[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-microsoft--RD-Agent":3,"tool-microsoft--RD-Agent":62},[4,18,26,36,46,54],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":17},4358,"openclaw","openclaw\u002Fopenclaw","OpenClaw 是一款专为个人打造的本地化 AI 助手，旨在让你在自己的设备上拥有完全可控的智能伙伴。它打破了传统 AI 助手局限于特定网页或应用的束缚，能够直接接入你日常使用的各类通讯渠道，包括微信、WhatsApp、Telegram、Discord、iMessage 等数十种平台。无论你在哪个聊天软件中发送消息，OpenClaw 都能即时响应，甚至支持在 macOS、iOS 和 Android 设备上进行语音交互，并提供实时的画布渲染功能供你操控。\n\n这款工具主要解决了用户对数据隐私、响应速度以及“始终在线”体验的需求。通过将 AI 部署在本地，用户无需依赖云端服务即可享受快速、私密的智能辅助，真正实现了“你的数据，你做主”。其独特的技术亮点在于强大的网关架构，将控制平面与核心助手分离，确保跨平台通信的流畅性与扩展性。\n\nOpenClaw 非常适合希望构建个性化工作流的技术爱好者、开发者，以及注重隐私保护且不愿被单一生态绑定的普通用户。只要具备基础的终端操作能力（支持 macOS、Linux 及 Windows WSL2），即可通过简单的命令行引导完成部署。如果你渴望拥有一个懂你",349277,3,"2026-04-06T06:32:30",[13,14,15,16],"Agent","开发框架","图像","数据工具","ready",{"id":19,"name":20,"github_repo":21,"description_zh":22,"stars":23,"difficulty_score":10,"last_commit_at":24,"category_tags":25,"status":17},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,"2026-04-05T11:01:52",[14,15,13],{"id":27,"name":28,"github_repo":29,"description_zh":30,"stars":31,"difficulty_score":32,"last_commit_at":33,"category_tags":34,"status":17},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",160784,2,"2026-04-19T11:32:54",[14,13,35],"语言模型",{"id":37,"name":38,"github_repo":39,"description_zh":40,"stars":41,"difficulty_score":42,"last_commit_at":43,"category_tags":44,"status":17},8272,"opencode","anomalyco\u002Fopencode","OpenCode 是一款开源的 AI 编程助手（Coding Agent），旨在像一位智能搭档一样融入您的开发流程。它不仅仅是一个代码补全插件，而是一个能够理解项目上下文、自主规划任务并执行复杂编码操作的智能体。无论是生成全新功能、重构现有代码，还是排查难以定位的 Bug，OpenCode 都能通过自然语言交互高效完成，显著减少开发者在重复性劳动和上下文切换上的时间消耗。\n\n这款工具专为软件开发者、工程师及技术研究人员设计，特别适合希望利用大模型能力来提升编码效率、加速原型开发或处理遗留代码维护的专业人群。其核心亮点在于完全开源的架构，这意味着用户可以审查代码逻辑、自定义行为策略，甚至私有化部署以保障数据安全，彻底打破了传统闭源 AI 助手的“黑盒”限制。\n\n在技术体验上，OpenCode 提供了灵活的终端界面（Terminal UI）和正在测试中的桌面应用程序，支持 macOS、Windows 及 Linux 全平台。它兼容多种包管理工具，安装便捷，并能无缝集成到现有的开发环境中。无论您是追求极致控制权的资深极客，还是渴望提升产出的独立开发者，OpenCode 都提供了一个透明、可信",144296,1,"2026-04-16T14:50:03",[13,45],"插件",{"id":47,"name":48,"github_repo":49,"description_zh":50,"stars":51,"difficulty_score":32,"last_commit_at":52,"category_tags":53,"status":17},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",109154,"2026-04-18T11:18:24",[14,15,13],{"id":55,"name":56,"github_repo":57,"description_zh":58,"stars":59,"difficulty_score":32,"last_commit_at":60,"category_tags":61,"status":17},6121,"gemini-cli","google-gemini\u002Fgemini-cli","gemini-cli 是一款由谷歌推出的开源 AI 命令行工具，它将强大的 Gemini 大模型能力直接集成到用户的终端环境中。对于习惯在命令行工作的开发者而言，它提供了一条从输入提示词到获取模型响应的最短路径，无需切换窗口即可享受智能辅助。\n\n这款工具主要解决了开发过程中频繁上下文切换的痛点，让用户能在熟悉的终端界面内直接完成代码理解、生成、调试以及自动化运维任务。无论是查询大型代码库、根据草图生成应用，还是执行复杂的 Git 操作，gemini-cli 都能通过自然语言指令高效处理。\n\n它特别适合广大软件工程师、DevOps 人员及技术研究人员使用。其核心亮点包括支持高达 100 万 token 的超长上下文窗口，具备出色的逻辑推理能力；内置 Google 搜索、文件操作及 Shell 命令执行等实用工具；更独特的是，它支持 MCP（模型上下文协议），允许用户灵活扩展自定义集成，连接如图像生成等外部能力。此外，个人谷歌账号即可享受免费的额度支持，且项目基于 Apache 2.0 协议完全开源，是提升终端工作效率的理想助手。",100752,"2026-04-10T01:20:03",[45,13,15,14],{"id":63,"github_repo":64,"name":65,"description_en":66,"description_zh":67,"ai_summary_zh":67,"readme_en":68,"readme_zh":69,"quickstart_zh":70,"use_case_zh":71,"hero_image_url":72,"owner_login":73,"owner_name":74,"owner_avatar_url":75,"owner_bio":76,"owner_company":77,"owner_location":77,"owner_email":78,"owner_twitter":79,"owner_website":80,"owner_url":81,"languages":82,"stars":120,"forks":121,"last_commit_at":122,"license":123,"difficulty_score":10,"env_os":124,"env_gpu":125,"env_ram":125,"env_deps":126,"category_tags":132,"github_topics":134,"view_count":32,"oss_zip_url":77,"oss_zip_packed_at":77,"status":17,"created_at":143,"updated_at":144,"faqs":145,"releases":175},8402,"microsoft\u002FRD-Agent","RD-Agent","Research and development (R&D) is crucial for the enhancement of industrial productivity, especially in the AI era, where the core aspects of R&D are mainly focused on data and models. We are committed to automating these high-value generic R&D processes through R&D-Agent, which lets AI drive data-driven AI. 🔗https:\u002F\u002Faka.ms\u002FRD-Agent-Tech-Report","RD-Agent 是一款由微软开源的智能研发助手，旨在利用人工智能自动化高价值的研发流程，真正实现“用数据驱动的 AI 来驱动 AI 研发”。在当前的 AI 时代，研发的核心往往集中在数据处理与模型优化上，RD-Agent 正是为了解决这些环节耗时费力、依赖人工经验的问题而生。它能够自主执行从数据探索、特征工程到模型构建与迭代的全链路任务，显著提升工业生产力。\n\n这款工具特别适合 AI 研究人员、机器学习工程师以及量化交易领域的开发者使用。无论是希望加速实验迭代的科研团队，还是寻求策略优化的金融技术专家，都能从中获益。RD-Agent 的独特亮点在于其强大的通用性与自适应能力，它不仅支持多种大语言模型后端（如 LiteLLM），还配备了直观的 Web 界面供用户实时交互与追踪任务进度。值得一提的是，RD-Agent 目前在权威的 MLE-bench 评测中表现卓越，成为领先的机器学习工程智能体，其相关研究成果也已获 NeurIPS 2025 收录。通过 RD-Agent，繁琐的重复性工作被交给智能体处理，让专业人士能更专注于核心创新。","\u003Ch4 align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_35ad2cce302a.png\" alt=\"RA-Agent logo\" style=\"width:70%; \">\n  \n  \u003Ca href=\"https:\u002F\u002Frdagent.azurewebsites.net\" target=\"_blank\">🖥️ Live Demo\u003C\u002Fa> |\n  \u003Ca href=\"https:\u002F\u002Frdagent.azurewebsites.net\u002Ffactor_loop\" target=\"_blank\">🎥 Demo Video\u003C\u002Fa> \u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=JJ4JYO3HscM&list=PLALmKB0_N3_i52fhUmPQiL4jsO354uopR\" target=\"_blank\">▶️YouTube\u003C\u002Fa>   |\n  \u003Ca href=\"https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Findex.html\" target=\"_blank\">📖 Documentation\u003C\u002Fa> |\n  \u003Ca href=\"https:\u002F\u002Faka.ms\u002FRD-Agent-Tech-Report\" target=\"_blank\">📄 Tech Report\u003C\u002Fa> |\n  \u003Ca href=\"#-paperwork-list\"> 📃 Papers \u003C\u002Fa>\n\u003C\u002Fh3>\n\n\n[![CI](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Fci.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Fci.yml)\n[![CodeQL](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Fgithub-code-scanning\u002Fcodeql\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Fgithub-code-scanning\u002Fcodeql)\n[![Dependabot Updates](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Fdependabot\u002Fdependabot-updates\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Fdependabot\u002Fdependabot-updates)\n[![Lint PR Title](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Fpr.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Fpr.yml)\n[![Release.yml](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Frelease.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Frelease.yml)\n[![Platform](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fplatform-Linux-blue)](https:\u002F\u002Fpypi.org\u002Fproject\u002Frdagent\u002F#files)\n[![PyPI](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Frdagent)](https:\u002F\u002Fpypi.org\u002Fproject\u002Frdagent\u002F)\n[![PyPI - Python Version](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fpyversions\u002Frdagent)](https:\u002F\u002Fpypi.org\u002Fproject\u002Frdagent\u002F)\n[![Release](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fv\u002Frelease\u002Fmicrosoft\u002FRD-Agent)](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Freleases)\n[![GitHub](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flicense\u002Fmicrosoft\u002FRD-Agent)](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fblob\u002Fmain\u002FLICENSE)\n[![pre-commit](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fpre--commit-enabled-brightgreen?logo=pre-commit)](https:\u002F\u002Fgithub.com\u002Fpre-commit\u002Fpre-commit)\n[![Checked with mypy](https:\u002F\u002Fwww.mypy-lang.org\u002Fstatic\u002Fmypy_badge.svg)](http:\u002F\u002Fmypy-lang.org\u002F)\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[![Chat](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fchat-discord-blue)](https:\u002F\u002Fdiscord.gg\u002FybQ97B6Jjy)\n[![Documentation Status](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_13d664e1afd7.png)](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002F?badge=latest)\n[![Readthedocs Preview](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Freadthedocs-preview.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Freadthedocs-preview.yml) \u003C!