[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-mlr-org--mlr3":3,"tool-mlr-org--mlr3":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",140436,2,"2026-04-05T23:32:43",[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":69,"readme_en":70,"readme_zh":71,"quickstart_zh":72,"use_case_zh":73,"hero_image_url":74,"owner_login":75,"owner_name":75,"owner_avatar_url":76,"owner_bio":77,"owner_company":78,"owner_location":78,"owner_email":78,"owner_twitter":78,"owner_website":79,"owner_url":80,"languages":81,"stars":90,"forks":91,"last_commit_at":92,"license":93,"difficulty_score":94,"env_os":95,"env_gpu":95,"env_ram":95,"env_deps":96,"category_tags":109,"github_topics":110,"view_count":117,"oss_zip_url":78,"oss_zip_packed_at":78,"status":16,"created_at":118,"updated_at":119,"faqs":120,"releases":148},1054,"mlr-org\u002Fmlr3","mlr3","mlr3: Machine Learning in R - next generation","mlr3 是一个基于 R 语言的机器学习框架，专注于通过面向对象的设计提升机器学习流程的效率和可维护性。它通过模块化的方式简化了数据预处理、模型训练、评估与部署等环节，为开发者提供了更清晰的代码结构和更灵活的扩展能力。相比传统方法，mlr3 降低了复杂任务的开发门槛，尤其适合需要快速迭代算法或整合多步骤工作流的场景。  \n\nmlr3 适合研究人员、数据科学家以及 R 语言开发者使用，其丰富的扩展包生态（如模型库、调参工具和可视化组件）能显著提升开发效率。核心亮点包括统一的接口设计、与 R 生态系统的深度集成，以及支持从基础算法到高级流水线的全链路构建。无论是学术研究还是工业应用，mlr3 都能提供稳定且可扩展的解决方案。","\n# mlr3 \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlr-org_mlr3_readme_82e14b4c5103.png\" align=\"right\" width = \"120\" \u002F>\n\nPackage website: [release](https:\u002F\u002Fmlr3.mlr-org.com\u002F) \\|\n[dev](https:\u002F\u002Fmlr3.mlr-org.com\u002Fdev\u002F)\n\nEfficient, object-oriented programming on the building blocks of machine\nlearning. Successor of [mlr](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr).\n\n\u003C!-- badges: start -->\n\n[![r-cmd-check](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Factions\u002Fworkflows\u002Fr-cmd-check.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Factions\u002Fworkflows\u002Fr-cmd-check.yml)\n[![DOI](https:\u002F\u002Fjoss.theoj.org\u002Fpapers\u002F10.21105\u002Fjoss.01903\u002Fstatus.svg)](https:\u002F\u002Fdoi.org\u002F10.21105\u002Fjoss.01903)\n[![CRAN\nStatus](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlr-org_mlr3_readme_e7f0a20dfe93.png)](https:\u002F\u002Fcran.r-project.org\u002Fpackage=mlr3)\n[![Mattermost](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fchat-mattermost-orange.svg)](https:\u002F\u002Flmmisld-lmu-stats-slds.srv.mwn.de\u002Fmlr_invite\u002F)\n\u003C!-- badges: end -->\n\n## Resources (for users and developers)\n\n- We have written a [book](https:\u002F\u002Fmlr3book.mlr-org.com\u002F). This should\n  be the central entry point to the package.\n- The [mlr-org website](https:\u002F\u002Fmlr-org.com\u002F) includes for example a\n  [gallery](https:\u002F\u002Fmlr-org.com\u002Fgallery.html) with case studies.\n- [Reference manual](https:\u002F\u002Fmlr3.mlr-org.com\u002Freference\u002F)\n- [FAQ](https:\u002F\u002Fmlr-org.com\u002Ffaq.html)\n- Ask questions on Stackoverflow (tag mlr3)\n- **Extension Learners**\n  - Recommended core regression, classification, and survival learners\n    are in [mlr3learners](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3learners)\n  - All others are in\n    [mlr3extralearners](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3extralearners)\n  - Use the [learner search](https:\u002F\u002Fmlr-org.com\u002Flearners.html) to get a\n    simple overview\n- **Cheatsheets**\n  - [Overview of cheatsheets](https:\u002F\u002Fcheatsheets.mlr-org.com)\n  - [mlr3](https:\u002F\u002Fcheatsheets.mlr-org.com\u002Fmlr3.pdf)\n  - [mlr3tuning](https:\u002F\u002Fcheatsheets.mlr-org.com\u002Fmlr3tuning.pdf)\n  - [mlr3pipelines](https:\u002F\u002Fcheatsheets.mlr-org.com\u002Fmlr3pipelines.pdf)\n- **Videos**:\n  - [useR2019 talk on mlr3](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=wsP2hiFnDQs)\n  - [useR2019 talk on mlr3pipelines and\n    mlr3tuning](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=gEW5RxkbQuQ)\n  - [useR2020 tutorial on mlr3, mlr3tuning and\n    mlr3pipelines](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=T43hO2o_nZw)\n    \u003C!--   - [Recorded talk about mlr3spatiotempcv and mlr3spatial at OpenDataScience Europe Conference 2021 in Wageningen, NL](https:\u002F\u002Fav.tib.eu\u002Fmedia\u002F55271) -->\n- **Courses\u002FLectures**\n  - The course [Introduction to Machine learning\n    (I2ML)](https:\u002F\u002Fslds-lmu.github.io\u002Fi2ml\u002F) is a free and open flipped\n    classroom course on the basics of machine learning. `mlr3` is used\n    in the\n    [demos](https:\u002F\u002Fgithub.com\u002Fslds-lmu\u002Flecture_i2ml\u002Ftree\u002Fmaster\u002Fcode-demos-pdf)\n    and\n    [exercises](https:\u002F\u002Fgithub.com\u002Fslds-lmu\u002Flecture_i2ml\u002Ftree\u002Fmaster\u002Fexercises).\n- **Templates\u002FTutorials**\n  - [mlr3-targets](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3-targets): Tutorial\n    showcasing how to use {mlr3} with\n    [targets](https:\u002F\u002Fdocs.ropensci.org\u002Ftargets\u002F) for reproducible ML\n    workflow automation.\n  - Using `mlr3viz` and `animint2` for [interactive\n    visualizations](https:\u002F\u002Fanimint-manual-en.netlify.app\u002Fch20\u002F).