance and explainability.\n\n\n**Docs:**\n[Model monitoring \u003Cmonitoring](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fconcepts\u002Fmonitoring.html), [Alerts and notifications](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fconcepts\u002Falerts-notifications.html)\n**Tutorials:**\n[Deploy LLM using MLRun](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002Fgenai-01-basic-tutorial.html), [Model monitoring using LLM](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002Fgenai-02-monitoring-llm.html)\n**Demo:**\n[Banking agent demo](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fdemo-banking-agent)\n\n\n\u003Ca id=\"mlops-tasks\">\u003C\u002Fa>\n## MLOps tasks\n\n\u003Cp align=\"center\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlrun_mlrun_readme_85b15a485554.png\" alt=\"mlrun-tasks\" width=\"800\"\u002F>\u003C\u002Fp>\u003Cbr>\n\nThe [**MLOps development workflow**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fmlops-dev-flow.html) section describes the different tasks and stages in detail.\nMLRun can be used to automate and orchestrate all the different tasks or just specific tasks (and integrate them with what you have already deployed).\n\n### Project management and CI\u002FCD automation\n\nIn MLRun the assets, metadata, and services (data, functions, jobs, artifacts, models, secrets, etc.) are organized into projects.\nProjects can be imported\u002Fexported as a whole, mapped to git repositories or IDE projects (in PyCharm, VSCode, etc.), which enables versioning, collaboration, and CI\u002FCD. \nProject access can be restricted to a set of users and roles. \n\n**Docs:** [Projects and Automation](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fprojects\u002Fproject.html), [CI\u002FCD Integration](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fprojects\u002Fci-integration.html)\n**Tutorials:** [Quick start](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002F01-mlrun-basics.html), [Automated ML Pipeline](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002F04-pipeline.html)\n**Video:** [Quick start](https:\u002F\u002Fyoutu.be\u002FxI8KVGLlj7Q).\n\n### Ingest and process data\n\nMLRun provides abstract interfaces to various offline and online [**data sources**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fstore\u002Fdatastore.html), supports batch or realtime data processing at scale, data lineage and versioning, structured and unstructured data, and more. \nIn addition, the MLRun [**Feature Store**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ffeature-store\u002Ffeature-store.html) automates the collection, transformation, storage, catalog, serving, and monitoring of data features across the ML lifecycle and enables feature reuse and sharing. \n\nSee: **Docs:** [Ingest and process data](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fdata-prep\u002Findex.html), [Feature Store](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ffeature-store\u002Ffeature-store.html), [Data & Artifacts](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fconcepts\u002Fdata.html)\n**Tutorials:** [Quick start](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002F01-mlrun-basics.html), [Feature Store](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ffeature-store\u002Fbasic-demo.html).\n\n### Develop and train models\n\nMLRun allows you to easily build ML pipelines that take data from various sources or the Feature Store and process it, train models at scale with multiple parameters, test models, tracks each experiments, register, version and deploy models, etc. MLRun provides scalable built-in or custom model training services, integrate with any framework and can work with 3rd party training\u002Fauto-ML services. You can also bring your own pre-trained model and use it in the pipeline.\n\n**Docs:** [Develop and train models](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fdevelopment\u002Findex.html), [Model Training and Tracking](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fdevelopment\u002Fmodel-training-tracking.html), [Batch Runs and Workflows](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fconcepts\u002Fruns-workflows.html)\n**Tutorials:** [Train, compare, and register models](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002F02-model-training.html), [Automated ML Pipeline](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002F04-pipeline.html)\n**Video:** [Train and compare models](https:\u002F\u002Fyoutu.be\u002FbZgBsmLMdQo).\n\n### Deploy models and applications\n\nMLRun rapidly deploys and manages production-grade real-time or batch application pipelines using elastic and resilient serverless functions. MLRun addresses the entire ML application: intercepting application\u002Fuser requests, running data processing tasks, inferencing using one or more models, driving actions, and integrating with the application logic.\n\n**Docs:** [Deploy models and applications](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fdeployment\u002Findex.html), [Realtime Pipelines](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fserving\u002Fserving-graph.html), [Batch Inference](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fdeployment\u002Fbatch_inference.html)\n**Tutorials:** [Realtime Serving](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002F03-model-serving.html), [Batch Inference](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002F07-batch-infer.html), [Advanced Pipeline](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002F07-batch-infer.html)\n**Video:** [Serving pre-trained models](https:\u002F\u002Fyoutu.be\u002FOUjOus4dZfw).\n\n### Model Monitoring\n\nObservability is built into the different MLRun objects (data, functions, jobs, models, pipelines, etc.), eliminating the need for complex integrations and code instrumentation. With MLRun, you can observe the application\u002Fmodel resource usage and model behavior (drift, performance, etc.), define custom app metrics, and trigger alerts or retraining jobs.\n\n**Docs:** [Model monitoring](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fconcepts\u002Fmodel-monitoring.html), [Model Monitoring Overview](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fmonitoring\u002Fmodel-monitoring-deployment.html)\n**Tutorials:** [Model Monitoring & Drift Detection](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002F05-model-monitoring.html).\n\n\n\u003Ca id=\"core-components\">\u003C\u002Fa>\n## MLRun core components\n\n\u003Cp align=\"center\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlrun_mlrun_readme_eb8fc253d65c.png\" alt=\"mlrun-core\" width=\"800\"\u002F>\u003C\u002Fp>\u003Cbr>\n\n\nMLRun includes the following major components:\n\n[**Project Management:**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fprojects\u002Fproject.html) A service (API, SDK, DB, UI) that manages the different project assets (data, functions, jobs, workflows, secrets, etc.) and provides central control and metadata layer.  \n\n[**Functions:**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fruntimes\u002Ffunctions.html) automatically deployed software package with one or more methods and runtime-specific attributes (such as image, libraries, command, arguments, resources, etc.).\n\n[**Data & Artifacts:**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fconcepts\u002Fdata.html) Glueless connectivity to various data sources, metadata management, catalog, and versioning for structures\u002Funstructured artifacts.\n\n[**Batch Runs & Workflows:**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fconcepts\u002Fruns-workflows.html) Execute one or more functions with specific parameters and collect, track, and compare all their results and artifacts.\n\n[**Real-Time Serving Pipeline:**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fserving\u002Fserving-graph.html) Rapid deployment of scalable data and ML pipelines using real-time serverless technology, including API handling, data preparation\u002Fenrichment, model serving, ensembles, driving and measuring actions, etc.