-- this badge is too long, please place it in the last one to make it pretty --> \n[![arXiv](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FarXiv-2505.14738-00ff00.svg)](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.14738)\n\n\n# 📰 News\n| 🗞️ News        | 📝 Description                 |\n| --            | ------      |\n| Web UI Release | We release a new frontend that can be built and served by `rdagent server_ui` for real-time interaction and trace viewing, currently excluding the `data_science` scenario. |\n| NeurIPS 2025 Acceptance | We are thrilled to announce that our paper [R&D-Agent-Quant](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.15155) has been accepted to NeurIPS 2025 | \n| [Technical Report Release](#overall-technical-report) | Overall framework description and results on MLE-bench | \n| [R&D-Agent-Quant Release](#deep-application-in-diverse-scenarios) | Apply R&D-Agent to quant trading | \n| MLE-Bench Results Released | R&D-Agent currently leads as the [top-performing machine learning engineering agent](#-the-best-machine-learning-engineering-agent) on MLE-bench |\n| Support LiteLLM Backend | We now fully support **[LiteLLM](https:\u002F\u002Fgithub.com\u002FBerriAI\u002Flitellm)** as our default backend for integration with multiple LLM providers. |\n| General Data Science Agent | [Data Science Agent](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Fscens\u002Fdata_science.html) |\n| Kaggle Scenario release | We release **[Kaggle Agent](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Fscens\u002Fdata_science.html)**, try the new features!                  |\n| Official WeChat group release  | We created a WeChat group, welcome to join! (🗪[QR Code](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F880)) |\n| Official Discord release  | We launch our first chatting channel in Discord (🗪[![Chat](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fchat-discord-blue)](https:\u002F\u002Fdiscord.gg\u002FybQ97B6Jjy)) |\n| First release | **R&D-Agent** is released on GitHub |\n\n\n\n# 🏆 The Best Machine Learning Engineering Agent!\n\n[MLE-bench](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fmle-bench) is a comprehensive benchmark evaluating the performance of AI agents on machine learning engineering tasks. Utilizing datasets from 75 Kaggle competitions, MLE-bench provides robust assessments of AI systems' capabilities in real-world ML engineering scenarios.\n\nR&D-Agent currently leads as the top-performing machine learning engineering agent on MLE-bench:\n\n| Agent | Low == Lite (%) | Medium (%) | High (%) | All (%) |\n|---------|--------|-----------|---------|----------|\n| R&D-Agent o3(R)+GPT-4.1(D) | 51.52 ± 6.9 | 19.3 ± 5.5 | 26.67 ± 0 | 30.22 ± 1.5 |\n| R&D-Agent o1-preview | 48.18 ± 2.49 | 8.95 ± 2.36 | 18.67 ± 2.98 | 22.4 ± 1.1 |\n| AIDE o1-preview | 34.3 ± 2.4 | 8.8 ± 1.1 | 10.0 ± 1.9 | 16.9 ± 1.1 |\n\n**Notes:**\n- **O3(R)+GPT-4.1(D)**: This version is designed to both reduce average time per loop and leverage a cost-effective combination of backend LLMs by seamlessly integrating Research Agent (o3) with Development Agent (GPT-4.1).\n- **AIDE o1-preview**: Represents the previously best public result on MLE-bench as reported in the original MLE-bench paper.\n- Average and standard deviation results for R&D-Agent o1-preview is based on a independent of 5 seeds and for R&D-Agent o3(R)+GPT-4.1(D) is based on 6 seeds.\n- According to MLE-Bench, the 75 competitions are categorized into three levels of complexity: **Low==Lite** if we estimate that an experienced ML engineer can produce a sensible solution in under 2 hours, excluding the time taken to train any models; **Medium** if it takes between 2 and 10 hours; and **High** if it takes more than 10 hours.\n\nYou can inspect the detailed runs of the above results online.\n- [R&D-Agent o1-preview detailed runs](https:\u002F\u002Faka.ms\u002FRD-Agent_MLE-Bench_O1-preview)\n- [R&D-Agent o3(R)+GPT-4.1(D) detailed runs](https:\u002F\u002Faka.ms\u002FRD-Agent_MLE-Bench_O3_GPT41)\n\nFor running R&D-Agent on MLE-bench, refer to **[MLE-bench Guide: Running ML Engineering via MLE-bench](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Fscens\u002Fdata_science.html)**\n\n# 🥇 The First Data-Centric Quant Multi-Agent Framework!\n\nR&D-Agent for Quantitative Finance, in short **RD-Agent(Q)**, is the first data-centric, multi-agent framework designed to automate the full-stack research and development of quantitative strategies via coordinated factor-model co-optimization.\n\n![image](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_7f402d19303b.png)\n\nExtensive experiments in real stock markets show that, at a cost under $10, RD-Agent(Q) achieves approximately 2× higher ARR than benchmark factor libraries while using over 70% fewer factors. It also surpasses state-of-the-art deep time-series models under smaller resource budgets. Its alternating factor–model optimization further delivers excellent trade-off between predictive accuracy and strategy robustness.\n\nYou can learn more details about **RD-Agent(Q)** through the [paper](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.15155) and reproduce it through the [documentation](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Fscens\u002Fquant_agent_fin.html).\n\n# Data Science Agent Preview\nCheck out our demo video showcasing the current progress of our Data Science Agent under development:\n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F3eccbecb-34a4-4c81-bce4-d3f8862f7305\n\n# 🌟 Introduction\n\u003Cdiv align=\"center\">\n      \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_3effcc89e0a2.png\" alt=\"Our focused scenario\" style=\"width:80%; \">\n\u003C\u002Fdiv>\n\nR&D-Agent aims to automate the most critical and valuable aspects of the industrial R&D process, and we begin with focusing on the data-driven scenarios to streamline the development of models and data. \nMethodologically, we have identified a framework with two key components: 'R' for proposing new ideas and 'D' for implementing them.\nWe believe that the automatic evolution of R&D will lead to solutions of significant industrial value.\n\n\n\u003C!-- Tag Cloud -->\nR&D is a very general scenario. The advent of R&D-Agent can be your\n- 💰 **Automatic Quant Factory** ([🎥Demo Video](https:\u002F\u002Frdagent.azurewebsites.net\u002Ffactor_loop)|[▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=X4DK2QZKaKY&t=6s))\n- 🤖 **Data Mining Agent:** Iteratively proposing data & models ([🎥Demo Video 1](https:\u002F\u002Frdagent.azurewebsites.net\u002Fmodel_loop)|[▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=dm0dWL49Bc0&t=104s)) ([🎥Demo Video 2](https:\u002F\u002Frdagent.azurewebsites.net\u002Fdmm)|[▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=VIaSTZuoZg4))  and implementing them by gaining knowledge from data.\n- 🦾 **Research Copilot:** Auto read research papers ([🎥Demo Video](https:\u002F\u002Frdagent.azurewebsites.net\u002Freport_model)|[▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=BiA2SfdKQ7o)) \u002F financial reports ([🎥Demo Video](https:\u002F\u002Frdagent.azurewebsites.net\u002Freport_factor)|[▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=ECLTXVcSx-c)) and implement model structures or building datasets.\n- 🤖 **Kaggle Agent:** Auto Model Tuning and Feature Engineering([🎥Demo Video Coming Soon...]()) and implementing them to achieve more in competitions.\n- ...\n\nYou can click the links above to view the demo. We're continuously adding more methods and scenarios to the project to enhance your R&D processes and boost productivity. \n\nAdditionally, you can take a closer look at the examples in our **[🖥️ Live Demo](https:\u002F\u002Frdagent.azurewebsites.net\u002F)**.\n\n\u003Cdiv align=\"center\">\n    \u003Ca href=\"https:\u002F\u002Frdagent.azurewebsites.net\u002F\" target=\"_blank\">\n        \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_3545cfc52eeb.png\" alt=\"Watch the demo\" width=\"80%\">\n    \u003C\u002Fa>\n\u003C\u002Fdiv>\n\n\n# ⚡ Quick start\n\n### RD-Agent currently only supports Linux.\n\nYou can try above demos by running the following command:\n\n### 🐳 Docker installation.\nUsers must ensure Docker is installed before attempting most scenarios. Please refer to the [official 🐳Docker page](https:\u002F\u002Fdocs.docker.com\u002Fengine\u002Finstall\u002F) for installation instructions.\nEnsure the current user can run Docker commands **without using sudo**. You can verify this by executing `docker run hello-world`.