\n- [List of extension packages](https:\u002F\u002Fmlr-org.com\u002Fecosystem.html)\n- [mlr-outreach](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr-outreach) contains\n  public talks and slides resources.\n- [Wiki](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Fwiki): Contains mainly\n  information for developers.\n\n## Installation\n\nInstall the last release from CRAN:\n\n``` r\ninstall.packages(\"mlr3\")\n```\n\nInstall the development version from GitHub:\n\n``` r\n# install.packages(\"pak\")\npak::pak(\"mlr-org\u002Fmlr3\")\n```\n\nIf you want to get started with `mlr3`, we recommend installing the\n[mlr3verse](https:\u002F\u002Fmlr3verse.mlr-org.com\u002F) meta-package which installs\n`mlr3` and some of the most important extension packages:\n\n``` r\ninstall.packages(\"mlr3verse\")\n```\n\n## Example\n\n### Constructing Learners and Tasks\n\n``` r\nlibrary(mlr3)\n\n# create learning task\ntask_penguins = as_task_classif(species ~ ., data = palmerpenguins::penguins)\ntask_penguins\n```\n\n    ## \n    ## ── \u003CTaskClassif> (344x8) ───────────────────────────────────────────────────────\n    ## • Target: species\n    ## • Target classes: Adelie (44%), Gentoo (36%), Chinstrap (20%)\n    ## • Properties: multiclass\n    ## • Features (7):\n    ##   • int (3): body_mass_g, flipper_length_mm, year\n    ##   • dbl (2): bill_depth_mm, bill_length_mm\n    ##   • fct (2): island, sex\n\n``` r\n# load learner and set hyperparameter\nlearner = lrn(\"classif.rpart\", cp = .01)\n```\n\n### Basic train + predict\n\n``` r\n# train\u002Ftest split\nsplit = partition(task_penguins, ratio = 0.67)\n\n# train the model\nlearner$train(task_penguins, split$train_set)\n\n# predict data\nprediction = learner$predict(task_penguins, split$test_set)\n\n# calculate performance\nprediction$confusion\n```\n\n    ##            truth\n    ## response    Adelie Chinstrap Gentoo\n    ##   Adelie       146         5      0\n    ##   Chinstrap      6        63      1\n    ##   Gentoo         0         0    123\n\n``` r\nmeasure = msr(\"classif.acc\")\nprediction$score(measure)\n```\n\n    ## classif.acc \n    ##   0.9651163\n\n### Resample\n\n``` r\n# 3-fold cross validation\nresampling = rsmp(\"cv\", folds = 3L)\n\n# run experiments\nrr = resample(task_penguins, learner, resampling)\n\n# access results\nrr$score(measure)[, .(task_id, learner_id, iteration, classif.acc)]\n```\n\n    ##                     task_id    learner_id iteration classif.acc\n    ## 1: palmerpenguins::penguins classif.rpart         1   0.8956522\n    ## 2: palmerpenguins::penguins classif.rpart         2   0.9478261\n    ## 3: palmerpenguins::penguins classif.rpart         3   0.9649123\n\n``` r\nrr$aggregate(measure)\n```\n\n    ## classif.acc \n    ##   0.9361302\n\n## Extension Packages\n\n\u003Ca href=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlr-org\u002Fmlr3\u002Fmain\u002Fman\u002Ffigures\u002Fmlr3verse.svg?sanitize=true\">\u003Cimg src=\"man\u002Ffigures\u002Fmlr3verse.svg\" \u002F>\u003C\u002Fa>\n\nConsult the\n[wiki](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Fwiki\u002FExtension-Packages) for\nshort descriptions and links to the respective repositories.\n\nFor beginners, we strongly recommend to install and load the\n[mlr3verse](https:\u002F\u002Fmlr3verse.mlr-org.com\u002F) package for a better user\nexperience.\n\n## Why a rewrite?\n\n[mlr](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr) was first released to\n[CRAN](https:\u002F\u002Fcran.r-project.org\u002Fpackage=mlr) in 2013. Its core design\nand architecture date back even further. The addition of many features\nhas led to a [feature\ncreep](https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FFeature_creep) which makes\n[mlr](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr) hard to maintain and hard to\nextend. We also think that while mlr was nicely extensible in some parts\n(learners, measures, etc.), other parts were less easy to extend from\nthe outside. Also, many helpful R libraries did not exist at the time\n[mlr](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr) was created, and their inclusion\nwould result in non-trivial API changes.\n\n## Design principles\n\n- Only the basic building blocks for machine learning are implemented in\n  this package.\n- Focus on computation here. No visualization or other stuff. That can\n  go in extra packages.\n- Overcome the limitations of R’s [S3\n  classes](https:\u002F\u002Fadv-r.hadley.nz\u002Fs3.html) with the help of\n  [R6](https:\u002F\u002Fcran.r-project.org\u002Fpackage=R6).\n- Embrace [R6](https:\u002F\u002Fcran.r-project.org\u002Fpackage=R6) for a clean\n  OO-design, object state-changes and reference semantics. This might be\n  less “traditional R”, but seems to fit `mlr` nicely.\n- Embrace [`data.table`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=data.table)\n  for fast and convenient data frame computations.\n- Combine `data.table` and `R6`, for this we will make heavy use of list\n  columns in data.tables.\n- Defensive programming and type safety. All user input is checked with\n  [`checkmate`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=checkmate). Return\n  types are documented, and mechanisms popular in base R which\n  “simplify” the result unpredictably (e.g., `sapply()` or `drop`\n  argument in `[.data.frame`) are avoided.