\n\n[**Model monitoring:**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fmonitoring\u002Findex.html) monitors data, models, resources, and production components and provides a feedback loop for exploring production data, identifying drift, alerting on anomalies or data quality issues, triggering retraining jobs, measuring business impact, etc.\n\n[**Alerts and notifications:**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fconcepts\u002Fmodel-monitoring.html) Use alerts to identify and inform you of possible problem situations. Use notifications to report status on runs and pipelines.\n\n[**Feature Store:**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ffeature-store\u002Ffeature-store.html) automatically collects, prepares, catalogs, and serves production data features for development (offline) and real-time (online) deployment using minimal engineering effort.\n","\u003Ca id=\"top\">\u003C\u002Fa>\n[![构建状态](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fmlrun\u002Factions\u002Fworkflows\u002Fbuild.yaml\u002Fbadge.svg?branch=development)](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fmlrun\u002Factions\u002Fworkflows\u002Fbuild.yaml?query=branch%3Adevelopment)\n[![许可证](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-Apache%202.0-blue.svg)](https:\u002F\u002Fopensource.org\u002Flicenses\u002FApache-2.0)\n[![PyPI版本fury.io](https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Fmlrun.svg)](https:\u002F\u002Fpypi.python.org\u002Fpypi\u002Fmlrun\u002F)\n[![文档](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlrun_mlrun_readme_13d664e1afd7.png)](https:\u002F\u002Fmlrun.readthedocs.io\u002Fen\u002Flatest\u002F?badge=latest)\n[![Ruff](https:\u002F\u002Fimg.shields.io\u002Fendpoint?url=https:\u002F\u002Fraw.githubusercontent.com\u002Fastral-sh\u002Fruff\u002Fmain\u002Fassets\u002Fbadge\u002Fv2.json)](https:\u002F\u002Fgithub.com\u002Fastral-sh\u002Fruff)\n[![GitHub提交活动](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fcommit-activity\u002Fw\u002Fmlrun\u002Fmlrun)](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fmlrun\u002Fcommits\u002Fmain)\n[![GitHub发布（最新SemVer）](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fv\u002Frelease\u002Fmlrun\u002Fmlrun?sort=semver)](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fmlrun\u002Freleases)\n[![加入MLOps Live](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fslack-join_chat-white.svg?logo=slack&style=social)](https:\u002F\u002Fmlopslive.slack.com)\n\n\u003Cdiv>\n  \u003Cspan>\n    \u003Cpicture>\n      \u003Cimg img align=\"left\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlrun_mlrun_readme_48179110f2aa.png\" alt=\"MLRun logo\" width=\"150\"\u002F>\n    \u003C\u002Fpicture>\n  \u003C\u002Fspan>\n  \u003Cspan>\n    \u003Cpicture>\n      \u003Cimg img align=\"right\" src=\".\u002Fdocs\u002F_static\u002Fimages\u002Fmaintenance_logo.svg\" alt=\"Maintenance logo\" width=\"250\"\u002F>\n    \u003C\u002Fpicture>\n  \u003C\u002Fspan>\n  \u003Cbr clear=\"all\"\u002F>\n\u003C\u002Fdiv>\n\n# 使用MLRun \n\nMLRun是一个开源的AI编排平台，用于在其整个生命周期内快速构建和管理连续的（生成式）AI应用。MLRun可以集成到您的开发和CI\u002FCD环境中，并自动化生产数据、ML流水线和在线应用的交付。\nMLRun显著减少了工程工作量、上线时间和计算资源。\n借助MLRun，您可以在本地或云端选择任何IDE。MLRun打破了数据、ML、软件以及DevOps\u002FMLOps团队之间的壁垒，从而实现协作和快速的持续改进。\n\n您可以从MLRun的[教程和示例](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002Findex.html)以及[设置客户端环境](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fsetup-guide.md)开始使用，或者阅读关于[MLRun架构](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Farchitecture.html)的内容。\n\n本页将解释MLRun如何解决[生成式AI任务](#genai-tasks)、[MLOps任务](#mlops-tasks)，并介绍[MLRun核心组件](#core-components)。\n\n有关MLRun开放架构支持的数据存储、开发工具、服务、平台等，请参阅**https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fecosystem.html**。\n\n## 生成式AI任务\n\n\u003Cp align=\"center\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlrun_mlrun_readme_a0e624040ffc.png\" alt=\"ai-tasks\" width=\"800\"\u002F>\u003C\u002Fp>\u003Cbr>\n\n使用MLRun在企业范围内开发、扩展、部署和监控您的AI模型。[生成式AI开发流程](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fgenai-flow.html)一节详细描述了不同的任务和阶段。\n\n### 数据管理\n\n\nMLRun支持大规模的批处理或实时数据处理、数据血缘与版本控制、结构化与非结构化数据等。在早期阶段去除不适当的数据，可以节省后续所需的资源。\n\n\n**文档：**\n[使用LLM处理非结构化数据](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fdata-mgmt\u002Funstructured-data.html),\n[向量数据库](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fdata-mgmt\u002Fvector-databases.html),\n[数据管理的护栏](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fdata-mgmt\u002Fguardrails-data.html)\n**演示：**\n[呼叫中心演示](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fdemo-call-center)\n**视频：**\n[呼叫中心](https:\u002F\u002Fyoutu.be\u002FYycMbxRgLBA)\n\n### 开发\n使用MLRun构建自动化的ML流水线，以收集数据、预处理（准备）数据、运行训练流水线并评估模型。\n\n**文档：**\n[使用RAG](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fdevelopment\u002Fworking-with-rag.html), [评估LLM](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fdevelopment\u002Fevaluating-llms.html), [微调LLM](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fdevelopment\u002Ffine-tuning-llms.html)\n**演示：**\n[呼叫中心演示](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fdemo-call-center),\n[银行代理演示](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fdemo-banking-agent)\n**视频：**\n[呼叫中心](https:\u002F\u002Fyoutu.be\u002FYycMbxRgLBA)\n\n\n### 部署\nMLRun的服务功能可以将新训练的LLM产品化为无服务器函数，利用实时自动伸缩的Nuclio无服务器函数。应用程序流水线包括从接收事件或数据、结合上下文准备所需模型特征、使用一个或多个模型进行推理，再到驱动行动的所有步骤。\n\n\n**文档：**\n[服务生成式AI模型](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fdeployment\u002Fgenai_serving.html), [GPU利用率](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fdeployment\u002Fgpu_utilization.html), [生成式AI实时服务图](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fgenai\u002Fdeployment\u002Fgenai_serving_graph.html)\n**教程：**\n[使用MLRun部署LLM](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002Fgenai-01-basic-tutorial.html)\n**演示：**\n[呼叫中心演示](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fdemo-call-center),\n[银行代理演示](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fdemo-banking-agent)\n**视频：**\n[呼叫中心](https:\u002F\u002Fyoutu.be\u002FYycMbxRgLBA)\n\n\n### 运营监控\n监控所有资源、数据、模型和应用指标，以确保性能。然后识别风险、控制成本并衡量业务KPI。\n收集生产数据、元数据和指标，以便进一步调整模型和应用，并实现治理和可解释性。\n\n\n**文档：**\n[模型监控\u003Cmonitoring](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fconcepts\u002Fmonitoring.html), [警报和通知](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fconcepts\u002Falerts-notifications.html)\n**教程：**\n[使用MLRun部署LLM](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002Fgenai-01-basic-tutorial.html), [使用LLM监控模型](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002Fgenai-02-monitoring-llm.html)\n**演示：**\n[银行代理演示](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fdemo-banking-agent)\n\n\n\u003Ca id=\"mlops-tasks\">\u003C\u002Fa>\n## MLOps任务\n\n\u003Cp align=\"center\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlrun_mlrun_readme_85b15a485554.png\" alt=\"mlrun-tasks\" width=\"800\"\u002F>\u003C\u002Fp>\u003Cbr>\n\n[**MLOps开发流程**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fmlops-dev-flow.html)一节详细描述了不同的任务和阶段。\nMLRun可用于自动化和编排所有不同的任务，也可以只编排特定的任务（并与您已部署的内容集成）。\n\n### 项目管理和 CI\u002FCD 自动化\n\n在 MLRun 中，资产、元数据和服务（数据、函数、作业、工件、模型、密钥等）都被组织到项目中。项目可以作为一个整体导入或导出，并映射到 Git 仓库或 IDE 项目（如 PyCharm、VSCode 等），从而实现版本控制、协作以及 CI\u002FCD 流程。项目访问权限可以限制为特定的用户和角色。\n\n**文档：** [项目与自动化](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fprojects\u002Fproject.html)、[CI\u002FCD 集成](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fprojects\u002Fci-integration.html)  \n**教程：** [快速入门](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002F01-mlrun-basics.html)、[自动化机器学习流水线](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002F04-pipeline.html)  \n**视频：** [快速入门](https:\u002F\u002Fyoutu.be\u002FxI8KVGLlj7Q)。\n\n### 数据摄取与处理\n\nMLRun 提供了针对各种离线和在线 **数据源** 的抽象接口（[文档](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fstore\u002Fdatastore.html)），支持大规模的批处理或实时数据处理、数据血缘与版本管理、结构化与非结构化数据等多种功能。此外，MLRun 的 **特征存储**（[文档](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ffeature-store\u002Ffeature-store.html)）能够自动完成数据特征在整个机器学习生命周期中的收集、转换、存储、目录化、服务化及监控工作，同时支持特征的复用与共享。