\n\n### 🐍 Create a Conda Environment\n- Create a new conda environment with Python (3.10 and 3.11 are well-tested in our CI):\n  ```sh\n  conda create -n rdagent python=3.10\n  ```\n- Activate the environment:\n  ```sh\n  conda activate rdagent\n  ```\n\n### 🛠️ Install the R&D-Agent\n\n#### For Users\n- You can directly install the R&D-Agent package from PyPI:\n  ```sh\n  pip install rdagent\n  ```\n\n#### For Developers\n- If you want to try the latest version or contribute to RD-Agent, you can install it from the source and follow the development setup:\n  ```sh\n  git clone https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\n  cd RD-Agent\n  make dev\n  ```\n\nMore details can be found in the [development setup](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Fdevelopment.html).\n\n### 💊 Health check\n- rdagent provides a health check that currently checks two things.\n  - whether the docker installation was successful.\n  - whether the default port used by the [rdagent ui](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent?tab=readme-ov-file#%EF%B8%8F-monitor-the-application-results) is occupied.\n  ```sh\n  rdagent health_check --no-check-env\n  ```\n\n\n### ⚙️ Configuration\n- The demos requires following ability:\n  - ChatCompletion\n  - json_mode\n  - embedding query\n\n  You can set your Chat Model and Embedding Model in the following ways:\n\n  > **🔥 Attention**: We now provide experimental support for **DeepSeek** models! You can use DeepSeek's official API for cost-effective and high-performance inference. See the configuration example below for DeepSeek setup.\n\n- **Using LiteLLM (Default)**: We now support LiteLLM as a backend for integration with multiple LLM providers. You can configure in multiple ways:\n\n  **Option 1: Unified API base for both models**\n\n  *Configuration Example: `OpenAI` Setup :*\n\n  ```bash\n  cat \u003C\u003C EOF  > .env\n  # Set to any model supported by LiteLLM.\n  CHAT_MODEL=gpt-4o \n  EMBEDDING_MODEL=text-embedding-3-small\n  # Configure unified API base\n  OPENAI_API_BASE=\u003Cyour_unified_api_base>\n  OPENAI_API_KEY=\u003Creplace_with_your_openai_api_key>\n  ```\n\n  *Configuration Example: `Azure OpenAI` Setup :*\n\n  > Before using this configuration, please confirm in advance that your `Azure OpenAI API key` supports `embedded models`.\n\n  ```bash\n  cat \u003C\u003C EOF  > .env\n  EMBEDDING_MODEL=azure\u002F\u003CModel deployment supporting embedding>\n  CHAT_MODEL=azure\u002F\u003Cyour deployment name>\n  AZURE_API_KEY=\u003Creplace_with_your_openai_api_key>\n  AZURE_API_BASE=\u003Cyour_unified_api_base>\n  AZURE_API_VERSION=\u003Cazure api version>\n  ```\n\n  **Option 2: Separate API bases for Chat and Embedding models**\n  ```bash\n  cat \u003C\u003C EOF  > .env\n  # Set to any model supported by LiteLLM.\n  # Configure separate API bases for chat and embedding\n  \n  # CHAT MODEL:\n  CHAT_MODEL=gpt-4o \n  OPENAI_API_BASE=\u003Cyour_chat_api_base>\n  OPENAI_API_KEY=\u003Creplace_with_your_openai_api_key>\n\n  # EMBEDDING MODEL:\n  # TAKE siliconflow as an example, you can use other providers.\n  # Note: embedding requires litellm_proxy prefix\n  EMBEDDING_MODEL=litellm_proxy\u002FBAAI\u002Fbge-large-en-v1.5\n  LITELLM_PROXY_API_KEY=\u003Creplace_with_your_siliconflow_api_key>\n  LITELLM_PROXY_API_BASE=https:\u002F\u002Fapi.siliconflow.cn\u002Fv1\n  ```\n\n  *Configuration Example: `DeepSeek` Setup :*\n\n  >Since many users encounter configuration errors when setting up DeepSeek. Here's a complete working example for DeepSeek Setup:\n  ```bash\n  cat \u003C\u003C EOF  > .env\n  # CHAT MODEL: Using DeepSeek Official API\n  CHAT_MODEL=deepseek\u002Fdeepseek-chat \n  DEEPSEEK_API_KEY=\u003Creplace_with_your_deepseek_api_key>\n\n  # EMBEDDING MODEL: Using SiliconFlow for embedding since deepseek has no embedding model.\n  # Note: embedding requires litellm_proxy prefix\n  EMBEDDING_MODEL=litellm_proxy\u002FBAAI\u002Fbge-m3\n  LITELLM_PROXY_API_KEY=\u003Creplace_with_your_siliconflow_api_key>\n  LITELLM_PROXY_API_BASE=https:\u002F\u002Fapi.siliconflow.cn\u002Fv1\n  ```\n\n  Notice: If you are using reasoning models that include thought processes in their responses (such as \\\u003Cthink> tags), you need to set the following environment variable:\n  ```bash\n  REASONING_THINK_RM=True\n  ```\n\n  You can also use a deprecated backend if you only use `OpenAI API` or `Azure OpenAI` directly. For this deprecated setting and more configuration information, please refer to the [documentation](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Finstallation_and_configuration.html). \n\n\n\n- If your environment configuration is complete, please execute the following commands to check if your configuration is valid. This step is necessary.\n\n  ```bash\n  rdagent health_check\n  ```\n\n### 🚀 Run the Application\n\nThe **[🖥️ Live Demo](https:\u002F\u002Frdagent.azurewebsites.net\u002F)** is implemented by the following commands(each item represents one demo, you can select the one you prefer):\n\n- Run the **Automated Quantitative Trading & Iterative Factors Model Joint Evolution**:  [Qlib](http:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fqlib) self-loop factor & model proposal and implementation application\n  ```sh\n  rdagent fin_quant\n  ```\n\n- Run the **Automated Quantitative Trading & Iterative Factors Evolution**:  [Qlib](http:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fqlib) self-loop factor proposal and implementation application\n  ```sh\n  rdagent fin_factor\n  ```\n\n- Run the **Automated Quantitative Trading & Iterative Model Evolution**: [Qlib](http:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fqlib) self-loop model proposal and implementation application\n  ```sh\n  rdagent fin_model\n  ```\n\n- Run the **Automated Quantitative Trading & Factors Extraction from Financial Reports**:  Run the [Qlib](http:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fqlib) factor extraction and implementation application based on financial reports\n  ```sh\n  # 1. Generally, you can run this scenario using the following command:\n  rdagent fin_factor_report --report-folder=\u003CYour financial reports folder path>\n\n  # 2. Specifically, you need to prepare some financial reports first. You can follow this concrete example:\n  wget https:\u002F\u002Fgithub.com\u002FSunsetWolf\u002Frdagent_resource\u002Freleases\u002Fdownload\u002Freports\u002Fall_reports.zip\n  unzip all_reports.zip -d git_ignore_folder\u002Freports\n  rdagent fin_factor_report --report-folder=git_ignore_folder\u002Freports\n  ```\n\n- Run the **Automated Model Research & Development Copilot**: model extraction and implementation application\n  ```sh\n  # 1. Generally, you can run your own papers\u002Freports with the following command:\n  rdagent general_model \u003CYour paper URL>\n\n  # 2. Specifically, you can do it like this. For more details and additional paper examples, use `rdagent general_model -h`:\n  rdagent general_model  \"https:\u002F\u002Farxiv.org\u002Fpdf\u002F2210.09789\"\n  ```\n\n- Run the **Automated Medical Prediction Model Evolution**: Medical self-loop model proposal and implementation application\n\n  ```bash\n  # Generally, you can run the data science program with the following command:\n  rdagent data_science --competition \u003Cyour competition name>\n\n  # Specifically, you need to create a folder for storing competition files (e.g., competition description file, competition datasets, etc.), and configure the path to the folder in your environment. In addition, you need to use chromedriver when you download the competition descriptors, which you can follow for this specific example:\n\n  # 1. Download the dataset, extract it to the target folder.\n  wget https:\u002F\u002Fgithub.com\u002FSunsetWolf\u002Frdagent_resource\u002Freleases\u002Fdownload\u002Fds_data\u002Farf-12-hours-prediction-task.zip\n  unzip arf-12-hours-prediction-task.zip -d .\u002Fgit_ignore_folder\u002Fds_data\u002F\n\n  # 2. Configure environment variables in the `.env` file\n  dotenv set DS_LOCAL_DATA_PATH \"$(pwd)\u002Fgit_ignore_folder\u002Fds_data\"\n  dotenv set DS_CODER_ON_WHOLE_PIPELINE True\n  dotenv set DS_IF_USING_MLE_DATA False\n  dotenv set DS_SAMPLE_DATA_BY_LLM False\n  dotenv set DS_SCEN rdagent.scenarios.data_science.scen.DataScienceScen\n\n  # 3. run the application\n  rdagent data_science --competition arf-12-hours-prediction-task\n  ```\n\n  **NOTE:** For more information about the dataset, please refer to the [documentation](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Fscens\u002Fdata_science.html).