\n- Be light on dependencies. `mlr3` requires the following packages at\n  runtime:\n  - [`parallelly`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=parallelly):\n    Helper functions for parallelization. No extra recursive\n    dependencies.\n  - [`future.apply`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=future.apply):\n    Resampling and benchmarking is parallelized with the\n    [`future`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=future) abstraction\n    interfacing many parallel backends.\n  - [`backports`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=backports): Ensures\n    backward compatibility with older R releases. Developed by members\n    of the `mlr` team. No recursive dependencies.\n  - [`checkmate`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=checkmate): Fast\n    argument checks. Developed by members of the `mlr` team. No extra\n    recursive dependencies.\n  - [`mlr3misc`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=mlr3misc):\n    Miscellaneous functions used in multiple mlr3 [extension\n    packages](https:\u002F\u002Fmlr-org.com\u002Fecosystem.html). Developed by the\n    `mlr` team.\n  - [`paradox`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=paradox):\n    Descriptions for parameters and parameter sets. Developed by the\n    `mlr` team. No extra recursive dependencies.\n  - [`R6`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=R6): Reference class\n    objects. No recursive dependencies.\n  - [`data.table`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=data.table):\n    Extension of R’s `data.frame`. No recursive dependencies.\n  - [`digest`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=digest) (via\n    `mlr3misc`): Hash digests. No recursive dependencies.\n  - [`uuid`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=uuid): Create unique\n    string identifiers. No recursive dependencies.\n  - [`lgr`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=lgr): Logging facility.\n    No extra recursive dependencies.\n  - [`mlr3measures`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=mlr3measures):\n    Performance measures. No extra recursive dependencies.\n  - [`mlbench`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=mlbench): A\n    collection of machine learning data sets. No dependencies.\n  - [`palmerpenguins`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=palmerpenguins):\n    A classification data set about penguins, used on examples and\n    provided as a toy task. No dependencies.\n- [Reflections](https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FReflection_%28computer_programming%29):\n  Objects are queryable for properties and capabilities, allowing you to\n  program on them.\n- Additional functionality that comes with extra dependencies:\n  - To capture output, warnings and exceptions,\n    [`evaluate`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=evaluate) and\n    [`callr`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=callr) can be used.\n\n## Contributing to mlr3\n\nThis R package is licensed under the\n[LGPL-3](https:\u002F\u002Fwww.gnu.org\u002Flicenses\u002Flgpl-3.0.html.en). If you\nencounter problems using this software (lack of documentation,\nmisleading or wrong documentation, unexpected behavior, bugs, …) or just\nwant to suggest features, please open an issue in the [issue\ntracker](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Fissues). Pull requests are\nwelcome and will be included at the discretion of the maintainers.\n\nPlease consult the [wiki](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Fwiki\u002F) for a\n[style guide](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Fwiki\u002FStyle-Guide), a\n[roxygen guide](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Fwiki\u002FRoxygen-Guide) and\na [pull request\nguide](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Fwiki\u002FPR-Guidelines).\n\n## Citing mlr3\n\nIf you use mlr3, please cite our [JOSS\narticle](https:\u002F\u002Fdoi.org\u002F10.21105\u002Fjoss.01903):\n\n    @Article{mlr3,\n      title = {{mlr3}: A modern object-oriented machine learning framework in {R}},\n      author = {Michel Lang and Martin Binder and Jakob Richter and Patrick Schratz and Florian Pfisterer and Stefan Coors and Quay Au and Giuseppe Casalicchio and Lars Kotthoff and Bernd Bischl},\n      journal = {Journal of Open Source Software},\n      year = {2019},\n      month = {dec},\n      doi = {10.21105\u002Fjoss.01903},\n      url = {https:\u002F\u002Fjoss.theoj.org\u002Fpapers\u002F10.21105\u002Fjoss.01903},\n    }\n","# mlr3 \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlr-org_mlr3_readme_82e14b4c5103.png\" align=\"right\" width = \"120\" \u002F>\n\n项目网站: [release](https:\u002F\u002Fmlr3.mlr-org.com\u002F) \\|  \n[dev](https:\u002F\u002Fmlr3.mlr-org.com\u002Fdev\u002F)\n\n机器学习基础组件的高效面向对象编程。[mlr](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr) 的继任者。\n\n\u003C!-- badges: start -->\n\n[![r-cmd-check](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Factions\u002Fworkflows\u002Fr-cmd-check.