\n\n更多信息请参阅：  \n**文档：** [数据摄取与处理](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fdata-prep\u002Findex.html)、[特征存储](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ffeature-store\u002Ffeature-store.html)、[数据与工件](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fconcepts\u002Fdata.html)  \n**教程：** [快速入门](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002F01-mlrun-basics.html)、[特征存储](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ffeature-store\u002Fbasic-demo.html)。\n\n### 模型开发与训练\n\nMLRun 允许您轻松构建机器学习流水线，从多种数据源或特征存储中获取数据并进行处理，以多种参数进行大规模模型训练，测试模型，跟踪每次实验，注册、版本化并部署模型等。MLRun 提供可扩展的内置或自定义模型训练服务，能够与任何框架集成，并可与第三方训练或自动机器学习服务协同工作。您还可以引入自己的预训练模型并在流水线中使用。\n\n**文档：** [模型开发与训练](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fdevelopment\u002Findex.html)、[模型训练与跟踪](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fdevelopment\u002Fmodel-training-tracking.html)、[批量运行与工作流](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fconcepts\u002Fruns-workflows.html)  \n**教程：** [训练、比较与注册模型](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002F02-model-training.html)、[自动化机器学习流水线](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002F04-pipeline.html)  \n**视频：** [训练与比较模型](https:\u002F\u002Fyoutu.be\u002FbZgBsmLMdQo)。\n\n### 模型与应用部署\n\nMLRun 可以利用弹性且高可用的无服务器函数，快速部署和管理生产级别的实时或批处理应用流水线。MLRun 覆盖整个机器学习应用流程：拦截应用\u002F用户请求，执行数据处理任务，使用一个或多个模型进行推理，驱动相应操作，并与应用逻辑无缝集成。\n\n**文档：** [模型与应用部署](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fdeployment\u002Findex.html)、[实时流水线](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fserving\u002Fserving-graph.html)、[批量推理](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fdeployment\u002Fbatch_inference.html)  \n**教程：** [实时服务](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002F03-model-serving.html)、[批量推理](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002F07-batch-infer.html)、[高级流水线](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002F07-batch-infer.html)  \n**视频：** [服务预训练模型](https:\u002F\u002Fyoutu.be\u002FOUjOus4dZfw)。\n\n### 模型监控\n\nMLRun 的各个对象（数据、函数、作业、模型、流水线等）都内置了可观性功能，无需复杂的集成或代码埋点。借助 MLRun，您可以监控应用\u002F模型的资源使用情况及模型行为（漂移、性能等），定义自定义应用指标，并触发警报或重新训练作业。\n\n**文档：** [模型监控](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fconcepts\u002Fmodel-monitoring.html)、[模型监控概述](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fmonitoring\u002Fmodel-monitoring-deployment.html)  \n**教程：** [模型监控与漂移检测](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002F05-model-monitoring.html)。\n\n\u003Ca id=\"core-components\">\u003C\u002Fa>\n## MLRun 核心组件\n\n\u003Cp align=\"center\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlrun_mlrun_readme_eb8fc253d65c.png\" alt=\"mlrun-core\" width=\"800\"\u002F>\u003C\u002Fp>\u003Cbr>\n\nMLRun 包括以下主要组件：\n\n[**项目管理：**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fprojects\u002Fproject.html) 一种服务（API、SDK、数据库、UI），用于管理项目的各类资产（数据、函数、作业、工作流、密钥等），并提供中央控制和元数据层。\n\n[**函数：**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fruntimes\u002Ffunctions.html) 自动部署的软件包，包含一个或多个方法以及运行时特定的属性（如镜像、库、命令、参数、资源等）。\n\n[**数据与工件：**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fconcepts\u002Fdata.html) 与各种数据源的无缝连接、元数据管理、目录化以及对结构化\u002F非结构化工件的版本控制。\n\n[**批量运行与工作流：**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fconcepts\u002Fruns-workflows.html) 使用特定参数执行一个或多个函数，并收集、跟踪和比较所有结果及工件。\n\n[**实时服务流水线：**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fserving\u002Fserving-graph.html) 利用实时无服务器技术快速部署可扩展的数据与机器学习流水线，包括 API 处理、数据准备\u002F增强、模型服务、集成模型、动作驱动与度量等。\n\n[**模型监控：**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fmonitoring\u002Findex.html) 监控数据、模型、资源及生产组件，提供反馈回路以探索生产数据、识别漂移、对异常或数据质量问题发出警报、触发重新训练作业、衡量业务影响等。\n\n[**告警与通知：**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Fconcepts\u002Fmodel-monitoring.html) 使用告警来识别并通知您潜在的问题；使用通知报告运行和流水线的状态。\n\n[**特征存储：**](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ffeature-store\u002Ffeature-store.html) 自动收集、准备、目录化并服务于开发（离线）和实时（在线）部署的生产级数据特征，同时最大限度地减少工程工作量。","# MLRun 快速上手指南\n\nMLRun 是一个开源的 AI 编排平台，旨在帮助用户快速构建和管理贯穿整个生命周期的连续（生成式）AI 应用。它能无缝集成到现有的开发和 CI\u002FCD 环境中，自动化交付生产数据、机器学习流水线及在线应用，显著降低工程成本并加速模型上线。\n\n## 1. 环境准备\n\n在开始之前，请确保您的开发环境满足以下要求：\n\n*   **操作系统**：Linux (推荐 Ubuntu 20.04+), macOS, 或 Windows (需使用 WSL2)。\n*   **Python 版本**：Python 3.9 - 3.11。\n*   **包管理工具**：`pip` (建议升级至最新版本)。\n*   **容器环境**：若需运行完整本地集群或部署服务，建议安装 **Docker**。\n*   **Kubernetes (可选)**：若要在生产级 K8s 集群中部署 MLRun 服务端，需具备 `kubectl` 访问权限及一个活跃的 K8s 集群（如 Minikube, EKS, GKE, ACK 等）。\n\n> **注意**：本指南主要侧重于客户端 SDK 的安装与基础使用。若需部署完整的 MLRun 服务端平台，请参考官方部署文档或使用 Helm Chart。\n\n## 2. 安装步骤\n\n### 安装 Python SDK\n\n使用 pip 安装最新稳定版的 MLRun 客户端库。\n\n```bash\npip install mlrun\n```\n\n**国内加速方案**：\n如果您在中国大陆地区，建议使用清华或阿里云镜像源以加快下载速度：\n\n```bash\npip install mlrun -i https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple\n```\n\n### 验证安装\n\n安装完成后，可通过以下命令验证版本：\n\n```bash\nmlrun --version\n```\n\n或者在 Python 交互环境中检查：\n\n```python\nimport mlrun\nprint(mlrun.__version__)\n```\n\n## 3. 基本使用\n\nMLRun 的核心概念包括 **项目 (Project)**、**函数 (Function)** 和 **任务 (Task)**。以下是最简单的入门流程，展示如何定义一个数据处理函数并将其作为任务运行。\n\n### 步骤 1: 初始化项目\n\n首先，导入库并创建一个项目上下文。项目用于组织代码、数据和模型等资源。\n\n```python\nimport mlrun\n\n# 初始化或获取名为 \"quick-start\" 的项目\nproject = mlrun.get_or_create_project(\"quick-start\", context=\".\u002F\")\n```\n\n### 步骤 2: 定义函数\n\n将您的 Python 代码封装为 MLRun 函数。您可以直接包装当前文件中的函数，或指向外部脚本。以下示例定义了一个简单的日志打印函数。\n\n```python\n# 定义一个简单的处理函数\ndef my_handler(context, name: str = \"world\"):\n    context.logger.info(f\"Hello {name}!\")\n    return {\"result\": f\"Hello {name}!\"}\n\n# 将函数注册为 MLRun 可执行对象\n# kind=\"job\" 表示这是一个批处理任务\nfn = mlrun.code_to_function(\n    name=\"hello-function\",\n    kind=\"job\",\n    filename=__file__, # 指向当前脚本文件\n    image=\"mlrun\u002Fmlrun:latest\" # 指定运行时的容器镜像\n)\n```\n\n### 步骤 3: 创建并运行任务\n\n创建一个任务实例，设置参数，并提交运行。\n\n```python\n# 创建任务，设置输入参数\ntask = mlrun.new_task(\n    name=\"my-first-run\",\n    params={\"name\": \"MLRun User\"},\n    outputs=[\"result\"]\n)\n\n# 提交运行\n# local=True 表示在本地 Docker 容器中运行（需安装 Docker）\n# 若无 Docker 环境，可设为 False 直接在本地进程运行（取决于配置）\nrun = fn.run(task, local=True)\n\n# 查看运行结果\nprint(run.outputs)\n```\n\n### 下一步\n\n完成上述步骤后，您已成功运行了第一个 MLRun 任务。接下来您可以探索：\n*   **数据摄入**：使用 `Feature Store` 管理特征数据。\n*   **流水线编排**：使用 `run_pipeline` 串联多个任务形成工作流。\n*   **模型服务**：将训练好的模型部署为实时 API (`kind=\"serving\"`)。\n*   **监控**：配置模型漂移检测和性能监控。\n\n更多详细教程和示例代码，请访问 [MLRun 官方文档](https:\u002F\u002Fdocs.mlrun.org\u002Fen\u002Fstable\u002Ftutorials\u002Findex.html)。","某金融科技公司的大数据团队正致力于构建一个实时反欺诈检测系统，需要频繁迭代模型并处理海量交易流数据。\n\n### 没有 mlrun 时\n- **环境割裂严重**：数据科学家在本地 Jupyter 笔记本中调试代码，而运维团队使用完全不同的 CI\u002FCD 脚本部署，导致“在我机器上能跑”的冲突频发。\n- **流水线手动拼接**：从数据清洗、特征工程到模型训练，每个环节都需要人工编写独立的 Shell 脚本串联，一旦数据源变更，整个链路极易断裂。\n- **资源浪费与监控缺失**：难以动态分配计算资源，常出现小任务占用大集群的情况，且模型上线后缺乏统一的生命周期追踪，故障排查耗时极长。\n- **协作效率低下**：数据版本、代码版本和模型版本分散管理，团队成员无法快速复现他人的实验结果，新成员上手成本极高。\n\n### 使用 mlrun 后\n- **无缝集成开发流**：mlrun 直接嵌入现有的 IDE 和 CI\u002FCD 环境，数据科学家提交的代码可自动转化为生产级管道，彻底消除环境与部署差异。\n- **自动化端到端管线**：通过 mlrun 定义统一的自动化工作流，自动完成数据预处理、训练及评估，数据源变动时管道自动触发重跑，稳定性大幅提升。\n- **弹性资源与全链路观测**：mlrun 根据任务负载动态调度计算资源，显著降低算力成本，并提供从数据血缘到模型在线服务的全生命周期监控面板。\n- **打破团队孤岛**：统一平台管理代码、数据和模型版本，任何成员均可一键复现实验，促进了数据、算法与运维团队的高效协作。\n\nmlrun 通过将碎片化的机器学习工程标准化和自动化，帮助团队将模型从实验到生产的交付周期缩短了 70%。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlrun_mlrun_48179110.png","MLRun","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fmlrun_53d1d971.png","MLRun is an open MLOps framework for quickly building and managing continuous ML and generative AI applications across their lifecycle.",null,"mlrun.org","https:\u002F\u002Fgithub.com\u002Fmlrun",[83,87,91,95,99,102,106,109],{"name":84,"color":85,"percentage":86},"Python","#3572A5",98.2,{"name":88,"color":89,"percentage":90},"Go","#00ADD8",0.9,{"name":92,"color":93,"percentage":94},"Makefile","#427819",0.4,{"name":96,"color":97,"percentage":98},"Dockerfile","#384d54",0.2,{"name":100,"color":101,"percentage":98},"Shell","#89e051",{"name":103,"color":104,"percentage":105},"Jupyter Notebook","#DA5B0B",0,{"name":107,"color":108,"percentage":105},"Mako","#7e858d",{"name":110,"color":111,"percentage":105},"HTML","#e34c26",1659,298,"2026-04-04T19:19:15","Apache-2.0",4,"未说明","非必需，但支持 GPU 加速（文档提及 GPU 利用相关章节），具体型号、显存及 CUDA 版本未说明",{"notes":120,"python":117,"dependencies":121},"MLRun 是一个编排平台，通常以客户端 - 服务器架构部署。