\n\n- Run the **Automated Kaggle Model Tuning & Feature Engineering**:  self-loop model proposal and feature engineering implementation application \u003Cbr \u002F>\n  > Using **tabular-playground-series-dec-2021** as an example. \u003Cbr \u002F>\n  > 1. Register and login on the [Kaggle](https:\u002F\u002Fwww.kaggle.com\u002F) website. \u003Cbr \u002F>\n  > 2. Configuring the Kaggle API. \u003Cbr \u002F>\n  > (1) Click on the avatar (usually in the top right corner of the page) -> `Settings` -> `Create New Token`, A file called `kaggle.json` will be downloaded. \u003Cbr \u002F>\n  > (2) Move `kaggle.json` to `~\u002F.config\u002Fkaggle\u002F` \u003Cbr \u002F>\n  > (3) Modify the permissions of the kaggle.json file. Reference command: `chmod 600 ~\u002F.config\u002Fkaggle\u002Fkaggle.json` \u003Cbr \u002F>\n  > 3. Join the competition: Click `Join the competition` -> `I Understand and Accept` at the bottom of the [competition details page](https:\u002F\u002Fwww.kaggle.com\u002Fcompetitions\u002Ftabular-playground-series-dec-2021\u002Fdata).\n  ```bash\n  # Generally, you can run the Kaggle competition program with the following command:\n  rdagent data_science --competition \u003Cyour competition name>\n\n  # 1. Configure environment variables in the `.env` file\n  mkdir -p .\u002Fgit_ignore_folder\u002Fds_data\n  dotenv set DS_LOCAL_DATA_PATH \"$(pwd)\u002Fgit_ignore_folder\u002Fds_data\"\n  dotenv set DS_CODER_ON_WHOLE_PIPELINE True\n  dotenv set DS_IF_USING_MLE_DATA True\n  dotenv set DS_SAMPLE_DATA_BY_LLM True\n  dotenv set DS_SCEN rdagent.scenarios.data_science.scen.KaggleScen\n\n  # 2. run the application\n  rdagent data_science --competition tabular-playground-series-dec-2021\n  ```\n\n### 🖥️ Monitor the Application Results\n#### Streamlit UI\n\nUse the Streamlit UI to view run logs, especially for the `data_science` scenario.\n\n```sh\nrdagent ui --port 19899 --log-dir \u003Cyour log folder like \"log\u002F\"> --data-science\n```\n\nAbout the `data_science` parameter: If you want to see the logs of the data science scenario, set the `data_science` parameter to `True`; otherwise set it to `False`.\n\n#### Web UI\n\nWe also provide a separate web frontend in `web\u002F` for the Flask backend started by `server_ui`.\n\n**NOTE:** This web UI is different from `rdagent ui`. The current web UI does not support the `data_science` scenario yet. For the `data_science` scenario, please continue to use `rdagent ui --data-science`.\n\n```sh\ncd web\nnpm install\n```\n\nTo build the frontend for the Flask backend, generate the static assets into the default directory used by `server_ui`:\n\n```sh\ncd web\nnpm run build:flask\n```\n\nBy default, `server_ui` serves static files from `.\u002Fgit_ignore_folder\u002Fstatic`. If you need a different location, set the `UI_STATIC_PATH` environment variable before starting the backend.\n\nStart the Flask backend and serve the built frontend together with the real-time APIs:\n\n```sh\nrdagent server_ui --port 19899\n```\n\nAfter that, open `http:\u002F\u002F127.0.0.1:19899` in your browser.\n\n#### Common Notes\n\nPort `19899` is used in the examples above. Before starting either UI, check whether this port is already occupied. If it is, please change it to another available port.\n\nYou can check whether the port is occupied by running:\n\n```sh\nrdagent health_check --no-check-env --no-check-docker\n```\n\n# 🏭 Scenarios\n\nWe have applied R&D-Agent to multiple valuable data-driven industrial scenarios.\n\n\n## 🎯 Goal: Agent for Data-driven R&D\n\nIn this project, we are aiming to build an Agent to automate Data-Driven R\\&D that can\n+ 📄 Read real-world material (reports, papers, etc.) and **extract** key formulas, descriptions of interested **features** and **models**, which are the key components of data-driven R&D .\n+ 🛠️ **Implement** the extracted formulas (e.g., features, factors, and models) in runnable codes.\n   + Due to the limited ability of LLM in implementing at once, build an evolving process for the agent to improve performance by learning from feedback and knowledge.\n+ 💡 Propose **new ideas** based on current knowledge and observations.\n\n\u003C!-- ![Data-Centric R&D Overview](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_1f93d86d3cd6.png) -->\n\n## 📈 Scenarios\u002FDemos\n\nIn the two key areas of data-driven scenarios, model implementation and data building, our system aims to serve two main roles: 🦾Copilot and 🤖Agent. \n- The 🦾Copilot follows human instructions to automate repetitive tasks. \n- The 🤖Agent, being more autonomous, actively proposes ideas for better results in the future.\n\nThe supported scenarios are listed below:\n\n| Scenario\u002FTarget | Model Implementation                   | Data Building                                                                      |\n| --              | --                                     | --                                                                                 |\n| **💹 Finance**      | 🤖 [Iteratively Proposing Ideas & Evolving](https:\u002F\u002Frdagent.azurewebsites.net\u002Fmodel_loop)[▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=dm0dWL49Bc0&t=104s) |  🤖 [Iteratively Proposing Ideas & Evolving](https:\u002F\u002Frdagent.azurewebsites.net\u002Ffactor_loop) [▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=X4DK2QZKaKY&t=6s) \u003Cbr\u002F>   🦾 [Auto reports reading & implementation](https:\u002F\u002Frdagent.azurewebsites.net\u002Freport_factor)[▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=ECLTXVcSx-c)  |\n| **🩺 Medical**      | 🤖 [Iteratively Proposing Ideas & Evolving](https:\u002F\u002Frdagent.azurewebsites.net\u002Fdmm)[▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=VIaSTZuoZg4) | -                                                                                  |\n| **🏭 General**      | 🦾 [Auto paper reading & implementation](https:\u002F\u002Frdagent.azurewebsites.net\u002Freport_model)[▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=BiA2SfdKQ7o) \u003Cbr\u002F> 🤖 Auto Kaggle Model Tuning   | 🤖Auto Kaggle feature Engineering |\n\n- **[RoadMap](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Fscens\u002Fdata_science.html#roadmap)**: Currently, we are working hard to add new features to the Kaggle scenario.\n\nDifferent scenarios vary in entrance and configuration. Please check the detailed setup tutorial in the scenarios documents.\n\nHere is a gallery of [successful explorations](https:\u002F\u002Fgithub.com\u002FSunsetWolf\u002Frdagent_resource\u002Freleases\u002Fdownload\u002Fdemo_traces\u002Fdemo_traces.zip) (5 traces showed in **[🖥️ Live Demo](https:\u002F\u002Frdagent.azurewebsites.net\u002F)**). You can download and view the execution trace using [this command](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent?tab=readme-ov-file#%EF%B8%8F-monitor-the-application-results) from the documentation.\n\nPlease refer to **[📖readthedocs_scen](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Fscens\u002Fcatalog.html)** for more details of the scenarios.\n\n# ⚙️ Framework\n\n\u003Cdiv align=\"center\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_1af740d37b7b.png\" alt=\"Framework-RDAgent\" width=\"85%\">\n\u003C\u002Fdiv>\n\n\nAutomating the R&D process in data science is a highly valuable yet underexplored area in industry. We propose a framework to push the boundaries of this important research field.\n\nThe research questions within this framework can be divided into three main categories:\n| Research Area | Paper\u002FWork List |\n|--------------------|-----------------|\n| **Benchmark the R&D abilities** | [Benchmark](#benchmark) |\n| **Idea proposal:** Explore new ideas or refine existing ones | [Research](#research) |\n| **Ability to realize ideas:** Implement and execute ideas | [Development](#development) |\n\nWe believe that the key to delivering high-quality solutions lies in the ability to evolve R&D capabilities. Agents should learn like human experts, continuously improving their R&D skills.\n\nMore documents can be found in the **[📖 readthedocs](https:\u002F\u002Frdagent.readthedocs.io\u002F)**.