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Factions\u002Fworkflows\u002Fr-cmd-check.yml)\n[![DOI](https:\u002F\u002Fjoss.theoj.org\u002Fpapers\u002F10.21105\u002Fjoss.01903\u002Fstatus.svg)](https:\u002F\u002Fdoi.org\u002F10.21105\u002Fjoss.01903)\n[![CRAN 状态](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlr-org_mlr3_readme_e7f0a20dfe93.png)](https:\u002F\u002Fcran.r-project.org\u002Fpackage=mlr3)\n[![Mattermost](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fchat-mattermost-orange.svg)](https:\u002F\u002Flmmisld-lmu-stats-slds.srv.mwn.de\u002Fmlr_invite\u002F)\n\u003C!-- badges: end -->\n\n## 资源（面向用户和开发者）\n\n- 我们编写了一本[书籍](https:\u002F\u002Fmlr3book.mlr-org.com\u002F)。这应该是该包的中心入口点。\n- [mlr-org 网站](https:\u002F\u002Fmlr-org.com\u002F) 包含例如[案例研究图库](https:\u002F\u002Fmlr-org.com\u002Fgallery.html)。\n- [参考手册](https:\u002F\u002Fmlr3.mlr-org.com\u002Freference\u002F)\n- [常见问题](https:\u002F\u002Fmlr-org.com\u002Ffaq.html)\n- 在 Stackoverflow 提问（标签 mlr3）\n- **扩展学习器**\n  - 推荐的核心回归、分类和生存学习器位于 [mlr3learners](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3learners)\n  - 其余学习器位于 [mlr3extralearners](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3extralearners)\n  - 使用 [学习器搜索](https:\u002F\u002Fmlr-org.com\u002Flearners.html) 获取简单概览\n- **速查表**\n  - [速查表概览](https:\u002F\u002Fcheatsheets.mlr-org.com)\n  - [mlr3](https:\u002F\u002Fcheatsheets.mlr-org.com\u002Fmlr3.pdf)\n  - [mlr3tuning](https:\u002F\u002Fcheatsheets.mlr-org.com\u002Fmlr3tuning.pdf)\n  - [mlr3pipelines](https:\u002F\u002Fcheatsheets.mlr-org.com\u002Fmlr3pipelines.pdf)\n- **视频**:\n  - [useR2019 关于 mlr3 的演讲](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=wsP2hiFnDQs)\n  - [useR2019 关于 mlr3pipelines 和 mlr3tuning 的演讲](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=gEW5RxkbQuQ)\n  - [useR2020 关于 mlr3、mlr3tuning 和 mlr3pipelines 的教程](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=T43hO2o_nZw)\n    \u003C!--   - [关于 mlr3spatiotempcv 和 mlr3spatial 在 OpenDataScience Europe 2021 大会的演讲录播](https:\u002F\u002Fav.tib.eu\u002Fmedia\u002F55271) -->\n- **课程\u002F讲座**\n  - 课程 [机器学习入门 (I2ML)](https:\u002F\u002Fslds-lmu.github.io\u002Fi2ml\u002F) 是一门免费开放的翻转课堂课程，介绍机器学习基础。`mlr3` 在 [演示](https:\u002F\u002Fgithub.com\u002Fslds-lmu\u002Flecture_i2ml\u002Ftree\u002Fmaster\u002Fcode-demos-pdf) 和 [练习](https:\u002F\u002Fgithub.com\u002Fslds-lmu\u002Flecture_i2ml\u002Ftree\u002Fmaster\u002Fexercises) 中使用。\n- **模板\u002F教程**\n  - [mlr3-targets](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3-targets): 展示如何使用 {mlr3} 与 [targets](https:\u002F\u002Fdocs.ropensci.org\u002Ftargets\u002F) 进行可重复的 ML 工作流自动化。\n  - 使用 `mlr3viz` 和 `animint2` 进行[交互式可视化](https:\u002F\u002Fanimint-manual-en.netlify.app\u002Fch20\u002F)。\n- [扩展包列表](https:\u002F\u002Fmlr-org.com\u002Fecosystem.html)\n- [mlr-outreach](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr-outreach) 包含公开演讲和幻灯片资源。\n- [Wiki](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Fwiki): 主要包含开发者信息。\n\n## 安装\n\n从 CRAN 安装最新版本：\n\n``` r\ninstall.packages(\"mlr3\")\n```\n\n从 GitHub 安装开发版本：\n\n``` r\n# install.packages(\"pak\")\npak::pak(\"mlr-org\u002Fmlr3\")\n```\n\n如果想开始使用 `mlr3`，我们推荐安装 [mlr3verse](https:\u002F\u002Fmlr3verse.mlr-org.com\u002F) 元包，它会安装 `mlr3` 和一些最重要的扩展包：\n\n``` r\ninstall.packages(\"mlr3verse\")\n```\n\n## 示例\n\n### 构建学习器和任务\n\n``` r\nlibrary(mlr3)\n\n# 创建学习任务\ntask_penguins = as_task_classif(species ~ ., data = palmerpenguins::penguins)\ntask_penguins\n```\n\n    ## \n    ## ── \u003CTaskClassif> (344x8) ───────────────────────────────────────────────────────\n    ## • 目标: species\n    ## • 目标类别: Adelie (44%), Gentoo (36%), Chinstrap (20%)\n    ## • 属性: 多类\n    ## • 特征 (7):\n    ##   • 整数 (3): body_mass_g, flipper_length_mm, year\n    ##   • 双精度 (2): bill_depth_mm, bill_length_mm\n    ##   • 因子 (2): island, sex\n\n``` r\n# 加载学习器并设置超参数\nlearner = lrn(\"classif.rpart\", cp = .01)\n```\n\n### 基础训练 + 预测\n\n``` r\n# 训练\u002F测试划分\nsplit = partition(task_penguins, ratio = 0.67)\n\n# 训练模型\nlearner$train(task_penguins, split$train_set)\n\n# 预测数据\nprediction = learner$predict(task_penguins, split$test_set)\n\n# 计算性能\nprediction$confusion\n```\n\n    ##            truth\n    ## response    Adelie Chinstrap Gentoo\n    ##   Adelie       146         5      0\n    ##   Chinstrap      6        63      1\n    ##   Gentoo         0         0    123\n\n``` r\nmeasure = msr(\"classif.acc\")\nprediction$score(measure)\n```\n\n    ## classif.acc \n    ##   0.9651163\n\n### 重采样\n\n``` r\n# 3折交叉验证\nresampling = rsmp(\"cv\", folds = 3L)\n\n# 运行实验\nrr = resample(task_penguins, learner, resampling)\n\n# 访问结果\nrr$score(measure)[, .(task_id, learner_id, iteration, classif.acc)]\n```\n\n    ##                     task_id    learner_id iteration classif.acc\n    ## 1: palmerpenguins::penguins classif.rpart         1   0.8956522\n    ## 2: palmerpenguins::penguins classif.rpart         2   0.9478261\n    ## 3: palmerpenguins::penguins classif.rpart         3   0.9649123\n\n``` r\nrr$aggregate(measure)\n```\n\n    ## classif.acc \n    ##   0.9361302\n\n## 扩展包\n\n\u003Ca href=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlr-org\u002Fmlr3\u002Fmain\u002Fman\u002Ffigures\u002Fmlr3verse.svg?sanitize=true\">\u003Cimg src=\"man\u002Ffigures\u002Fmlr3verse.svg\" \u002F>\u003C\u002Fa>\n\n请查阅 [wiki](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Fwiki\u002FExtension-Packages) 获取简要描述和相应仓库的链接。