核心组件包括项目管理服务、函数运行时（支持自动部署）、数据存储和特征存储。它深度集成 Nuclio 用于实时无服务器功能，并支持多种数据源和机器学习框架。具体的运行环境（如 Python 版本、OS）通常取决于所调用的具体函数容器或客户端安装需求，需参考官方设置指南（Set up your client environment）获取详细依赖信息。",[122,123],"nuclio (用于无服务器函数)","未说明其他具体库及版本",[51,54,15,13],[126,127,128,129,130,131,132,133,134,135],"mlops","python","data-science","machine-learning","data-engineering","experiment-tracking","model-serving","mlops-workflow","workflow","kubernetes","2026-03-27T02:49:30.150509","2026-04-06T05:16:42.237852",[139,144,149,154,159,163],{"id":140,"question_zh":141,"answer_zh":142,"source_url":143},9604,"在 Kubernetes 集群上运行工作流时遇到 Docker 镜像拉取或构建错误，如何解决？","需要同时设置 `DEFAULT_DOCKER_REGISTRY`（例如 `https:\u002F\u002Findex.docker.io\u002Fv1\u002F`）和 `DEFAULT_DOCKER_SECRET`（同一命名空间下的 Kubernetes Secret 名称）。另一种方法是为每个函数单独配置注册表，在函数对象中添加以下代码：`fn.build_config(image='target\u002Fimage:tag', secret='my_docker')`。如果问题仍然存在，可能是开源版本中 Pipeline 区域的近期变更导致的临时性问题，建议关注官方修复更新。","https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fmlrun\u002Fissues\u002F711",{"id":145,"question_zh":146,"answer_zh":147,"source_url":148},9605,"在本地 Minikube 运行工作流时出现 'Name or service not known' 连接错误怎么办？","该错误通常是因为无法解析 `ml-pipeline.default.svc.cluster.local` 服务地址。请确保已按照官方文档正确安装 Kubeflow Pipelines 组件，并且 ml-pipeline 服务在对应的命名空间中正常运行。如果是单用户模式或非标准安装，可能需要手动指定正确的命名空间或调整服务发现配置。此外，检查 Nuclio 安装时的环境变量配置，特别是关于外部地址和 Ingress 的设置，可以在 `nuclio-dashboard` Deployment 中修改相关环境变量，或在新的安装中通过 Nuclio Helm values 进行配置。","https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fmlrun\u002Fissues\u002F244",{"id":150,"question_zh":151,"answer_zh":152,"source_url":153},9606,"在 Mac (M1\u002FM2) 上使用 Python 3.9+ 安装 mlrun 时遇到 PyYAML 构建失败错误，如何解决？","这是 PyYAML 在 ARM64 架构上的已知问题（参考 https:\u002F\u002Fgithub.com\u002Fyaml\u002Fpyyaml\u002Fissues\u002F724）。解决方案有两种：1) 将 PyYAML 降级到 5.3 版本（`pip install pyyaml==5.3`），然后再安装 mlrun，并确保安装过程不会自动升级 PyYAML；2) 遵循 mlrun 官方贡献指南中关于 ARM64 机器开发的说明（https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fmlrun\u002Fblob\u002Fdevelopment\u002FCONTRIBUTING.md#developing-with-arm64-machines），这样可以正常使用 PyYAML 5.4.1 及以上版本。使用 PyEnv 管理 Python 版本通常不会导致此问题。","https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fmlrun\u002Fissues\u002F5470",{"id":155,"question_zh":156,"answer_zh":157,"source_url":158},9607,"在 Kubeflow 上运行 Pipeline 时出现 'Invalid resource references for experiment' 错误，原因是什么？","该错误表明实验的资源引用配置不正确，系统期望有一个带有所有者关系的命名空间类型，但实际获取为空数组 `[]`。这通常发生在命名空间配置错误或权限不足的情况下。请确认：1) 使用的命名空间是否正确（如从 `mlrun` 改为 `kubeflow`）；2) 该命名空间下是否存在有效的 Experiment 资源；3) 当前用户或服务账户是否有在该命名空间创建 Experiment 的权限。如果是在单用户模式下运行，可能不完全兼容，建议切换到多用户模式或检查 Kubeflow 的 RBAC 配置。","https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fmlrun\u002Fissues\u002F1479",{"id":160,"question_zh":161,"answer_zh":162,"source_url":148},9608,"如何配置 Nuclio Dashboard 的外部访问地址和 Ingress？","Nuclio Dashboard 的外部地址和 Ingress 配置是通过环境变量设置的，这些变量告诉 Nuclio 如何暴露其服务。对于已有安装，可以编辑 `nuclio-dashboard` 的 Deployment 资源，修改相应的环境变量值。对于新安装，推荐在使用 Helm 部署 Nuclio 时，直接在 `values.yaml` 文件中配置相关参数，这样更易于管理和维护。具体变量名和格式可参考 Nuclio 官方文档或 mlrun 的安装指南。",{"id":164,"question_zh":165,"answer_zh":166,"source_url":143},9609,"工作流执行时报错 'open \u002Ftmp\u002Fmymodel: no such file or directory'，可能的原因是什么？","此错误表示容器内尝试读取 `\u002Ftmp\u002Fmymodel` 文件但该文件不存在。常见原因包括：1) 前序步骤未正确生成或输出该文件；2) 文件路径配置错误，实际输出路径与预期不符；3) 容器间共享卷（volume）未正确挂载，导致文件无法传递。解决方法：检查 Pod 的 YAML 定义，确认 `volumeMounts` 和 `volumes` 配置是否正确；查看前序任务的日志，验证文件是否成功生成并写入指定路径；确保所有相关容器都挂载了相同的持久化卷（PVC）以共享数据。",[168,173,178,183,188,193,198,203,208,213,218,223,228,233,238,243,248,253,258,263],{"id":169,"version":170,"summary_zh":171,"released_at":172},106929,"v1.11.0-rc48","### Features \u002F Enhancements\n* **Mounts**: Explicit auto_mount_type config takes precedence over mlrun_pvc_mount env var [1.11.x], #9531, @alxtkr77\n* **Tutorials**: Increase timeout and replace deprecated parameter [1.11.x], #9510, @EdmondIguazio\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc48#features-and-enhancements)\n\n### Bug fixes\n* **Auth**: Fix project-summaries endpoint bypassing authorization filter [1.11.x], #9529, @liranbg\n* **Auth**: Fix project-summaries endpoint bypassing authorization filter [1.11.x], #9529, @liranbg\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc48#bug-fixes)\n\n\n#### Pull requests:\nf4b854eac [Mounts] Explicit auto_mount_type config takes precedence over MLRUN_PVC_MOUNT env var [1.11.x] (#9531)\n9be50e305 [Auth] Fix project-summaries endpoint bypassing authorization filter [1.11.x] (#9529)\n390bec2a1 [Auth] Fix project-summaries endpoint bypassing authorization filter [1.11.x] (#9529)\n5503981e2 [Tutorials] Increase timeout and replace deprecated parameter [1.11.x] (#9510)","2026-03-31T21:46:33",{"id":174,"version":175,"summary_zh":176,"released_at":177},106930,"v1.11.0-rc47","### Features \u002F Enhancements\n* **Auth**: Return resolved username in delete secret token responses, #9508, @elbamit\n* **Auth**: Persist user_id for schedules and retries, #9502, @elbamit\n* **Alerts**: Split other_alerts_count into application and infra counters, #9488, @alxtkr77\n* **Unknown**: Revert per-project kafka consumer group (#9378), #9504, @alxtkr77\n* **Mounts**: Support s3 and secret_env auto mount types in client-side auto_mount(), #9485, @alxtkr77\n* **CI**: Install aiohttp and click on data node before enterprise cleanup, #9499, @Yacouby\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc47#features-and-enhancements)\n\n### Bug fixes\n* **Tutorials**: Tutorials fixes, #9500, @EdmondIguazio\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc47#bug-fixes)\n\n\n#### Pull requests:\n1ac84e919 [Auth] Return resolved username in delete secret token responses (#9508)\n3810f8465 [Auth] Persist user_id for schedules and retries (#9502)\ncba0b02ac [Alerts] Split other_alerts_count into application and infra counters (#9488)\na6dc079ad [Tutorials] Tutorials fixes (#9500)\nfe2df6739 Revert per-project Kafka consumer group (#9378) (#9504)\nf8b4e61dd [Mounts] Support S3 and secret_env auto mount types in client-side auto_mount() (#9485)\n5ff0f83c2 [CI] Install aiohttp and click on data node before enterprise cleanup (#9499)","2026-03-27T12:46:13",{"id":179,"version":180,"summary_zh":181,"released_at":182},106931,"v1.11.0-rc46","### Features \u002F Enhancements\n* **Smoke-test**: Install yq for k3s cluster setup, #9492, @assaf758\n* **Docs**: Update components in ce install, #9408, @jillnogold\n* **Docs**: Add details about direct_port_access=true in application runtime, #9495, @jillnogold\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc46#features-and-enhancements)\n\n### Bug fixes\n* **API**: Fix project-summaries endpoint bypassing project permissions, #9498, @liranbg\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc46#bug-fixes)\n\n\n#### Pull requests:\n552daf6fc [Smoke-test] Install yq for k3s cluster setup (#9492)\n68883856c [Docs] Update components in CE install (#9408)\n38bdee3b0 [API] Fix project-summaries endpoint bypassing project permissions (#9498)\ncf07566d1 [Docs] Add details about direct_port_access=True in application runtime (#9495)","2026-03-25T10:52:59",{"id":184,"version":185,"summary_zh":186,"released_at":187},106932,"v1.11.0-rc44","### Features \u002F Enhancements\n* **Requirements**: Bump storey ot 1.11.24, #9487, @tomerm-iguazio\n* **Model Monitoring**: Add configurable absolute hpa target for stream function, #9478, @alxtkr77\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc44#features-and-enhancements)\n\n### Bug fixes\n* **Chief**: Fix nonetype iteration error in pod container status resolution, #9479, @elbamit\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc44#bug-fixes)\n\n\n#### Pull requests:\n40cb24e07 [Chief] Fix NoneType iteration error in pod container status resolution (#9479)\ncd1e9196f [Requirements] Bump storey ot 1.11.24 (#9487)\n9d8226142 [Model Monitoring] Add configurable absolute HPA target for stream function (#9478)","2026-03-23T23:14:57",{"id":189,"version":190,"summary_zh":191,"released_at":192},106933,"v1.11.0-rc42","### Features \u002F Enhancements\n* **Requirements**: Bump storey 1.11.22, #9443, @royischoss\n* **Auto-mount**: Use semicolon instead of comma as a delimeter for secret_env type, #9439, @rokatyy\n* **Docs**: Add config default references for lag detection parameters, #9437, @alxtkr77\n* **Docs**: Add with_sidecar  to nuclio page, #9436, @jillnogold\n* **Application**: Skip directory sources in local file validation, #9434, @Yacouby\n* **Project**: Add mgmt permission check for project owner changes, #9418, @Yacouby\n* **Docs**: Document model monitoring lag detection (ml-11648), #9428, @alxtkr77\n* **Serving**: Block custom handler usage in jobs from serving functions, #9410, @royischoss\n* **Application**: Validate local source file exists before deploy, #9429, @Yacouby\n* **Python**: Bump pip version, #9426, @liranbg\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc42#features-and-enhancements)\n\n### Bug fixes\n* **Serving**: Fix _mappedbody unpacking bypassed in async engine, #9425, @jond01\n* **Serving**: Fix - enforce step max_iterations on cyclic graph, #9441, @royischoss\n* **Tests**: Fix async mode system tests, #9430, @royischoss\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc42#bug-fixes)\n\n\n#### Pull requests:\n54abe01b8 [Serving] Fix _MappedBody unpacking bypassed in async engine (#9425)\ne28e1b737 [Requirements] Bump storey 1.11.22 (#9443)\nef167db8d [Auto-mount] Use semicolon instead of comma as a delimeter for secret_env type (#9439)\n991dbfa66 [Serving] Fix - enforce step max_iterations on cyclic graph (#9441)\n6ab8f0d99 [Docs] Add config default references for lag detection parameters (#9437)\nfd2099ff3 [Docs] Add with_sidecar  to Nuclio page (#9436)\n39782bc48 [Tests] Fix async mode system tests (#9430)\n736017a80 [Application] Skip directory sources in local file validation (#9434)\n97c9b4e08 [Project] Add mgmt permission check for project owner changes (#9418)\n7f3d3d4ae [Docs] Document model monitoring lag detection (ML-11648) (#9428)\n17a36bf16 [Serving] Block custom handler usage in jobs from serving functions (#9410)\n5e6a49fdd [Application] Validate local source file exists before deploy (#9429)\nce8abf116 [Python] Bump pip version (#9426)","2026-03-12T18:34:27",{"id":194,"version":195,"summary_zh":196,"released_at":197},106934,"v1.11.0-rc41","### Features \u002F Enhancements\n* **Dependabot-automated**: Bump manusa\u002Factions-setup-minikube from 2.15.0 to 2.16.0, #9421, @dependabot[bot]\n* **Model Monitoring**: Bump storey to 1.11.21 and add predictions uniqueness, #9424, @alxtkr77\n* **Model Monitoring**: Optimize timescaledb storage for app_results table, #9237, @alxtkr77\n* **CI**: Update uv version to 0.10.9, #9420, @liranbg\n* **Dockerfiles**: Remove setuptools 82 constraint, #9419, @liranbg\n* **CI**: Adding welcome and readme images [docs], #9414, @daniels290813\n* **Auth**: Skip invalid tokens during sync, #9396, @elbamit\n* **CI**: Add claude code commands for system test workflows, #9416, @liranbg\n* **Doc**: Serving function example with kafka queue and nuclio function, #9391, @guylei-code\n* **Requirements-automated**: Upgrade lock files, #9413, @iguazio-cicd\n* **Dependabot-automated**: Bump docker\u002Flogin-action from 3.7.0 to 4.0.0, #9407, @dependabot[bot]\n* **Dependabot-automated**: Bump google.golang.org\u002Fgrpc from 1.79.1 to 1.79.2 in \u002Fserver\u002Fgo, #9411, @dependabot[bot]\n* **Docs**: Add cyclic graph, #9142, @jillnogold\n* **Application**: Add workdir support for pull-at-runtime, #9400, @Yacouby\n* **Docs**: Delete the feature store demo from the docs, #9406, @jillnogold\n* **CI**: Adjust post-coverage-comment for branch that created from fork repo, #9405, @tomerm-iguazio\n* **Requirements**: Bump storey to 1.11.20, #9403, @tomerm-iguazio\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc41#features-and-enhancements)\n\n### Bug fixes\n* **Alerts**: Fix wildcard alert activation storing \"*\" as entity_id, #9417, @alxtkr77\n* **CI**: Fix system tests, #9415, @liranbg\n* **Model Monitoring**: Fix timescaledb pool hanging 120s on bad credentials, #9409, @alxtkr77\n* **Nuclio**: Fix jsondecodeerror when invoking function with head method, #9404, @jond01\n* **Launcher**: Fix unexpected `local` argument, #9402, @liranbg\n* **CI**: Fix secret system test, #9401, @liranbg\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc41#bug-fixes)\n\n\n#### Pull requests:\n31ae38933 [Dependabot-automated] Bump manusa\u002Factions-setup-minikube from 2.15.0 to 2.16.0 (#9421)\nf9ff10ec8 [Model Monitoring] Bump storey to 1.11.21 and add predictions uniqueness (#9424)\n6b64c0fc2 [Alerts] Fix wildcard alert activation storing \"*\" as entity_id (#9417)\n41d273170 [Model Monitoring] Optimize TimescaleDB storage for app_results table (#9237)\n32e0e4419 [CI] Update uv version to 0.10.9 (#9420)\n000b7e1a0 [Dockerfiles] Remove setuptools 82 constraint (#9419)\n6f01a38bf [CI] Adding welcome and readme images [Docs] (#9414)\nd3461ca61 [Auth] Skip invalid tokens during sync (#9396)\ne05b3c2a7 [CI] Add Claude Code commands for system test workflows (#9416)\n5a6be27ec [CI] Fix system tests (#9415)\n9e6b07351 [Model Monitoring] Fix TimescaleDB pool hanging 120s on bad credentials (#9409)\n35bc0b885 [Nuclio] Fix JSONDecodeError when invoking function with HEAD method (#9404)\n7f9fa1bf0 [Doc] Serving function example with Kafka queue and Nuclio function (#9391)\n586207aac [Requirements-automated] Upgrade lock files (#9413)\n0cee37e62 [Dependabot-automated] Bump docker\u002Flogin-action from 3.7.0 to 4.0.0 (#9407)\n95f5664f8 [Dependabot-automated] Bump google.golang.org\u002Fgrpc from 1.79.1 to 1.79.2 in \u002Fserver\u002Fgo (#9411)\n99e0fa364 [Docs] Add cyclic graph (#9142)\nc8a3a3b48 [Application] Add workdir support for pull-at-runtime (#9400)\nd8f7fa014 [Docs] Delete the Feature store demo from the docs (#9406)\n3b7b8db96 [CI] Adjust post-coverage-comment for branch that created from fork repo (#9405)\nd9b60ffc5 [Requirements] Bump storey to 1.11.20 (#9403)\n17f29fa2d [Launcher] Fix unexpected `local` argument (#9402)\nc36f62de1 [CI] Fix secret system test (#9401)","2026-03-09T16:40:32",{"id":199,"version":200,"summary_zh":201,"released_at":202},106935,"v1.11.0-rc40","### Features \u002F Enhancements\n* **Projects**: Control follower execution order on project deletion, #9398, @yaelgen\n* **Auth**: Change error type for missing auth token during token enrichment, #9395, @elbamit\n* **KFP**: Enhance pipeline run deletion with retry logic for transient kfp errors, #9393, @liranbg\n* **Serving**: Remove `context.current_event` that prevented last stream offset from being committed, #9392, @gtopper\n* **Serving**: Add  `remotefunctionstep` graph step, #8741, @daniels290813\n* **CI**: Add coverage comment in pr page, #9364, @tomerm-iguazio\n* **CI**: Remove legacy system-test devutils, #9394, @liranbg\n* **Docs**: Add datastore profiles page, #9222, @jillnogold\n* **Dependabot-automated**: Bump astral-sh\u002Fsetup-uv from 7.2.1 to 7.3.1, #9388, @dependabot[bot]\n* **Dependabot-automated**: Bump the kubernetes group across 1 directory with 3 updates, #9389, @dependabot[bot]\n* **Dependabot-automated**: Bump anchore\u002Fscan-action from 7.3.1 to 7.3.2, #9281, @dependabot[bot]\n* **Dependabot-automated**: Bump aws-actions\u002Fconfigure-aws-credentials from 5.1.1 to 6.0.0, #9282, @dependabot[bot]\n* **Serving**: Detect body_map vs path template param name conflicts at init time, #9382, @jond01\n* **Dependabot-automated**: Bump actions\u002Fupload-artifact from 6 to 7, #9385, @dependabot[bot]\n* **Dependabot-automated**: Bump actions\u002Fdownload-artifact from 7 to 8, #9386, @dependabot[bot]\n* **Requirements-automated**: Upgrade lock files, #9387, @iguazio-cicd\n* **Requirments**: Revert storey to 1.11.16 (mm \u002F test_app_flow is failing), #9384, @assaf758\n* **Requirements**: Bump storey to 1.11.18, #9380, @tomerm-iguazio\n* **Application runtime**: Validate sidecar probes in client side, #9374, @weilerN\n* **Testing**: Ga-workflow