\n\n# 📃 Paper\u002FWork list\n\n## Overall Technical Report\n- [R&D-Agent: An LLM-Agent Framework Towards Autonomous Data Science](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.14738)\n```BibTeX\n@misc{yang2025rdagentllmagentframeworkautonomous,\n      title={R&D-Agent: An LLM-Agent Framework Towards Autonomous Data Science}, \n      author={Xu Yang and Xiao Yang and Shikai Fang and Yifei Zhang and Jian Wang and Bowen Xian and Qizheng Li and Jingyuan Li and Minrui Xu and Yuante Li and Haoran Pan and Yuge Zhang and Weiqing Liu and Yelong Shen and Weizhu Chen and Jiang Bian},\n      year={2025},\n      eprint={2505.14738},\n      archivePrefix={arXiv},\n      primaryClass={cs.AI},\n      url={https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.14738}, \n}\n```\n![image](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_c41df67c53fe.png)\n\n## 📊 Benchmark\n- [Towards Data-Centric Automatic R&D](https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.11276)\n```BibTeX\n@misc{chen2024datacentric,\n    title={Towards Data-Centric Automatic R&D},\n    author={Haotian Chen and Xinjie Shen and Zeqi Ye and Wenjun Feng and Haoxue Wang and Xiao Yang and Xu Yang and Weiqing Liu and Jiang Bian},\n    year={2024},\n    eprint={2404.11276},\n    archivePrefix={arXiv},\n    primaryClass={cs.AI}\n}\n```\n![image](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_d524d2e40195.png)\n\n## 🔍 Research\n\nIn a data mining expert's daily research and development process, they propose a hypothesis (e.g., a model structure like RNN can capture patterns in time-series data), design experiments (e.g., finance data contains time-series and we can verify the hypothesis in this scenario), implement the experiment as code (e.g., Pytorch model structure), and then execute the code to get feedback (e.g., metrics, loss curve, etc.). The experts learn from the feedback and improve in the next iteration.\n\nBased on the principles above, we have established a basic method framework that continuously proposes hypotheses, verifies them, and gets feedback from the real-world practice. This is the first scientific research automation framework that supports linking with real-world verification.\n\nFor more detail, please refer to our **[🖥️ Live Demo page](https:\u002F\u002Frdagent.azurewebsites.net)**.\n\n## 🛠️ Development\n\n- [Collaborative Evolving Strategy for Automatic Data-Centric Development](https:\u002F\u002Farxiv.org\u002Fabs\u002F2407.18690)\n```BibTeX\n@misc{yang2024collaborative,\n    title={Collaborative Evolving Strategy for Automatic Data-Centric Development},\n    author={Xu Yang and Haotian Chen and Wenjun Feng and Haoxue Wang and Zeqi Ye and Xinjie Shen and Xiao Yang and Shizhao Sun and Weiqing Liu and Jiang Bian},\n    year={2024},\n    eprint={2407.18690},\n    archivePrefix={arXiv},\n    primaryClass={cs.AI}\n}\n```\n![image](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_3d09526afcc2.png)\n\n## Deep Application in Diverse Scenarios\n\n- [R&D-Agent-Quant: A Multi-Agent Framework for Data-Centric Factors and Model Joint Optimization](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.15155)\n```BibTeX\n@misc{li2025rdagentquantmultiagentframeworkdatacentric,\n      title={R&D-Agent-Quant: A Multi-Agent Framework for Data-Centric Factors and Model Joint Optimization}, \n      author={Yuante Li and Xu Yang and Xiao Yang and Minrui Xu and Xisen Wang and Weiqing Liu and Jiang Bian},\n      year={2025},\n      eprint={2505.15155},\n      archivePrefix={arXiv},\n      primaryClass={q-fin.CP},\n      url={https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.15155}, \n}\n```\n![image](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_d5537d0b6c1a.png)\n\n\n# 🤝 Contributing\n\nWe welcome contributions and suggestions to improve R&D-Agent. Please refer to the [Contributing Guide](CONTRIBUTING.md) for more details on how to contribute.\n\nBefore submitting a pull request, ensure that your code passes the automatic CI checks.\n\n## 📝 Guidelines\nThis project welcomes contributions and suggestions.\nContributing to this project is straightforward and rewarding. Whether it's solving an issue, addressing a bug, enhancing documentation, or even correcting a typo, every contribution is valuable and helps improve R&D-Agent.\n\nTo get started, you can explore the issues list, or search for `TODO:` comments in the codebase by running the command `grep -r \"TODO:\"`.\n\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fcontributors-anon\u002Fmicrosoft\u002FRD-Agent\"\u002F>\n\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fgraphs\u002Fcontributors\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_fc84cbb1b1b8.png\" \u002F>\n\u003C\u002Fa>\n\nBefore we released R&D-Agent as an open-source project on GitHub, it was an internal project within our group. Unfortunately, the internal commit history was not preserved when we removed some confidential code. As a result, some contributions from our group members, including Haotian Chen, Wenjun Feng, Haoxue Wang, Zeqi Ye, Xinjie Shen, and Jinhui Li, were not included in the public commits.\n\n# ⚖️ Legal disclaimer\n\u003Cp style=\"line-height: 1; font-style: italic;\">The RD-agent is provided “as is”, without warranty of any kind, express or implied, including but not limited to the warranties of merchantability, fitness for a particular purpose and noninfringement. The RD-agent is aimed to facilitate research and development process in the financial industry and not ready-to-use for any financial investment or advice. Users shall independently assess and test the risks of the RD-agent in a specific use scenario, ensure the responsible use of AI technology, including but not limited to developing and integrating risk mitigation measures, and comply with all applicable laws and regulations in all applicable jurisdictions. The RD-agent does not provide financial opinions or reflect the opinions of Microsoft, nor is it designed to replace the role of qualified financial professionals in formulating, assessing, and approving finance products. The inputs and outputs of the RD-agent belong to the users and users shall assume all liability under any theory of liability, whether in contract, torts, regulatory, negligence, products liability, or otherwise, associated with use of the RD-agent and any inputs and outputs thereof.\u003C\u002Fp>\n","\u003Ch4 align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_35ad2cce302a.png\" alt=\"RA-Agent logo\" style=\"width:70%; \">\n  \n  \u003Ca href=\"https:\u002F\u002Frdagent.azurewebsites.net\" target=\"_blank\">🖥️ 在线演示\u003C\u002Fa> |\n  \u003Ca href=\"https:\u002F\u002Frdagent.azurewebsites.net\u002Ffactor_loop\" target=\"_blank\">🎥 演示视频\u003C\u002Fa> \u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=JJ4JYO3HscM&list=PLALmKB0_N3_i52fhUmPQiL4jsO354uopR\" target=\"_blank\">▶️YouTube\u003C\u002Fa>   |\n  \u003Ca href=\"https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Findex.html\" target=\"_blank\">📖 文档\u003C\u002Fa> |\n  \u003Ca href=\"https:\u002F\u002Faka.ms\u002FRD-Agent-Tech-Report\" target=\"_blank\">📄 技术报告\u003C\u002Fa> |\n  \u003Ca href=\"#-paperwork-list\"> 📃 论文 \u003C\u002Fa>\n\u003C\u002Fh3>\n\n\n[![CI](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Fci.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Fci.yml)\n[![CodeQL](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Fgithub-code-scanning\u002Fcodeql\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Fgithub-code-scanning\u002Fcodeql)\n[![Dependabot Updates](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Fdependabot\u002Fdependabot-updates\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Fdependabot\u002Fdependabot-updates)\n[![Lint PR Title](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Fpr.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Fpr.yml)\n[![Release.yml](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Frelease.