\n\n对于初学者，我们强烈建议安装并加载 [mlr3verse](https:\u002F\u002Fmlr3verse.mlr-org.com\u002F) 包以获得更好的用户体验。\n\n## 为什么进行重写？\n\n[mlr](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr) 最初于 2013 年发布到 [CRAN](https:\u002F\u002Fcran.r-project.org\u002Fpackage=mlr)。其核心设计和架构可以追溯到更早的时期。随着许多功能的增加，导致了[功能蔓延](https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FFeature_creep)，使[mlr](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr)难以维护和扩展。我们认为虽然 mlr 在某些部分（如学习器、度量等）具有良好的可扩展性，但其他部分从外部扩展起来较为困难。此外，当 [mlr](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr) 被创建时，许多有用的 R 库尚未存在，它们的引入会导致非平凡的 API 变更。\n\n## 设计原则\n\n- 本包仅实现机器学习的基本构建块。\n- 专注于计算。不包含可视化或其他功能。这些可以放在其他包中。\n- 通过 [R6](https:\u002F\u002Fcran.r-project.org\u002Fpackage=R6) 克服 R 的 [S3 类](https:\u002F\u002Fadv-r.hadley.nz\u002Fs3.html) 的限制。\n- 使用 [R6](https:\u002F\u002Fcran.r-project.org\u002Fpackage=R6) 实现清晰的面向对象设计（OO-design）、对象状态变更和引用语义。这可能与传统的 R 用法不同，但似乎与 `mlr` 非常契合。\n- 使用 [`data.table`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=data.table) 实现快速且便捷的数据框计算。\n- 结合 `data.table` 和 `R6`，我们将大量使用数据表中的列表列。\n- 防御性编程和类型安全。所有用户输入都通过 [`checkmate`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=checkmate) 进行检查。返回类型会进行文档说明，并避免使用 base R 中可能导致结果不可预测简化的机制（例如 `sapply()` 或 `[.data.frame` 中的 `drop` 参数）。\n- 依赖项尽量轻量。`mlr3` 运行时需要以下包：\n  - [`parallelly`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=parallelly)：并行化的辅助函数。无额外递归依赖。\n  - [`future.apply`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=future.apply)：通过 [`future`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=future) 抽象接口对重采样和基准测试进行并行化，该接口支持多种并行后端。\n  - [`backports`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=backports)：确保与旧版 R 的向后兼容性。由 `mlr` 团队成员开发。无递归依赖。\n  - [`checkmate`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=checkmate)：快速参数检查。由 `mlr` 团队成员开发。无额外递归依赖。\n  - [`mlr3misc`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=mlr3misc)：用于多个 `mlr3` [扩展包](https:\u002F\u002Fmlr-org.com\u002Fecosystem.html) 的杂项功能。由 `mlr` 团队开发。\n  - [`paradox`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=paradox)：参数和参数集的描述。由 `mlr` 团队开发。无额外递归依赖。\n  - [`R6`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=R6)：引用类对象。无递归依赖。\n  - [`data.table`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=data.table)：R 的 `data.frame` 扩展。无递归依赖。\n  - [`digest`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=digest)（通过 `mlr3misc`）：哈希摘要。无递归依赖。\n  - [`uuid`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=uuid)：生成唯一字符串标识符。无递归依赖。\n  - [`lgr`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=lgr)：日志功能。无额外递归依赖。\n  - [`mlr3measures`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=mlr3measures)：性能度量标准。无额外递归依赖。\n  - [`mlbench`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=mlbench)：机器学习数据集集合。无依赖。\n  - [`palmerpenguins`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=palmerpenguins)：关于企鹅的分类数据集，用于示例并作为玩具任务提供。无依赖。\n- [反射](https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FReflection_%28computer_programming%29)：对象可以查询属性和功能，允许对其进行编程。\n- 需要额外依赖项的附加功能：\n  - 为了捕获输出、警告和异常，可以使用 [`evaluate`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=evaluate) 和 [`callr`](https:\u002F\u002Fcran.r-project.org\u002Fpackage=callr)。\n\n## 贡献于 mlr3\n\n这个 R 包采用 [LGPL-3](https:\u002F\u002Fwww.gnu.org\u002Flicenses\u002Flgpl-3.0.html.en) 许可证。如果您在使用此软件时遇到问题（文档缺失、误导性或错误文档、意外行为、bug 等），或只是想提出功能建议，请在 [问题跟踪器](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Fissues) 中打开一个问题。欢迎提交 Pull Request，是否采纳由维护者决定。\n\n请参阅 [wiki](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Fwiki\u002F) 中的 [风格指南](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Fwiki\u002FStyle-Guide)、[roxygen 指南](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Fwiki\u002FRoxygen-Guide) 和 [Pull Request 指南](https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Fwiki\u002FPR-Guidelines)。\n\n## 引用 mlr3\n\n如果您使用 mlr3，请引用我们的 [JOSS 文章](https:\u002F\u002Fdoi.org\u002F10.21105\u002Fjoss.01903):\n\n    @Article{mlr3,\n      title = {{mlr3}: A modern object-oriented machine learning framework in {R}},\n      author = {Michel Lang and Martin Binder and Jakob Richter and Patrick Schratz and Florian Pfisterer and Stefan Coors and Quay Au and Giuseppe Casalicchio and Lars Kotthoff and Bernd Bischl},\n      journal = {Journal of Open Source Software},\n      year = {2019},\n      month = {dec},\n      doi = {10.21105\u002Fjoss.01903},\n      url = {https:\u002F\u002Fjoss.theoj.org\u002Fpapers\u002F10.21105\u002Fjoss.01903},\n    }","# mlr3 快速上手指南\n\n## 环境准备\n- 系统要求：R语言环境（≥3.5.0）\n- 前置依赖：\n  - R6（R6包）\n  - data.table（数据处理）\n  - 无需额外配置，标准R环境即可运行\n\n## 安装步骤\n安装稳定版（推荐国内镜像加速）：\n```r\ninstall.packages(\"mlr3\", repos = \"https:\u002F\u002Fmirrors.tuna.tsinghua.edu.cn\u002FCRAN\u002F\")\n```\n\n安装开发版（GitHub）：\n```r\n# 安装pak包（若未安装）\ninstall.packages(\"pak\")\npak::pak(\"mlr-org\u002Fmlr3\")\n```\n\n安装完整生态包（mlr3verse）：\n```r\ninstall.packages(\"mlr3verse\")\n```\n\n## 基本使用\n```r\nlibrary(mlr3)\nlibrary(palmerpenguins)  # 确保数据集已加载\n\n# 创建分类任务\ntask_penguins = as_task_classif(species ~ ., data = palmerpenguins::penguins)\n\n# 加载决策树学习器\nlearner = lrn(\"classif.rpart\", cp = 0.01)\n\n# 训练与预测\nsplit = partition(task_penguins, ratio = 0.67)\nlearner$train(task_penguins, split$train_set)\nprediction = learner$predict(task_penguins, split$test_set)\n\n# 评估性能\nprediction$confusion  # 混淆矩阵\nmeasure = msr(\"classif.acc\")\nprediction$score(measure)  # 准确率\n```\n\n交叉验证示例：\n```r\nresampling = rsmp(\"cv\", folds = 3L)\nrr = resample(task_penguins, learner, resampling)\nrr$aggregate(msr(\"classif.acc\"))  # 3折交叉验证平均准确率\n```","某电商平台数据科学家需构建用户流失预测模型，需整合多种算法并优化流程。