system-tests-opensource availble for pr smoketests, setting \"smoke tests: pass\" label on success., #9361, @assaf758\n* **Requirements-automated**: Upgrade lock files, #9381, @iguazio-cicd\n* **Auto-mount**: Add new auto-mount type secret_env, #9372, @rokatyy\n* **CE**: Support flows where mlrun is installed without spark\u002Fdask and others, #9371, @liranbg\n* **Docs**: Update serving graph coverage, #9252, @jillnogold\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc40#features-and-enhancements)\n\n### Bug fixes\n* **Tests**: Fix `testservingapihandler::test_api_handler_forbidden`, #9397, @gtopper\n* **Docs**: Fix delete_runtime_resources docstring to reflect actual authorization behavior, #9369, @yaelgen\n* **SDK**: Fix warning on launcher consumed arguments, #9373, @liranbg\n* **Package**: Bug fixes and logging kwargs, #9377, @guy1992l\n* **Secrets**: Fix auth secret username setting when running kfp pipeline, #9379, @rokatyy\n* **Docs**: Fix filename in serving graph, #9375, @jillnogold\n* **Model Monitoring**: Fix wrong kwarg in delete_model_monitoring_lag_alert, #9376, @alxtkr77\n* **Serving**: Fix lost outlet step in child function graph, #9370, @royischoss\n* **Testing**: Ruff formatting fixes, #9368, @assaf758\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc40#bug-fixes)\n\n\n#### Pull requests:\n1262e19aa [Projects] Control follower execution order on project deletion (#9398)\n9785e61c7 [Auth] Change error type for missing auth token during token enrichment (#9395)\n9cddaa6c0 [KFP] Enhance pipeline run deletion with retry logic for transient KFP errors (#9393)\nb1fe99db2 [Serving] Remove `context.current_event` that prevented last stream offset from being committed (#9392)\nf75599434 [Serving] Add  `RemoteFunctionStep` Graph Step (#8741)\nf5bd4f19f [CI] Add coverage comment in pr page (#9364)\n9dcdf8911 [Tests] Fix `TestServingAPIHandler::test_api_handler_forbidden` (#9397)\nfb4df9199 [CI] Remove legacy system-test devutils (#9394)\n77a631d0d [Docs] Add datastore profiles page (#9222)\n4f5aecec6 [Dependabot-automated] Bump astral-sh\u002Fsetup-uv from 7.2.1 to 7.3.1 (#9388)\n5d706a5a6 [Dependabot-automated] Bump the kubernetes group across 1 directory with 3 updates (#9389)\n5b9ecefad [Dependabot-automated] Bump anchore\u002Fscan-action from 7.3.1 to 7.3.2 (#9281)\na631e6392 [Dependabot-automated] Bump aws-actions\u002Fconfigure-aws-credentials from 5.1.1 to 6.0.0 (#9282)\n51cc3d489 [Serving] Detect body_map vs path template param name conflicts at init time (#9382)\n48c68da2c [Docs] Fix delete_runtime_resources docstring to reflect actual authorization behavior (#9369)\nfff235a08 [Dependabot-automated] Bump actions\u002Fupload-artifact from 6 to 7 (#9385)\nf40707996 [Dependabot-automated] Bump actions\u002Fdownload-artifact from 7 to 8 (#9386)\n1691d654d [Requirements-automated] Upgrade lock files (#9387)\n21328f6b0 [Requirments] Revert Storey to 1.11.16 (MM \u002F test_app_flow is failing) (#9384)\n64330c422 [Requirements] Bump storey to 1.11.18 (#9380)\nf2b9b564b [Application runtime] Validate sidecar probes in client side (#9374)\n138adf5a9 [Testing] GA-workflow system-tests-opensource availble for PR smoketests, setting \"Smoke tests: Pass\" label on success. (#9361)\n41480c734 [SDK] Fix warning on","2026-03-03T22:50:42",{"id":204,"version":205,"summary_zh":206,"released_at":207},106936,"v1.11.0-rc39","### Features \u002F Enhancements\n* **CI**: Upgrade ruff from 0.8.0 to 0.15.2, #9362, @jond01\n* **Requirements-automated**: Upgrade lock files, #9367, @iguazio-cicd\n* **Serving**: Add wildcard star path matching to api handler, #9358, @jond01\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc39#features-and-enhancements)\n\n### Bug fixes\n* **Pipelines**: Fix parameter name, #9366, @liranbg\n* **Model Monitoring**: Fix tracking for streaming `llmodel` responses, #9343, @gtopper\n* **FeatureStore**: Fix plot with_targets=true, #9359, @tomerm-iguazio\n* **IG4**: Fix http status code mapping in v4 client error handling, #9363, @yaelgen\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc39#bug-fixes)\n\n\n#### Pull requests:\na1ae24c0d [CI] Upgrade ruff from 0.8.0 to 0.15.2 (#9362)\na783c134b [Requirements-automated] Upgrade lock files (#9367)\nd7908116e [Pipelines] Fix parameter name (#9366)\n4ecf9a00a [Model Monitoring] Fix tracking for streaming `LLModel` responses (#9343)\n1ccd9f4e3 [Serving] Add wildcard star path matching to API handler (#9358)\nbf24f5b19 [FeatureStore] Fix plot with_targets=True (#9359)\nac9a8e0f1 [IG4] Fix HTTP status code mapping in v4 client error handling (#9363)","2026-02-24T22:49:01",{"id":209,"version":210,"summary_zh":211,"released_at":212},106937,"v1.11.0-rc38","### Features \u002F Enhancements\n* **K8s**: Enhance kubernetes api timeout handling, #9356, @liranbg\n* **Summary**: Default summaries to 0 if not yet calculated, #9360, @rokatyy\n* **Application**: Support any file extension in for single-file source, #9357, @Yacouby\n* **RemoteModels**: Add note for slow internet connection with hf, #9334, @tomerm-iguazio\n* **Requirements-automated**: Upgrade lock files, #9353, @liranbg\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc38#features-and-enhancements)\n\n### Bug fixes\n* **Alembic**: Fix background_task_labels migration failure, #9344, @Yacouby\n* **Config**: Fix set_env_from_file values overridden by ~\u002F.mlrun.env, #9354, @liranbg\n* **Alerts**: Fix docs to reflect auto reset behavior, #9355, @Yacouby\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc38#bug-fixes)\n\n\n#### Pull requests:\n04fa1443e [K8s] Enhance Kubernetes API timeout handling (#9356)\nc54431ad8 [Summary] Default summaries to 0 if not yet calculated (#9360)\nc550a7b1a [Application] Support any file extension in for single-file source (#9357)\n1ad7ac9eb [Alembic] Fix background_task_labels migration failure (#9344)\n4f2315770 [Config] Fix set_env_from_file values overridden by ~\u002F.mlrun.env (#9354)\nab58afe75 [Alerts] Fix docs to reflect auto reset behavior (#9355)\ne294f6d29 [RemoteModels] Add note for slow internet connection with hf (#9334)\n48b739ae8 [Requirements-automated] Upgrade lock files (#9353)","2026-02-23T23:47:28",{"id":214,"version":215,"summary_zh":216,"released_at":217},106938,"v1.11.0-rc37","### Features \u002F Enhancements\n* **Tests**: Add system test for streaming error termination, #9345, @gtopper\n* **Serving**: Add path and query parameter support to api handler, #9347, @jond01\n* **Application**: Improve single-file source artifact user experience, #9346, @Yacouby\n* **Requirements-automated**: Upgrade lock files, #9314, @iguazio-cicd\n* **Artifact**: Enrich artifact owner on be, #9350, @rokatyy\n* **Package**: Docs and pack \u002F unpack artifact types separation, #9321, @guy1992l\n* **Model Providers**: Add streaming support, #9335, @gtopper\n* **Requirements**: Bump locked requirements, #9341, @liranbg\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc37#features-and-enhancements)\n\n### Bug fixes\n* **Serving**: Fix remove error raise step from function.spec.graph, #9340, @royischoss\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc37#bug-fixes)\n\n\n#### Pull requests:\n2cb3f3619 [Tests] Add system test for streaming error termination (#9345)\nfb16e0b0c [Serving] Add path and query parameter support to API handler (#9347)\n83722d42a [Application] Improve single-file source artifact user experience (#9346)\n4926c92bf [Requirements-automated] Upgrade lock files (#9314)\nf5ee44cac [Artifact] Enrich artifact owner on BE (#9350)\n69ddd011f [Package] Docs and Pack \u002F Unpack Artifact Types Separation (#9321)\n8fa2aacb9 [Serving] Fix remove error raise step from function.spec.graph (#9340)\nf5643049d [Model Providers] Add streaming support (#9335)\n14f9c5e3f [Requirements] Bump locked requirements (#9341)","2026-02-22T15:47:26",{"id":219,"version":220,"summary_zh":221,"released_at":222},106939,"v1.11.0-rc36","### Features \u002F Enhancements\n* **Auth**: Add bulk token deletion endpoint for user cleanup, #9326, @yaelgen\n* **System tests**: Add retry to new application