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Frelease.yml)\n[![Platform](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fplatform-Linux-blue)](https:\u002F\u002Fpypi.org\u002Fproject\u002Frdagent\u002F#files)\n[![PyPI](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Frdagent)](https:\u002F\u002Fpypi.org\u002Fproject\u002Frdagent\u002F)\n[![PyPI - Python Version](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fpyversions\u002Frdagent)](https:\u002F\u002Fpypi.org\u002Fproject\u002Frdagent\u002F)\n[![Release](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fv\u002Frelease\u002Fmicrosoft\u002FRD-Agent)](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Freleases)\n[![GitHub](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flicense\u002Fmicrosoft\u002FRD-Agent)](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fblob\u002Fmain\u002FLICENSE)\n[![pre-commit](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fpre--commit-enabled-brightgreen?logo=pre-commit)](https:\u002F\u002Fgithub.com\u002Fpre-commit\u002Fpre-commit)\n[![Checked with mypy](https:\u002F\u002Fwww.mypy-lang.org\u002Fstatic\u002Fmypy_badge.svg)](http:\u002F\u002Fmypy-lang.org\u002F)\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[![Chat](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fchat-discord-blue)](https:\u002F\u002Fdiscord.gg\u002FybQ97B6Jjy)\n[![Documentation Status](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_13d664e1afd7.png)](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002F?badge=latest)\n[![Readthedocs Preview](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Freadthedocs-preview.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Factions\u002Fworkflows\u002Freadthedocs-preview.yml) \u003C!-- 这个徽章太长了，为了美观请把它放在最后一位 --> \n[![arXiv](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FarXiv-2505.14738-00ff00.svg)](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.14738)\n\n\n# 📰 新闻\n| 🗞️ 新闻        | 📝 描述                 |\n| --            | ------      |\n| Web UI 发布 | 我们发布了一个新的前端界面，可以通过 `rdagent server_ui` 构建并提供服务，用于实时交互和追踪查看，目前暂不包括 `data_science` 场景。 |\n| NeurIPS 2025 录用 | 我们非常高兴地宣布，我们的论文 [R&D-Agent-Quant](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.15155) 已被 NeurIPS 2025 接收 | \n| [技术报告发布](#overall-technical-report) | 整体框架描述及 MLE-bench 上的结果 | \n| [R&D-Agent-Quant 发布](#deep-application-in-diverse-scenarios) | 将 R&D-Agent 应用于量化交易 | \n| MLE-Bench 结果公布 | R&D-Agent 目前在 MLE-bench 上以 [最佳机器学习工程代理] 的身份领先 | \n| 支持 LiteLLM 后端 | 我们现在完全支持 **[LiteLLM](https:\u002F\u002Fgithub.com\u002FBerriAI\u002Flitellm)** 作为默认后端，以便与多个 LLM 提供商集成。 |\n| 通用数据科学代理 | [数据科学代理](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Fscens\u002Fdata_science.html) |\n| Kaggle 场景发布 | 我们发布了 **[Kaggle 代理](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Fscens\u002Fdata_science.html)**，快来体验新功能吧！                  |\n| 官方微信群发布  | 我们创建了一个微信群，欢迎大家加入！ (🗪[二维码](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F880)) |\n| 官方 Discord 发布  | 我们在 Discord 上推出了第一个聊天频道 (🗪[![Chat](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fchat-discord-blue)](https:\u002F\u002Fdiscord.gg\u002FybQ97B6Jjy)) |\n| 首次发布 | **R&D-Agent** 在 GitHub 上首次发布 |\n\n\n\n# 🏆 最佳机器学习工程代理！\n\n[MLE-bench](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fmle-bench) 是一个全面的基准测试，用于评估 AI 代理在机器学习工程任务中的表现。它使用来自 75 个 Kaggle 竞赛的数据集，为 AI 系统在真实世界 ML 工程场景中的能力提供了可靠的评估。\n\nR&D-Agent 目前在 MLE-bench 上以最佳机器学习工程代理的身份领先：\n\n| 代理 | 低 == 轻量级 (%) | 中等 (%) | 高 (%) | 总计 (%) |\n|---------|--------|-----------|---------|----------|\n| R&D-Agent o3(R)+GPT-4.1(D) | 51.52 ± 6.9 | 19.3 ± 5.5 | 26.67 ± 0 | 30.22 ± 1.5 |\n| R&D-Agent o1-preview | 48.18 ± 2.49 | 8.95 ± 2.36 | 18.67 ± 2.98 | 22.4 ± 1.1 |\n| AIDE o1-preview | 34.3 ± 2.4 | 8.8 ± 1.1 | 10.0 ± 1.9 | 16.9 ± 1.1 |\n\n**注释：**\n- **O3(R)+GPT-4.1(D)**：此版本旨在通过无缝整合研究代理（o3）与开发代理（GPT-4.1），既减少每次循环的平均时间，又利用经济高效的后端 LLM 组合。\n- **AIDE o1-preview**：代表了原始 MLE-bench 论文中报道的先前在 MLE-bench 上的最佳公开结果。\n- R&D-Agent o1-preview 的平均值和标准差结果基于 5 个独立种子，而 R&D-Agent o3(R)+GPT-4.1(D) 的结果则基于 6 个种子。\n- 根据 MLE-Bench 的分类，这 75 个竞赛被分为三个复杂度级别：如果我们估计一位经验丰富的 ML 工程师可以在不到 2 小时内提出合理的解决方案（不包括训练任何模型的时间），则归为 **低 == 轻量级**；如果需要 2 到 10 小时，则为 **中等**；如果需要超过 10 小时，则为 **高**。\n\n您可以在网上查看上述结果的详细运行记录。\n- [R&D-Agent o1-preview 详细运行](https:\u002F\u002Faka.ms\u002FRD-Agent_MLE-Bench_O1-preview)\n- [R&D-Agent o3(R)+GPT-4.1(D) 详细运行](https:\u002F\u002Faka.ms\u002FRD-Agent_MLE-Bench_O3_GPT41)\n\n如需在 MLE-bench 上运行 R&D-Agent，请参阅 **[MLE-bench 指南：通过 MLE-bench 运行 ML 工程](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Fscens\u002Fdata_science.html)**\n\n# 🥇 首个以数据为中心的量化多智能体框架！\n\n量化金融研发智能体，简称 **RD-Agent(Q)**，是首个以数据为中心的多智能体框架，旨在通过协调一致的因子-模型联合优化，实现量化策略全栈研发的自动化。\n\n![image](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_7f402d19303b.png)\n\n在真实股票市场中的大量实验表明，在成本低于10美元的情况下，RD-Agent(Q) 的年化收益率（ARR）比基准因子库高出约2倍，同时使用的因子数量减少了70%以上。此外，在资源预算较小的情况下，它也超越了当前最先进的深度时间序列模型。其交替进行的因子与模型优化机制，进一步实现了预测精度与策略稳健性之间的良好权衡。\n\n您可以通过 [论文](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.15155) 了解 **RD-Agent(Q)** 的更多细节，并通过 [文档](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Fscens\u002Fquant_agent_fin.html) 进行复现。\n\n# 数据科学智能体预览\n请观看我们的演示视频，了解正在开发中的数据科学智能体的最新进展：\n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F3eccbecb-34a4-4c81-bce4-d3f8862f7305\n\n# 🌟 简介\n\u003Cdiv align=\"center\">\n      \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_3effcc89e0a2.png\" alt=\"我们的聚焦场景\" style=\"width:80%; \">\n\u003C\u002Fdiv>\n\nR&D-Agent 致力于自动化工业研发流程中最为关键和有价值的环节，我们首先聚焦于数据驱动型场景，以简化模型与数据的开发过程。\n从方法论上看，我们确定了一个包含两个核心组件的框架：“R”代表提出新想法，“D”代表将这些想法付诸实施。\n我们相信，研发过程的自动化演进将带来具有重大工业价值的解决方案。\n\n\n\u003C!-- 标签云 -->\nR&D 是一个非常通用的场景。随着 R&D-Agent 的出现，它可以成为您的：\n- 💰 **自动量化工厂** ([🎥演示视频](https:\u002F\u002Frdagent.azurewebsites.net\u002Ffactor_loop)|[▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=X4DK2QZKaKY&t=6s))\n- 🤖 **数据挖掘智能体：** 迭代式地提出数据与模型方案（[🎥演示视频 1](https:\u002F\u002Frdagent.azurewebsites.net\u002Fmodel_loop)|[▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=dm0dWL49Bc0&t=104s))（[🎥演示视频 2](https:\u002F\u002Frdagent.azurewebsites.net\u002Fdmm)|[▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=VIaSTZuoZg4))，并通过从数据中获取知识来加以实现。\n- 🦾 **研究助手：** 自动阅读科研论文（[🎥演示视频](https:\u002F\u002Frdagent.azurewebsites.net\u002Freport_model)|[▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=BiA2SfdKQ7o)) \u002F 财务报告（[🎥演示视频](https:\u002F\u002Frdagent.azurewebsites.net\u002Freport_factor)|[▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=ECLTXVcSx-c))，并据此实现模型结构或构建数据集。\n- 🤖 **Kaggle 智能体：** 自动进行模型调优与特征工程（[🎥演示视频即将发布...]())，并在竞赛中加以应用以取得更好的成绩。\n- …\n\n您可以点击上方链接观看演示视频。我们正不断为该项目添加更多方法和场景，以优化您的研发流程并提升生产力。\n\n此外，您还可以在我们的 **[🖥️ 实时演示](https:\u002F\u002Frdagent.azurewebsites.net\u002F)** 中查看更多示例。\n\n\u003Cdiv align=\"center\">\n    \u003Ca href=\"https:\u002F\u002Frdagent.azurewebsites.net\u002F\" target=\"_blank\">\n        \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_3545cfc52eeb.png\" alt=\"观看演示\" width=\"80%\">\n    \u003C\u002Fa>\n\u003C\u002Fdiv>\n\n\n# ⚡ 快速入门\n\n### RD-Agent 目前仅支持 Linux 系统。\n\n您可以通过运行以下命令来尝试上述演示：\n\n### 🐳 Docker 安装。\n大多数场景的运行都需要先安装 Docker，请务必确保已正确安装。安装说明请参考 [官方 🐳Docker 页面](https:\u002F\u002Fdocs.docker.com\u002Fengine\u002Finstall\u002F)。\n请确保当前用户无需使用 `sudo` 即可执行 Docker 命令。您可以通过运行 `docker run hello-world` 来验证这一点。\n\n### 🐍 创建 Conda 环境\n- 使用 Python 创建一个新的 Conda 环境（我们在 CI 测试中对 Python 3.10 和 3.11 进行了充分验证）：\n  ```sh\n  conda create -n rdagent python=3.10\n  ```\n- 激活该环境：\n  ```sh\n  conda activate rdagent\n  ```\n\n### 🛠️ 安装 R&D-Agent\n\n#### 对于普通用户\n- 您可以直接从 PyPI 安装 R&D-Agent 包：\n  ```sh\n  pip install rdagent\n  ```\n\n#### 对于开发者\n- 如果您想体验最新版本或参与 RD-Agent 的开发，可以从源代码安装并按照开发环境配置步骤操作：\n  ```sh\n  git clone https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\n  cd RD-Agent\n  make dev\n  ```\n\n更多详细信息请参阅 [开发环境设置](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Fdevelopment.html)。\n\n### 💊 健康检查\n- rdagent 提供一项健康检查功能，目前主要检测两方面：\n  - Docker 是否成功安装。\n  - [rdagent UI](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent?tab=readme-ov-file#%EF%B8%8F-monitor-the-application-results) 所使用的默认端口是否已被占用。\n  ```sh\n  rdagent health_check --no-check-env\n  ```\n\n### ⚙️ 配置\n- 本示例需要以下能力：\n  - ChatCompletion\n  - json_mode\n  - embedding query\n\n  您可以通过以下方式设置聊天模型和嵌入模型：\n\n  > **🔥 注意**：我们现在提供对 **DeepSeek** 模型的实验性支持！您可以使用 DeepSeek 的官方 API 进行经济高效且高性能的推理。请参阅下方的 DeepSeek 配置示例。\n\n- **使用 LiteLLM（默认）**：我们现在支持 LiteLLM 作为后端，以集成多个 LLM 提供商。您可以通过多种方式进行配置：\n\n  **选项 1：两种模型共用统一的 API 基础地址**\n\n  *配置示例：`OpenAI` 设置：*\n\n  ```bash\n  cat \u003C\u003C EOF  > .env\n  # 设置为 LiteLLM 支持的任意模型。\n  CHAT_MODEL=gpt-4o \n  EMBEDDING_MODEL=text-embedding-3-small\n  # 配置统一的 API 基础地址\n  OPENAI_API_BASE=\u003Cyour_unified_api_base>\n  OPENAI_API_KEY=\u003Creplace_with_your_openai_api_key>\n  ```\n\n  *配置示例：`Azure OpenAI` 设置：*\n\n  > 在使用此配置之前，请提前确认您的 `Azure OpenAI API 密钥` 是否支持 `嵌入模型`。\n\n  ```bash\n  cat \u003C\u003C EOF  > .env\n  EMBEDDING_MODEL=azure\u002F\u003C支持嵌入的模型部署名称>\n  CHAT_MODEL=azure\u002F\u003C您的部署名称>\n  AZURE_API_KEY=\u003Creplace_with_your_openai_api_key>\n  AZURE_API_BASE=\u003Cyour_unified_api_base>\n  AZURE_API_VERSION=\u003Cazure api 版本>\n  ```\n\n  **选项 2：聊天模型和嵌入模型分别使用不同的 API 基础地址**\n  ```bash\n  cat \u003C\u003C EOF  > .env\n  # 设置为 LiteLLM 支持的任意模型。