\n\n### 没有 mlr3 时\n- 手动编写大量重复代码处理不同算法的接口差异，代码冗余度高\n- 特征工程与模型训练耦合紧密，难以快速迭代不同特征组合\n- 超参数调优依赖分散的函数库，需分别学习不同框架的调参语法\n- 模型评估需手动整合多个指标，结果可比性差\n- 缺乏统一的工作流管理，实验记录混乱易出错\n\n### 使用 mlr3 后\n- 通过统一的 `Task` 和 `Learner` 接口简化算法接入，代码量减少40%\n- 基于 `mlr3pipelines` 实现特征工程与模型的模块化组合，迭代效率提升60%\n- 集成 `mlr3tuning` 自动化超参数搜索，调参时间缩短50%\n- 标准化 `Performance` 对象自动聚合多个评估指标，结果分析更高效\n- 通过 `mlr3viz` 可视化模块实现训练过程的全程追踪，实验可复现性显著增强\n\nmlr3 通过模块化设计和标准化接口，将机器学习流程的复杂度降低40%，显著提升数据科学家的开发效率。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlr-org_mlr3_82e14b4c.png","mlr-org","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fmlr-org_347544af.png","",null,"https:\u002F\u002Fmlr-org.com","https:\u002F\u002Fgithub.com\u002Fmlr-org",[82,86],{"name":83,"color":84,"percentage":85},"R","#198CE7",99.7,{"name":87,"color":88,"percentage":89},"TeX","#3D6117",0.3,1063,96,"2026-04-05T22:42:32","LGPL-3.0",1,"未说明",{"notes":97,"python":95,"dependencies":98},"需要R环境（≥3.5）和Rtools工具链，建议通过CRAN安装。扩展包需额外安装，部分功能依赖外部工具如CUDA（若使用GPU加速时）。",[99,100,101,102,103,104,105,106,107,108],"parallelly","future.apply","checkmate","data.table","R6","mlr3misc","paradox","mlr3measures","mlbench","palmerpenguins",[54,15,55,26,14,53,52,51,13],[111,112,113,114,115,67,116],"machine-learning","data-science","classification","regression","r","r-package",4,"2026-03-27T02:49:30.150509","2026-04-06T08:40:10.016573",[121,126,131,136,140,144],{"id":122,"question_zh":123,"answer_zh":124,"source_url":125},4698,"基准测试在外部服务器上运行缓慢且崩溃如何解决？","该问题由并行化配置导致，建议参考 Issue #546 进行排查。同时可尝试使用 `leanify` 和新的内部数据结构优化性能。","https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Fissues\u002F501",{"id":127,"question_zh":128,"answer_zh":129,"source_url":130},4699,"保存 resample 结果时 saveRDS 非常慢如何优化？","需检查 `mlr3misc` 包版本，建议通过 `remotes::install_github(\"mlr-org\u002Fmlr3\", dependencies = TRUE, force = TRUE)` 重新安装所有依赖包以修复潜在兼容性问题。","https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Fissues\u002F482",{"id":132,"question_zh":133,"answer_zh":134,"source_url":135},4700,"生成基准网格设计时出现 'zero-length inputs cannot be mixed' 错误如何解决？","此问题可能由环境配置异常导致，建议通过 `remotes::install_github(\"mlr-org\u002Fmlr3\", dependencies = TRUE, force = TRUE)` 重新安装 mlr3 及其依赖包。","https:\u002F\u002Fgithub.com\u002Fmlr-org\u002Fmlr3\u002Fissues\u002F504",{"id":137,"question_zh":138,"answer_zh":139,"source_url":135},4701,"如何解决基准网格设计中参数传递的零长度输入错误？","需确保任务、学习器和重采样方法的参数传递正确。建议检查 `benchmark_grid` 的输入参数是否符合要求，避免混合不同长度的数据结构。",{"id":141,"question_zh":142,"answer_zh":143,"source_url":125},4702,"并行化运行时 R 会话频繁崩溃如何处理？","应避免在服务器端使用 `futures` 进行并行化操作，改用串行模式运行。同时确保所有依赖包版本一致，避免因环境差异导致的兼容性问题。",{"id":145,"question_zh":146,"answer_zh":147,"source_url":130},4703,"读取 saveRDS 保存的 resample 结果时内存占用过高如何优化？","建议在保存前对数据进行轻量化处理（如 `leanify`），并确保使用最新版 `mlr3misc` 包优化数据存储结构。同时可尝试分批次读取数据以降低内存压力。",[149,154,159,164,169,174,179,184,189,194,199,204,209,214,219,224,229,234,239,244],{"id":150,"version":151,"summary_zh":152,"released_at":153},113911,"v1.6.0","* feat: `Learner` gains a `predict_raw` flag and `Prediction` gains a `raw` field to store the raw prediction object from the upstream model alongside standardized predictions.\n* fix: IDs containing `%` (e.g., from `deparse1(substitute(x))` with pipe expressions) now produce a clear error instead of causing cryptic `sprintf` failures downstream (#1461).\n* fix: `Learner$predict_newdata()` now preserves the target's factor level ordering from training, fixing inverted probabilities for binary classification when the positive class was not the first alphabetical level (#1459).\n","2026-04-02T10:10:24",{"id":155,"version":156,"summary_zh":157,"released_at":158},113912,"v1.5.0","* feat: Replace messages with conditions in logs.\n","2026-02-27T18:04:17",{"id":160,"version":161,"summary_zh":162,"released_at":163},113913,"v1.4.0","* feat: Add `$native_model` active binding to `Learner` to access the model from the upstream package.\n* feat: Learner store the condition of warnings and errors in the `log` field.\n* feat: No supervised tasks with missing target values are allowed anymore.\n* refactor: Validate the input of fields.\n* fix: Assert list input in `assert_learners`, `assert_tasks`, `assert_measures` and `assert_resamplings`.\n* fix: `convert_task` converts the internal validation task now.\n* fix: disable printing of class ratios for large tasks.\n* fix: Make quantiles resettable.\n","2026-02-19T15:03:09",{"id":165,"version":166,"summary_zh":167,"released_at":168},113914,"v1.3.0","* feat: `Learner$predict()` can now add additional data to `PredictionClassif` and `PredictionRegr` objects via the `extra` field.\n* feat: `Mlr3Error` and `Mlr3Warning` classes for errors and warnings.\n* refactor: `$obs_loss` methods in `Measure`, `Prediction`, `ResampleResult`, and `BenchmarkResult`.\n* perf: Use more `lengths()` and `data.table::setattr()` in tests.\n* perf: Use `attr()` instead of `attributes()` for extracting single attributes.\n* test: Use more specialised test functions.