runtime system tests, #9338, @Yacouby\n* **Auth**: Infer auth info from selected user auth token, #9336, @liranbg\n* **Tests**: Drop timing assertion in new system test, #9337, @gtopper\n* **Run**: Warn on unexpected run kwargs and suggest typos, #9331, @liranbg\n* **Model Monitoring**: Add sdk methods and wildcard cache for lag detection alerts, #9315, @alxtkr77\n* **Serving**: Support proper http error codes, #9325, @jond01\n* **Spark**: Mount auth token secret to driver and executor pods, #9327, @liranbg\n* **Client**: Per thread client and dummy cookie jar, #9320, @rokatyy\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc36#features-and-enhancements)\n\n### Bug fixes\n* **RemoteRuntime**: Fix adding http trigger mode for sync, #9332, @royischoss\n* **Serving**: Fix asyncio execution mechanism in streaming models, #9333, @gtopper\n* **Tests**: Fix mrs batch step system test, #9330, @tomerm-iguazio\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc36#bug-fixes)\n\n\n#### Pull requests:\nc2ac0b1ac [RemoteRuntime] Fix adding http trigger mode for sync (#9332)\n2a77fd291 [Auth] Add bulk token deletion endpoint for user cleanup (#9326)\n2a0f3c00d [System tests] Add retry to new application runtime system tests (#9338)\nc9bfb1ccf [Auth] Infer auth info from selected user auth token (#9336)\n8064006e4 [Tests] Drop timing assertion in new system test (#9337)\nbb8c400e5 [Run] Warn on unexpected run kwargs and suggest typos (#9331)\n742fef252 [Model Monitoring] Add SDK methods and wildcard cache for lag detection alerts (#9315)\n87694a226 [Serving] Fix asyncio execution mechanism in streaming models (#9333)\n9faa35920 [Tests] Fix mrs batch step system test (#9330)\n58f742833 [Serving] Support proper HTTP error codes (#9325)\n3a34759b1 [Spark] Mount auth token secret to driver and executor pods (#9327)\nc9826b522 [Client] Per thread client and dummy cookie jar (#9320)","2026-02-18T17:58:47",{"id":224,"version":225,"summary_zh":226,"released_at":227},106940,"v1.11.0-rc35","### Features \u002F Enhancements\n* **Docs**: Add 1.10.2 to change log, #9329, @jillnogold\n* **Model Monitoring**: Preserve auth token when updating controller, #9323, @alxtkr77\n* **Dependabot-automated**: Bump manusa\u002Factions-setup-minikube from 2.14.0 to 2.15.0, #9318, @dependabot[bot]\n* **Docs**: Add load single code file and pull at runtime, #9300, @jillnogold\n* **Tests**: Update nuclio system test, #9319, @gtopper\n* **Model Monitoring**: Add support for streaming mrs, #9278, @gtopper\n* **Serving**: Add jsonpath body mapping support for api handler, #9311, @jond01\n* **Package**: Added `loghint` model to pass in a function's `returns`, #9294, @guy1992l\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc35#features-and-enhancements)\n\n### Bug fixes\n* **docs**: Fix the issue build_image pod consumes too much ram and being evicted, #9296, @xsqian\n* **API**: Fix ig4 project delete strategy handling, #9324, @quaark\n* **Model Monitoring**: Fix drift histogram over-counting with timescaledb, #9284, @alxtkr77\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc35#bug-fixes)\n\n\n#### Pull requests:\ncbd27ce31 [Docs] Add 1.10.2 to change log (#9329)\n96c8046a8 [docs] fix the issue build_image pod consumes too much RAM and being evicted (#9296)\n96edbeee3 [Model Monitoring] Preserve auth token when updating controller (#9323)\n8d49e43f5 [API] Fix IG4 Project Delete Strategy Handling (#9324)\nf9f86fbf1 [Dependabot-automated] Bump manusa\u002Factions-setup-minikube from 2.14.0 to 2.15.0 (#9318)\n4d0ddf505 [Docs] Add load single code file and pull at runtime (#9300)\n84a722064 [Model Monitoring] Fix drift histogram over-counting with TimescaleDB (#9284)\nab190bf69 [Tests] Update Nuclio system test (#9319)\nd301c69f4 [Model Monitoring] Add support for streaming MRS (#9278)\n1c6cfcd88 [Serving] Add JSONPath body mapping support for API handler (#9311)\n0b566d4b6 [Package] Added `LogHint` model to pass in a function's `returns` (#9294)","2026-02-17T09:57:39",{"id":229,"version":230,"summary_zh":231,"released_at":232},106941,"v1.11.0-rc34","### Features \u002F Enhancements\n* **Model Monitoring**: Wire lag detection params through sdk\u002Fapi\u002Fcrud, #9285, @alxtkr77\n* **Application**: Delete source artifacts on function deletion, #9312, @Yacouby\n* **Dependabot-automated**: Bump google.golang.org\u002Fgrpc from 1.78.0 to 1.79.1 in \u002Fserver\u002Fgo, #9313, @dependabot[bot]\n* **RemoteModels**: Batch step error handling, #9280, @tomerm-iguazio\n* **Application**: Support source archives & git in init container, #9306, @Yacouby\n* **Application**: Change default_worker_number to 100, #9310, @rokatyy\n* **Authentication**: Create a runtimeconfigurationcontext for passing auth token name to runtimes, #9240, @elbamit\n* **Render**: Added support for nested inputs and outputs, #9308, @guy1992l\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc34#features-and-enhancements)\n\n### Bug fixes\n* **Tests**: Fix test_mep_with_remote_model, #9309, @tomerm-iguazio\n* **API**: Fix missing project permission checks, #9303, @quaark\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc34#bug-fixes)\n\n\n#### Pull requests:\n63a39f271 [Model Monitoring] Wire lag detection params through SDK\u002FAPI\u002FCRUD (#9285)\nac9093314 [Application] Delete source artifacts on function deletion (#9312)\n443e648c3 [Dependabot-automated] Bump google.golang.org\u002Fgrpc from 1.78.0 to 1.79.1 in \u002Fserver\u002Fgo (#9313)\n308771058 [RemoteModels] Batch step error handling (#9280)\n045aadae0 [Application] Support source archives & Git in init container (#9306)\nd58f41401 [Tests] Fix test_mep_with_remote_model (#9309)\n863680c93 [Application] Change default_worker_number to 100 (#9310)\n23a7583b3 [API] Fix Missing Project Permission Checks (#9303)\n9b6993be6 [Authentication] Create a RuntimeConfigurationContext for passing auth token name to runtimes (#9240)\ndf3c642a3 [Render] Added support for nested inputs and outputs (#9308)","2026-02-15T14:39:21",{"id":234,"version":235,"summary_zh":236,"released_at":237},106942,"v1.11.0-rc33","### Features \u002F Enhancements\n* **Dependabot-automated**: Bump the kubernetes group across 1 directory with 3 updates, #9301, @dependabot[bot]\n* **Go**: Bump go version to 1.25.7, #9302, @Copilot\n* **Package**: Multiple `handler` decorators won't harm the run, #9295, @guy1992l\n* **Requirements**: Bump storey 1.11.13, #9298, @tomerm-iguazio\n* **Docs**: Add and improve ce upgrade docs, #9276, @GiladShapira94\n* **Docs**: Adjust linkcheck configuration, #9297, @rokatyy\n* **Requirements-automated**: Upgrade lock files, #9288, @iguazio-cicd\n* **Requirements**: Remove obsoleted conda install commands, #9287, @liranbg\n* **Docs**: Add rabbitmq, #9204, @jillnogold\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc33#features-and-enhancements)\n\n### Bug fixes\n* **RemoteModels**: Fix batch_input_data import, #9307, @tomerm-iguazio\n* **Build**: Bump mpi4py~=3.1 to ~=4.1 to fix setuptools>=81 build failure, #9305, @jond01\n* **Serving**: Api handler fixes, #9286, @jond01\n* **Build**: Fix pkg_resources breakage from setuptools 82, #9292, @jond01\n* **Security**: Potential fix for code scanning alerts, #9289, @liranbg\n* **CI**: Fix broken azure dataframe urls, #9290, @liranbg\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc33#bug-fixes)\n\n\n#### Pull requests:\n04f2250fc [RemoteModels] fix BATCH_INPUT_DATA import (#9307)\nb455f7896 [Build] Bump mpi4py~=3.1 to ~=4.1 to fix setuptools>=81 build failure (#9305)\n928406b80 [Dependabot-automated] Bump the kubernetes group across 1 directory with 3 updates (#9301)\n1fcb385ca [Go] Bump Go version to 1.25.7 (#9302)\nc62954ab4 [Package] Multiple `handler` decorators won't harm the run (#9295)\n3e859d103 [Requirements] Bump storey 1.11.13 (#9298)\n25d0bba7d [Docs] Add and Improve CE upgrade docs (#9276)\n8f653fd0e [Docs] Adjust linkcheck configuration (#9297)\nd46567480 [Serving] API handler fixes (#9286)\n4ae9e9e2f [Build] Fix pkg_resources breakage from setuptools 82 (#9292)\nbe8be2263 [Security] Potential