\n  # 分别配置聊天和嵌入模型的 API 基础地址\n  \n  # 聊天模型：\n  CHAT_MODEL=gpt-4o \n  OPENAI_API_BASE=\u003Cyour_chat_api_base>\n  OPENAI_API_KEY=\u003Creplace_with_your_openai_api_key>\n\n  # 嵌入模型：\n  # 以 Siliconflow 为例，您也可以使用其他提供商。\n  # 注意：嵌入需要添加 litellm_proxy 前缀\n  EMBEDDING_MODEL=litellm_proxy\u002FBAAI\u002Fbge-large-en-v1.5\n  LITELLM_PROXY_API_KEY=\u003Creplace_with_your_siliconflow_api_key>\n  LITELLM_PROXY_API_BASE=https:\u002F\u002Fapi.siliconflow.cn\u002Fv1\n  ```\n\n  *配置示例：`DeepSeek` 设置：*\n\n  > 由于许多用户在配置 DeepSeek 时会遇到错误，这里提供一个完整的 DeepSeek 配置示例：\n  ```bash\n  cat \u003C\u003C EOF  > .env\n  # 聊天模型：使用 DeepSeek 官方 API\n  CHAT_MODEL=deepseek\u002Fdeepseek-chat \n  DEEPSEEK_API_KEY=\u003Creplace_with_your_deepseek_api_key>\n\n  # 嵌入模型：由于 DeepSeek 没有嵌入模型，因此使用 SiliconFlow 提供的嵌入服务。\n  # 注意：嵌入需要添加 litellm_proxy 前缀\n  EMBEDDING_MODEL=litellm_proxy\u002FBAAI\u002Fbge-m3\n  LITELLM_PROXY_API_KEY=\u003Creplace_with_your_siliconflow_api_key>\n  LITELLM_PROXY_API_BASE=https:\u002F\u002Fapi.siliconflow.cn\u002Fv1\n  ```\n\n  注意：如果您使用的推理模型会在响应中包含思考过程（例如 `\u003Cthink>` 标签），则需要设置以下环境变量：\n  ```bash\n  REASONING_THINK_RM=True\n  ```\n\n  如果您仅直接使用 `OpenAI API` 或 `Azure OpenAI`，也可以使用已弃用的后端。有关此弃用设置及更多配置信息，请参阅 [文档](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Finstallation_and_configuration.html)。\n\n\n\n- 如果您的环境配置已完成，请执行以下命令以检查配置是否有效。此步骤是必要的。\n\n  ```bash\n  rdagent health_check\n  ```\n\n### 🚀 运行应用\n\n**[🖥️ 实时演示](https:\u002F\u002Frdagent.azurewebsites.net\u002F)** 由以下命令实现（每项代表一个演示，您可以选择自己喜欢的）：\n\n- 运行 **自动化量化交易与迭代因子模型联合进化**：[Qlib](http:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fqlib) 自循环因子与模型方案及实现应用\n  ```sh\n  rdagent fin_quant\n  ```\n\n- 运行 **自动化量化交易与迭代因子进化**：[Qlib](http:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fqlib) 自循环因子方案及实现应用\n  ```sh\n  rdagent fin_factor\n  ```\n\n- 运行 **自动化量化交易与迭代模型进化**：[Qlib](http:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fqlib) 自循环模型方案及实现应用\n  ```sh\n  rdagent fin_model\n  ```\n\n- 运行 **自动化量化交易与财报中因子提取**：基于财务报表运行 [Qlib](http:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fqlib) 因子提取与实现应用\n  ```sh\n  # 1. 一般情况下，您可以通过以下命令运行此场景：\n  rdagent fin_factor_report --report-folder=\u003C您的财报文件夹路径>\n\n  # 2. 具体来说，您需要先准备一些财务报表。可以按照以下具体示例操作：\n  wget https:\u002F\u002Fgithub.com\u002FSunsetWolf\u002Frdagent_resource\u002Freleases\u002Fdownload\u002Freports\u002Fall_reports.zip\n  unzip all_reports.zip -d git_ignore_folder\u002Freports\n  rdagent fin_factor_report --report-folder=git_ignore_folder\u002Freports\n  ```\n\n- 运行 **自动化模型研发协作助手**：模型提取与实现应用\n  ```sh\n  # 1. 一般情况下，您可以通过以下命令运行自己的论文\u002F报告：\n  rdagent general_model \u003C您的论文URL>\n\n  # 2. 具体操作如下。更多详细信息和额外的论文示例，请使用 `rdagent general_model -h`：\n  rdagent general_model  \"https:\u002F\u002Farxiv.org\u002Fpdf\u002F2210.09789\"\n  ```\n\n- 运行 **自动化医学预测模型进化**：医学自循环模型方案及实现应用\n\n  ```bash\n  # 一般情况下，您可以通过以下命令运行数据科学项目：\n  rdagent data_science --competition \u003C您的竞赛名称>\n\n  # 具体来说，您需要创建一个用于存放竞赛文件的文件夹（例如竞赛说明文件、竞赛数据集等），并在环境中配置该文件夹的路径。此外，在下载竞赛说明时，您还需要使用 chromedriver，具体操作可参考以下示例：\n\n  # 1. 下载数据集，并将其解压到目标文件夹。\n  wget https:\u002F\u002Fgithub.com\u002FSunsetWolf\u002Frdagent_resource\u002Freleases\u002Fdownload\u002Fds_data\u002Farf-12-hours-prediction-task.zip\n  unzip arf-12-hours-prediction-task.zip -d .\u002Fgit_ignore_folder\u002Fds_data\u002F\n\n  # 2. 在 `.env` 文件中配置环境变量\n  dotenv set DS_LOCAL_DATA_PATH \"$(pwd)\u002Fgit_ignore_folder\u002Fds_data\"\n  dotenv set DS_CODER_ON_WHOLE_PIPELINE True\n  dotenv set DS_IF_USING_MLE_DATA False\n  dotenv set DS_SAMPLE_DATA_BY_LLM False\n  dotenv set DS_SCEN rdagent.scenarios.data_science.scen.DataScienceScen\n\n  # 3. 运行应用\n  rdagent data_science --competition arf-12-hours-prediction-task\n  ```\n\n  **注意**：有关数据集的更多信息，请参阅 [文档](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Fscens\u002Fdata_science.html)。\n\n- 运行 **自动化 Kaggle 模型调优与特征工程**：自循环模型方案及特征工程实现应用 \u003Cbr \u002F>\n  > 以 **tabular-playground-series-dec-2021** 为例。\u003Cbr \u002F>\n  > 1. 在 [Kaggle](https:\u002F\u002Fwww.kaggle.com\u002F) 网站上注册并登录。\u003Cbr \u002F>\n  > 2. 配置 Kaggle API。\u003Cbr \u002F>\n  > (1) 点击头像（通常位于页面右上角）-> `Settings` -> `Create New Token`，会下载一个名为 `kaggle.json` 的文件。\u003Cbr \u002F>\n  > (2) 将 `kaggle.json` 移动到 `~\u002F.config\u002Fkaggle\u002F`。\u003Cbr \u002F>\n  > (3) 修改 `kaggle.json` 文件的权限。参考命令：`chmod 600 ~\u002F.config\u002Fkaggle\u002Fkaggle.json`。\u003Cbr \u002F>\n  > 3. 加入竞赛：在 [竞赛详情页](https:\u002F\u002Fwww.kaggle.com\u002Fcompetitions\u002Ftabular-playground-series-dec-2021\u002Fdata) 底部点击 `Join the competition` -> `I Understand and Accept`。\n  ```bash\n  # 一般情况下，您可以通过以下命令运行 Kaggle 竞赛程序：\n  rdagent data_science --competition \u003C您的竞赛名称>\n\n  # 1. 在 `.env` 文件中配置环境变量\n  mkdir -p .\u002Fgit_ignore_folder\u002Fds_data\n  dotenv set DS_LOCAL_DATA_PATH \"$(pwd)\u002Fgit_ignore_folder\u002Fds_data\"\n  dotenv set DS_CODER_ON_WHOLE_PIPELINE True\n  dotenv set DS_IF_USING_MLE_DATA True\n  dotenv set DS_SAMPLE_DATA_BY_LLM True\n  dotenv set DS_SCEN rdagent.scenarios.data_science.scen.KaggleScen\n\n  # 2. 运行应用\n  rdagent data_science --competition tabular-playground-series-dec-2021\n  ```\n\n### 🖥️ 监控应用结果\n#### Streamlit UI\n\n使用 Streamlit UI 查看运行日志，尤其是 `data_science` 场景的日志。\n\n```sh\nrdagent ui --port 19899 --log-dir \u003C您的日志文件夹，如 \"log\u002F\"> --data-science\n```\n\n关于 `data_science` 参数：如果您想查看数据科学场景的日志，请将 `data_science` 参数设置为 `True`；否则设置为 `False`。\n\n#### Web UI\n\n我们还在 `web\u002F` 中提供了一个独立的 Web 前端，用于由 `server_ui` 启动的 Flask 后端。\n\n**注意**：此 Web UI 与 `rdagent ui` 不同。目前的 Web UI 尚不支持 `data_science` 场景。对于 `data_science` 场景，请继续使用 `rdagent ui --data-science`。\n\n```sh\ncd web\nnpm install\n```\n\n为了构建 Flask 后端的前端，将静态资源生成到 `server_ui` 默认使用的目录中：\n\n```sh\ncd web\nnpm run build:flask\n```\n\n默认情况下，`server_ui` 会从 `.\u002Fgit_ignore_folder\u002Fstatic` 提供静态文件。如果您需要其他位置，请在启动后端之前设置 `UI_STATIC_PATH` 环境变量。\n\n启动 Flask 后端，并将构建好的前端与实时 API 一起提供：\n\n```sh\nrdagent server_ui --port 19899\n```\n\n之后，在浏览器中打开 `http:\u002F\u002F127.0.0.1:19899`。\n\n#### 常见注意事项\n\n上述示例中使用了端口 `19899`。在启动任何 UI 之前，请检查该端口是否已被占用。如果已被占用，请更换为其他可用端口。\n\n您可以通过运行以下命令来检查端口是否被占用：\n\n```sh\nrdagent health_check --no-check-env --no-check-docker\n```\n\n# 🏭 场景\n\n我们已将 R&D-Agent 应用于多个有价值的以数据驱动的工业场景。\n\n## 🎯 目标：数据驱动研发的智能体\n\n在本项目中，我们的目标是构建一个用于自动化数据驱动研发的智能体，它能够：\n+ 📄 阅读真实世界中的材料（报告、论文等），并**提取**关键公式、感兴趣的**特征**和**模型**的描述，这些正是数据驱动研发的核心组成部分。\n+ 🛠️ 将提取出的公式（如特征、因子和模型）**实现**为可运行的代码。\n   + 由于大语言模型一次性实现的能力有限，我们将构建一个演进式流程，使智能体能够通过学习反馈和知识不断提升性能。\n+ 💡 基于现有知识和观察提出**新想法**。\n\n\u003C!-- ![以数据为中心的研发概览](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_1f93d86d3cd6.png) -->\n\n## 📈 场景\u002F演示\n\n在数据驱动场景的两个关键领域——模型实现和数据构建——我们的系统旨在扮演两种主要角色：🦾协作助手和🤖智能体。\n- 🦾协作助手会根据人类指令自动执行重复性任务。\n- 🤖智能体则更具自主性，能够主动提出改进方案，以获得更好的结果。\n\n支持的场景如下：\n\n| 场景\u002F目标 | 模型实现                   | 数据构建                                                                      |\n| --              | --                                     | --                                                                                 |\n| **💹 金融**      | 🤖 [迭代提出想法并演进](https:\u002F\u002Frdagent.azurewebsites.net\u002Fmodel_loop)[▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=dm0dWL49Bc0&t=104s) |  🤖 [迭代提出想法并演进](https:\u002F\u002Frdagent.azurewebsites.net\u002Ffactor_loop) [▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=X4DK2QZKaKY&t=6s) \u003Cbr\u002F>   🦾 [自动阅读报告并实现](https:\u002F\u002Frdagent.azurewebsites.net\u002Freport_factor)[▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=ECLTXVcSx-c)  |\n| **🩺 医疗**      | 🤖 [迭代提出想法并演进](https:\u002F\u002Frdagent.azurewebsites.net\u002Fdmm)[▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=VIaSTZuoZg4) | -                                                                                  |\n| **🏭 通用**      | 🦾 [自动阅读论文并实现](https:\u002F\u002Frdagent.azurewebsites.net\u002Freport_model)[▶️YouTube](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=BiA2SfdKQ7o) \u003Cbr\u002F> 🤖 自动Kaggle模型调优   | 🤖自动Kaggle特征工程 |\n\n- **[路线图](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Fscens\u002Fdata_science.html#roadmap)**：目前，我们正努力为Kaggle场景添加新功能。\n\n不同场景的入口和配置有所不同。请参阅场景文档中的详细设置教程。\n\n这里提供一组[成功探索案例](https:\u002F\u002Fgithub.com\u002FSunsetWolf\u002Frdagent_resource\u002Freleases\u002Fdownload\u002Fdemo_traces\u002Fdemo_traces.zip)（其中5个案例可在**[🖥️在线演示](https:\u002F\u002Frdagent.azurewebsites.net\u002F)**中查看）。您可以使用文档中的[此命令](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent?tab=readme-ov-file#%EF%B8%8F-monitor-the-application-results)下载并查看执行轨迹。