\n* fix: `weights_measure` now work with `stratum`.\n* fix: Encapsulation loads loaded packages on the workers.\n* fix: Learners can handle new factor levels.\n* refactor: Use more implicit returns.\n* feat: Learner can pass extra data to predictions.\n","2025-12-03T19:06:09",{"id":170,"version":171,"summary_zh":172,"released_at":173},113915,"v1.2.0","* feat: Add `mirai` support for parallelization and encapsulation.\n* feat: Fallback can now be configured to only be used in case of certain errors via the `when` argument.\n* feat: Custom error and warning classes.\n* fix: `$selected_features` returns error when model is not trained yet.\n* docs: Missing values during scoring.\n* BREAKING CHANGE: Removed `data_format` argument of `$data()` method of `DataBackend`.\n* BREAKING CHANGE: Remove `data_formats` field from `Learner`.\n* BREAKING CHANGE: Remove `DataBackendMatrix` class.\n* feat: Add `materialize_view()` method to `Task` to replace the internal `DataBackend` with a new one after operations like `$select()` and `$filter()`.\n* docs: Information about quantile prediction.\n* perf: Use `fget` in `assert_predictable`.\n* feat: Store oob error in state without requiring storing the model.\n* fix: `$levels()` of `Task` returns in the correct order.\n* chore: Only print up to 10 classes in the `Task` printer.\n* fix: Check if `quantiles` and `quantile_response` are set.\n","2025-09-14T06:08:46",{"id":175,"version":176,"summary_zh":177,"released_at":178},113916,"v1.1.0","* feat: Add new measure `MeasureRegrRQR` for quantile regression.\n* feat: Add `$predict_newdata_fast()` method to `Learner` to speed up prediction.\n* fix: `configure_learner` is passed on `run_experiment()` for autotest learners.\n","2025-07-30T21:22:40",{"id":180,"version":181,"summary_zh":182,"released_at":183},113917,"v1.0.1","* fix: The printer of `Learner` failed when the `validate` field was set.\n* fix: Avoid printing empty line for feature less tasks.\n* perf: Use `data.table::setattr()` for less copying.\n","2025-07-07T07:57:06",{"id":185,"version":186,"summary_zh":187,"released_at":188},113918,"v1.0.0","* BREAKING CHANGE: The mlr3 ecosystem has a base logger now which is named `mlr3`.\n  The `mlr3\u002Fcore` logger is a child of the `mlr3` logger and is used for logging messages from the `mlr3` package.\n  Some extension packages have their own loggers which are children of the mlr3 logger e.g. mlr3\u002Fmlr3pipelines and mlr3\u002Fbbotk for tuning.\n* BREAKING CHANGE: `weights` property and functionality is split into `weights_learner` and `weights_measure`:\n\n  * `weights_learner`: Weights used during training by the Learner.\n  * `weights_measure`: Weights used during scoring predictions via measures.\n\n  Each of these can be disabled via the new field `use_weights` in `Learner` and `Measure` objects.\n* feat: Add `$confusion_weighted` field to `PredictionClassif`.\n* feat: Add `$weights` field to `Prediction`. It contains the `weights_measure` weights from the `Task` that was used for prediction.\n* feat: Add `\"macro_weighted\"` option to `Measure$average` field.\n* feat: `MeasureRegrRSQ` and `MeasureClassifCost` gain `\"weights\"` property.\n* feat: `LearnerClassifFeatureless`, `LearnerRegrFeatureless`, `LearnerClassifDebug`, `LearnerRegrDebug` gain `\"weights\"` property.\n* feat: `Learner` printer now prints information about encapsulation and weights use.\n* feat: Add `score_roc_measures()` to score a prediction on various roc measures.\n* feat: A better error message is thrown, which often happens when incorrectly configuring the `validate` field\n  of a `GraphLearner`\n* feat: Added method `$set_threshold()` to `BenchmarkResult` and `ResamplingResult`, which allows to set the threshold for the response prediction of classification learners, given they have output a probability prediction (#1270).\n* feat: Added field `$uhash_table` to `BenchmarkResult` and functions `uhash()` and `uhashes()`\n  to easily compute uhashes for given learner, task, or resampling ids (#1270).\n* feat: You can now change the default predict type of classification learners to `\"prob\"` by setting\n  the option `mlr3.prob_as_default` to `TRUE` (#1273).\n* feat: `benchmark_grid()` will now throw a warning if you mix different predict types in the\n  design (#1273).\n* feat: Converting a `BenchmarkResult` to a `data.table` now includes the `task_id`, `learner_id`, and `resampling_id` columns (#1275).\n* fix: Add missing parameters for `\"regr.pinball\"` and `\"sim.phi\"` measures.\n","2025-06-18T08:59:58",{"id":190,"version":191,"summary_zh":192,"released_at":193},113919,"v0.23.0","* feat: Add new `col_role` offset in `Task` and offset `Learner` property.\n  A warning is produced if a learner that doesn't support offsets is trained with a task that has an offset column.\n* fix: The `$predict_newdata()` method of `Learner` now automatically conducts type conversions (#685).\n* BREAKING_CHANGE: Predicting on a `Task` with the wrong column information is now an error and not a warning.\n* Column names with UTF-8 characters are now allowed by default.\n  The option `mlr3.allow_utf8_names` is removed.\n* BREAKING CHANGE: `Learner$predict_types` is read-only now.