fix for code scanning alerts (#9289)\na3f568d0e [Requirements-automated] Upgrade lock files (#9288)\nd1cc4ce85 [CI] Fix broken azure dataframe urls (#9290)\nd78a8e1d9 [Requirements] Remove obsoleted conda install commands (#9287)\nd84eb135d [Docs] Add RabbitMQ  (#9204)","2026-02-11T19:20:18",{"id":239,"version":240,"summary_zh":241,"released_at":242},106943,"v1.10.2","### Features \u002F Enhancements\r\n* **Client**: Per thread client instance and dummy cookie jar, #9258, @rokatyy\r\n* **Version**: Bump 1.10.x to 1.10.2, #9230, @liranbg\r\n* **Docs**: Backport refactored ce install [1.10.x], #9232, @jillnogold\r\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.10.2#features-and-enhancements)\r\n\r\n### Bug fixes\r\n* **CI**: Fix building images [1.10.x], #9304, @liranbg\r\n* **CI**: Fix build [1.10.x], #9299, @liranbg\r\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.10.2#bug-fixes)\r\n\r\n\r\n#### Pull requests:\r\nc4199f542 [CI] Fix building images [1.10.x] (#9304)\r\n2156f413a [CI] Fix build [1.10.x] (#9299)\r\na839ba858 [Client] Per thread client instance and dummy cookie jar (#9258)\r\n5db7d9f17 [CI] get_demos use 1.10.x branch instead of development as quick fix\r\nfad1bba2b [CI] Private release demo for 1.10.2\r\nb418427ca [Version] Bump 1.10.x to 1.10.2 (#9230)\r\nfe85c7c32 [CI] Private release improvements (#9231) - backport to 1.10.x\r\ne0a0cbd9b [CI] Ubuntu private runners and skopeo for private releases\r\n5c0359a1c [Docs] Backport refactored CE install [1.10.x] (#9232)\r\n\r\n#### Failed parsing:\r\n5db7d9f17 {Alessandro Pio Ardizio} [CI] get_demos use 1.10.x branch instead of development as quick fix\r\nfad1bba2b {Alessandro Pio Ardizio} [CI] Private release demo for 1.10.2\r\nfe85c7c32 {Alessandro Pio Ardizio} [CI] Private release improvements (#9231) - backport to 1.10.x\r\ne0a0cbd9b {Alessandro Pio Ardizio} [CI] Ubuntu private runners and skopeo for private releases","2026-02-16T10:53:18",{"id":244,"version":245,"summary_zh":246,"released_at":247},106944,"v1.11.0-rc32","### Features \u002F Enhancements\n* **Package**: Support collection as input, #9190, @guy1992l\n* **Model Monitoring**: Add writer lag detection events and graph step, #9272, @alxtkr77\n* **Application**: Support skipping image build for source changes, #9279, @Yacouby\n* **Auth**: Ensure owner is derived from authentication object, #9273, @liranbg\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc32#features-and-enhancements)\n\n### Bug fixes\n* **Serving**: Fix mlrunapiremotestep authorization on ig4 sys, #9283, @davesh0812\n* **Model Monitoring**: Fix incorrect json fallback for v2+ writer (ml-11979), #9221, @alxtkr77\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc32#bug-fixes)\n\n\n#### Pull requests:\nbcd67f693 [Serving] Fix MLRunAPIRemoteStep authorization on IG4 sys (#9283)\n1f7ee3ca8 [Package] Support Collection as Input (#9190)\nf76b6f97a [Model Monitoring] Add writer lag detection events and graph step (#9272)\n2eb4f95f1 [Application] Support skipping image build for source changes (#9279)\ne825d11d7 [Model Monitoring] Fix incorrect JSON fallback for v2+ writer (ML-11979) (#9221)\n4b401450c [Auth] Ensure owner is derived from authentication object (#9273)","2026-02-05T15:27:30",{"id":249,"version":250,"summary_zh":251,"released_at":252},106945,"v1.11.0-rc31","### Features \u002F Enhancements\n* **Serving**: Add a basic api handler option to the graph, #9262, @jond01\n* **Requirements-automated**: Upgrade lock files, #9269, @iguazio-cicd\n* **Application**: Add init container config for source loading, #9267, @Yacouby\n* **Requirements**: Install iguazio sdk from pypi, #9266, @TomerShor\n* **Projects**: Prevent load_project failure for non-owner users, #9271, @yaelgen\n* **Serving**: Support mrs with batch step, #9264, @tomerm-iguazio\n* **Serving**: Add support for streaming models in mrs, #9263, @gtopper\n* **Requirements**: Bump storey to 1.11.12, #9270, @tomerm-iguazio\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc31#features-and-enhancements)\n\n### Bug fixes\n* **Serving**: Fix `onlinevectorservice` following changes in #9248, #9277, @gtopper\n* **Serving**: Fix error in async engine with no `respond()`, #9274, @gtopper\n* **Feature Store**: Fix ingest failing on ig4 due to missing v3io_access_key, #9268, @alxtkr77\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc31#bug-fixes)\n\n\n#### Pull requests:\n7d9199e3e [Serving] Fix `OnlineVectorService` following changes in #9248 (#9277)\n97dffec67 [Serving] Fix error in async engine with no `respond()` (#9274)\ndc1676d24 [Feature Store] Fix ingest failing on IG4 due to missing V3IO_ACCESS_KEY (#9268)\n60ecedc33 [Serving] Add a basic API handler option to the graph (#9262)\nf28363c53 [Requirements-automated] Upgrade lock files (#9269)\n9b13a0313 [Application] Add init container config for source loading (#9267)\n7c89d17c8 [Requirements] Install iguazio SDK from PyPI (#9266)\ne0814b1a2 [Projects] Prevent load_project failure for non-owner users (#9271)\n16db264df [Serving] Support MRS with batch step (#9264)\n57261d557 [Serving] Add support for streaming models in MRS (#9263)\n74ddfef1b [Requirements] Bump storey to 1.11.12 (#9270)","2026-02-05T10:11:30",{"id":254,"version":255,"summary_zh":256,"released_at":257},106946,"v1.11.0-rc30","### Features \u002F Enhancements\n* **Serving**: Improve error in case of a step without a `do()` method, #9226, @gtopper\n* **Requirements**: Bump storey, #9259, @gtopper\n* **Dependabot-automated**: Bump astral-sh\u002Fsetup-uv from 7.2.0 to 7.2.1, #9257, @dependabot[bot]\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc30#features-and-enhancements)\n\n### Bug fixes\n* **Authentication**: Fix wrong parameter order, #9261, @elbamit\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc30#bug-fixes)\n\n\n#### Pull requests:\ndbdcb36c6 [Authentication] Fix wrong parameter order (#9261)\n236395f5b [Serving] Improve error in case of a step without a `do()` method (#9226)\n5c2b1ac0a [Requirements] Bump storey (#9259)\n619918cbf [Dependabot-automated] Bump astral-sh\u002Fsetup-uv from 7.2.0 to 7.2.1 (#9257)","2026-02-01T20:10:06",{"id":259,"version":260,"summary_zh":261,"released_at":262},106947,"v1.11.0-rc29","### Features \u002F Enhancements\n* **Serving**: Implement support for stream responses, #9248, @gtopper\n* **Dependabot-automated**: Bump docker\u002Flogin-action from 3.6.0 to 3.7.0, #9253, @dependabot[bot]\n* **Auth**: Token resolution logic for runtimes, #9227, @elbamit\n* **Dependabot-automated**: Bump anchore\u002Fscan-action from 7.3.0 to 7.3.1, #9251, @dependabot[bot]\n* **Application**: Support single-file source deployment with artifacts, #9244, @Yacouby\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc29#features-and-enhancements)\n\n### Bug fixes\n* **Authz**: Fix attaching resources, #9254, @liranbg\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc29#bug-fixes)\n\n\n#### Pull requests:\n8185f6dfe [Authz] Fix attaching resources (#9254)\n9215a0d08 [Serving] Implement support for stream responses (#9248)\n9c089df71 [Dependabot-automated] Bump docker\u002Flogin-action from 3.6.0 to 3.7.0 (#9253)\n3935d15ac [Auth] Token resolution logic for runtimes (#9227)\n756224f85 [Dependabot-automated] Bump anchore\u002Fscan-action from 7.3.0 to 7.3.1 (#9251)\ndcebd3a61 [Application] Support single-file source deployment with artifacts (#9244)","2026-01-30T14:35:49",{"id":264,"version":265,"summary_zh":266,"released_at":267},106948,"v1.11.0-rc28","### Features \u002F Enhancements\n* **Docs**: Remove python 3.9, #9177, @jillnogold\n* **UI**: [Features & enhancement](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc28#features-and-enhancements)\n\n### Bug fixes\n* **Auth**: Fix service account configuration validation logic, #9250, @quaark\n* **UI**: [Bug fixes](https:\u002F\u002Fgithub.com\u002Fmlrun\u002Fui\u002Freleases\u002Ftag\u002Fv1.11.0-rc28#bug-fixes)\n\n\n#### Pull requests:\n458ba2783 [Auth] Fix Service Account Configuration Validation Logic (#9250)\n63611fb3b [Docs] Remove python 3.9 (#9177)","2026-01-28T13:04:18"]