\n* docs: Clear up behavior of `Learner$predict_type` after training.\n* feat: Add callbacks to `resample()` and `benchmark()`.\n* fix: Internal tuning and validation now works when the model requires marshaling (#1256).\n","2025-03-12T13:43:37",{"id":195,"version":196,"summary_zh":197,"released_at":198},113920,"v0.22.1","* fix: Extend `assert_measure()` with checks for trained models in `assert_scorable()`.\n","2024-11-27T16:38:43",{"id":200,"version":201,"summary_zh":202,"released_at":203},113921,"v0.22.0","* fix: Quantiles must not ascend with probabilities.\n* refactor: Replace `tsk(\"boston_housing\")` with `tsk(\"california_housing\")`.\n* feat: Require unique learner ids in `benchmark_grid()`.\n* BREAKING CHANGE: Remove ``$loglik()`` method from all learners.\n* fix: Ignore `future.globals.maxSize` when `future::plan(\"sequential\")` is used.\n* feat: Add `$characteristics` field to `Task` to store additional information.\n","2024-11-25T07:55:13",{"id":205,"version":206,"summary_zh":207,"released_at":208},113922,"v0.21.1","* feat: Throw warning when prediction and measure type do not match.\n* fix: The `mlr_reflections` were broken when an extension package was not loaded on the workers.\n  Extension packages must now register themselves in the `mlr_reflections$loaded_packages` field.\n","2024-10-18T10:37:16",{"id":210,"version":211,"summary_zh":212,"released_at":213},113923,"v0.21.0","* BREAKING CHANGE: Deprecated `data_format` and `data_formats` for `Learner`, `Task`, and `DataBackend` classes.\n* feat: The `partition()` function creates training, test and validation sets now.\n* perf: Optimize the runtime of fixing factor levels.\n* perf: Optimize the runtime of setting row roles.\n* perf: Optimize the runtime of marshalling.\n* perf: Optimize the runtime of `Task$col_info`.\n* fix: column info is now checked for compatibility during `Learner$predict` (#943).\n* BREAKING CHANGE: The predict time of the learner now stores the cumulative duration for all predict sets (#992).\n* feat: `$internal_valid_task` can now be set to an `integer` vector.\n* feat: Measures can now have an empty `$predict_sets` (#1094).\n  This is relevant for measures that only extract information from the model of a learner (such as internal validation scores or AIC \u002F BIC)\n* BREAKING CHANGE: Deprecated the `$divide()` method\n* fix: `Task$cbind()` now works with non-standard primary keys for `data.frames` (#961).\n* fix: Triggering of fallback learner now has log-level `\"info\"` instead of `\"debug\"` (#972).\n* feat: Added new measure `regr.pinball` here and in mlr3measures.\n* feat: Added new measure `mu_auc` here and in mlr3measures.\n* feat: Add option to calculate the mean of the true values on the train set in `msr(\"regr.rsq\")`.\n* feat: Default fallback learner is set when encapsulation is activated.\n* feat: Learners `classif.debug` and `regr.debug` have new methods `$importance()` and `$selected_features()` for testing, also in downstream packages.\n* feat: Create default fallback learner with `default_fallback()`.\n* feat: Check column roles when using `$set_col_roles()` and `$col_roles`.\n* fix: Add predict set to learner hash.\n* BREAKING CHANGE: Encapsulation and the fallback learner are now set with the `$encapsulate(method, fallback)` method.\n  The `$fallback` field is read-only now and the encapsulate status can be retrieved from the `$encapsulation` field.\n","2024-09-24T07:53:58",{"id":215,"version":216,"summary_zh":217,"released_at":218},113924,"v0.20.2","* refactor: move RhpcBLASctl to suggest.\n","2024-07-29T14:38:25",{"id":220,"version":221,"summary_zh":222,"released_at":223},113925,"v0.20.1","* feat: Add multiclass Matthews correlation coefficient `msr(\"classif.mcc\")`.\n","2024-07-22T15:35:01",{"id":225,"version":226,"summary_zh":227,"released_at":228},113926,"v0.20.0","* Added support for learner-internal validation and tuning.\n","2024-06-28T08:49:20",{"id":230,"version":231,"summary_zh":232,"released_at":233},113927,"v0.19.0","* Added support for `\"marshal\"` property, which allows learners to process models so they can be serialized.\nThis happens automatically during `resample()` and `benchmark()`.\n* Encapsulation methods use the same RNG state now.\n* Fix missing values in `default_values.Learner()` function.\n* Encapsulated error messages are now printed with the `lgr` package.\n","2024-04-29T07:43:53",{"id":235,"version":236,"summary_zh":237,"released_at":238},113928,"v0.18.0","* Prepare compatibility with new paradox version.\n* feat: dictionary conversion of `mlr_learners` respects prototype arguments\nrecently added in mlr3misc\n* perf: skip unnecessary clone of learner's state in `resample()`\n","2024-03-05T17:20:09",{"id":240,"version":241,"summary_zh":242,"released_at":243},113929,"v0.17.2","* Skip new `data.table` tests on mac.\n","2024-01-12T07:37:05",{"id":245,"version":246,"summary_zh":247,"released_at":248},113930,"v0.17.1","* Remove `data_prototype` when resampling from `learner$state` to reduce memory consumption.\n* Reduce number of threads used by `data.table` and BLAS to 1 when running `resample()` or `benchmark()` in parallel.\n* Optimize runtime of `resample()` and `benchmark()` by reducing the number of hashing operations.\n","2023-12-21T17:57:42"]