\n\n更多场景详情，请参阅**[📖readthedocs_scen](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Fscens\u002Fcatalog.html)**。\n\n# ⚙️ 框架\n\n\u003Cdiv align=\"center\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_1af740d37b7b.png\" alt=\"Framework-RDAgent\" width=\"85%\">\n\u003C\u002Fdiv>\n\n\n自动化数据科学领域的研发过程是一项极具价值但尚未充分开发的课题。我们提出了一套框架，旨在推动这一重要研究领域的边界。\n\n该框架下的研究问题可分为三大类：\n| 研究领域 | 论文\u002F工作列表 |\n|--------------------|-----------------|\n| **评估研发能力** | [基准测试](#benchmark) |\n| **想法提出：** 探索新思路或优化现有思路 | [研究](#research) |\n| **实现能力：** 将想法付诸实践并执行 | [开发](#development) |\n\n我们认为，提供高质量解决方案的关键在于不断进化研发能力。智能体应像人类专家一样学习，持续提升其研发技能。\n\n更多文档请参见**[📖 readthedocs](https:\u002F\u002Frdagent.readthedocs.io\u002F)**。\n\n# 📃 论文\u002F工作列表\n\n## 总体技术报告\n- [R&D-Agent：迈向自主数据科学的大语言模型智能体框架](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.14738)\n```BibTeX\n@misc{yang2025rdagentllmagentframeworkautonomous,\n      title={R&D-Agent: An LLM-Agent Framework Towards Autonomous Data Science}, \n      author={Xu Yang and Xiao Yang and Shikai Fang and Yifei Zhang and Jian Wang and Bowen Xian and Qizheng Li and Jingyuan Li and Minrui Xu and Yuante Li and Haoran Pan and Yuge Zhang and Weiqing Liu and Yelong Shen and Weizhu Chen and Jiang Bian},\n      year={2025},\n      eprint={2505.14738},\n      archivePrefix={arXiv},\n      primaryClass={cs.AI},\n      url={https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.14738}, \n}\n```\n![image](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_c41df67c53fe.png)\n\n## 📊 基准测试\n- [迈向以数据为中心的自动化研发](https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.11276)\n```BibTeX\n@misc{chen2024datacentric,\n    title={Towards Data-Centric Automatic R&D},\n    author={Haotian Chen and Xinjie Shen and Zeqi Ye and Wenjun Feng and Haoxue Wang and Xiao Yang and Xu Yang and Weiqing Liu and Jiang Bian},\n    year={2024},\n    eprint={2404.11276},\n    archivePrefix={arXiv},\n    primaryClass={cs.AI}\n}\n```\n![image](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_d524d2e40195.png)\n\n## 🔍 研究\n\n在数据挖掘专家的日常研发过程中，他们会提出假设（例如，RNN这样的模型结构可以捕捉时间序列数据中的模式）、设计实验（比如金融数据包含时间序列，可以在该场景下验证假设）、将实验转化为代码实现（如PyTorch模型结构），然后执行代码以获取反馈（如指标、损失曲线等）。专家们会根据反馈进行调整，并在下一轮迭代中改进。\n\n基于上述原则，我们建立了一个基础方法框架，能够持续提出假设、验证假设，并从实际应用中获取反馈。这是首个支持与现实验证相结合的科学研究自动化框架。\n\n欲了解更多详情，请访问我们的**[🖥️在线演示页面](https:\u002F\u002Frdagent.azurewebsites.net)**。\n\n## 🛠️ Development\n\n- [Collaborative Evolving Strategy for Automatic Data-Centric Development](https:\u002F\u002Farxiv.org\u002Fabs\u002F2407.18690)\n```BibTeX\n@misc{yang2024collaborative,\n    title={Collaborative Evolving Strategy for Automatic Data-Centric Development},\n    author={Xu Yang and Haotian Chen and Wenjun Feng and Haoxue Wang and Zeqi Ye and Xinjie Shen and Xiao Yang and Shizhao Sun and Weiqing Liu and Jiang Bian},\n    year={2024},\n    eprint={2407.18690},\n    archivePrefix={arXiv},\n    primaryClass={cs.AI}\n}\n```\n![image](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_3d09526afcc2.png)\n\n## Deep Application in Diverse Scenarios\n\n- [R&D-Agent-Quant: A Multi-Agent Framework for Data-Centric Factors and Model Joint Optimization](https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.15155)\n```BibTeX\n@misc{li2025rdagentquantmultiagentframeworkdatacentric,\n      title={R&D-Agent-Quant: A Multi-Agent Framework for Data-Centric Factors and Model Joint Optimization}, \n      author={Yuante Li and Xu Yang and Xiao Yang and Minrui Xu and Xisen Wang and Weiqing Liu and Jiang Bian},\n      year={2025},\n      eprint={2505.15155},\n      archivePrefix={arXiv},\n      primaryClass={q-fin.CP},\n      url={https:\u002F\u002Farxiv.org\u002Fabs\u002F2505.15155}, \n}\n```\n![image](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_d5537d0b6c1a.png)\n\n\n# 🤝 Contributing\n\nWe welcome contributions and suggestions to improve R&D-Agent. Please refer to the [Contributing Guide](CONTRIBUTING.md) for more details on how to contribute.\n\nBefore submitting a pull request, ensure that your code passes the automatic CI checks.\n\n## 📝 Guidelines\nThis project welcomes contributions and suggestions.\nContributing to this project is straightforward and rewarding. Whether it's solving an issue, addressing a bug, enhancing documentation, or even correcting a typo, every contribution is valuable and helps improve R&D-Agent.\n\nTo get started, you can explore the issues list, or search for `TODO:` comments in the codebase by running the command `grep -r \"TODO:\"`.\n\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fcontributors-anon\u002Fmicrosoft\u002FRD-Agent\"\u002F>\n\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fgraphs\u002Fcontributors\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_readme_fc84cbb1b1b8.png\" \u002F>\n\u003C\u002Fa>\n\nBefore we released R&D-Agent as an open-source project on GitHub, it was an internal project within our group. Unfortunately, the internal commit history was not preserved when we removed some confidential code. As a result, some contributions from our group members, including Haotian Chen, Wenjun Feng, Haoxue Wang, Zeqi Ye, Xinjie Shen, and Jinhui Li, were not included in the public commits.\n\n# ⚖️ Legal disclaimer\n\u003Cp style=\"line-height: 1; font-style: italic;\">The RD-agent is provided “as is”, without warranty of any kind, express or implied, including but not limited to the warranties of merchantability, fitness for a particular purpose and noninfringement. The RD-agent is aimed to facilitate research and development process in the financial industry and not ready-to-use for any financial investment or advice. Users shall independently assess and test the risks of the RD-agent in a specific use scenario, ensure the responsible use of AI technology, including but not limited to developing and integrating risk mitigation measures, and comply with all applicable laws and regulations in all applicable jurisdictions. The RD-agent does not provide financial opinions or reflect the opinions of Microsoft, nor is it designed to replace the role of qualified financial professionals in formulating, assessing, and approving finance products. The inputs and outputs of the RD-agent belong to the users and users shall assume all liability under any theory of liability, whether in contract, torts, regulatory, negligence, products liability, or otherwise, associated with use of the RD-agent and any inputs and outputs thereof.\u003C\u002Fp>","# RD-Agent 快速上手指南\n\nRD-Agent 是由微软开源的自动化研发智能体框架，专注于数据驱动场景（如量化交易、Kaggle 竞赛、数据挖掘），能够自动提出新想法（Research）并实现代码落地（Development）。目前该工具在 MLE-bench 基准测试中表现优异。\n\n## 1. 环境准备\n\n**系统要求：**\n- **操作系统**：目前仅支持 **Linux** 系统。\n- **Docker**：必须安装 Docker 并确保当前用户无需 `sudo` 即可运行 Docker 命令。\n  - 验证方法：执行 `docker run hello-world` 确认安装成功且权限正常。\n  - 安装参考：[Docker 官方文档](https:\u002F\u002Fdocs.docker.com\u002Fengine\u002Finstall\u002F)\n- **Python 版本**：推荐 Python 3.10 或 3.11（已在 CI 中充分测试）。\n- **Conda**：建议使用 Conda 管理虚拟环境。\n\n**前置依赖检查：**\n确保已安装 Git 和 Make（用于开发者模式安装）。\n\n## 2. 安装步骤\n\n### 第一步：创建 Conda 环境\n```sh\nconda create -n rdagent python=3.10\nconda activate rdagent\n```\n\n### 第二步：安装 RD-Agent\n\n**方式 A：普通用户（推荐）**\n直接从 PyPI 安装稳定版：\n```sh\npip install rdagent\n```\n> 💡 **国内加速建议**：如遇下载缓慢，可使用清华或阿里镜像源：\n> ```sh\n> pip install rdagent -i https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple\n> ```\n\n**方式 B：开发者（获取最新功能）**\n从源码安装以便贡献代码或体验最新特性：\n```sh\ngit clone https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\ncd RD-Agent\nmake dev\n```\n\n### 第三步：健康检查\n安装完成后，运行以下命令检查 Docker 状态及端口占用情况：\n```sh\nrdagent health_check --no-check-env\n```\n\n## 3. 基本使用\n\n### 配置模型\nRD-Agent 依赖大语言模型（LLM）进行推理，默认支持 **LiteLLM** 后端，可对接多种模型提供商。在使用前需配置 Chat 模型和 Embedding 模型。\n\n你需要设置环境变量或配置文件以启用以下能力：\n- `ChatCompletion`\n- `json_mode`\n- `embedding query`\n\n*(注：具体配置方式请参考官方文档中关于 LiteLLM 的设置指南)*\n\n### 运行示例\n配置完成后，你可以尝试运行内置的演示场景。例如，启动量化因子挖掘或 Kaggle 竞赛代理：\n\n```sh\n# 示例：运行特定场景（具体场景名称请参考文档）\nrdagent run --scenario quant_agent_fin\n```\n\n### 监控与可视化\nRD-Agent 提供了 Web UI 用于实时交互和查看运行轨迹：\n```sh\nrdagent server_ui\n```\n启动后，访问浏览器即可查看任务进度、日志及生成的代码结果。\n\n---\n*更多详细场景（如 Data Science Agent, Quant Agent）及高级配置，请参阅 [官方文档](https:\u002F\u002Frdagent.readthedocs.io\u002Fen\u002Flatest\u002Findex.html)。*","某量化对冲基金的算法团队正致力于从海量另类数据中挖掘新的 Alpha 因子，以优化其高频交易策略。\n\n### 没有 RD-Agent 时\n- **人工试错效率低**：研究员需手动编写代码清洗数据、构建特征并回测，一个因子的验证周期长达数天，难以覆盖广阔的搜索空间。\n- **知识复用困难**：过往成功的建模经验散落在不同人员的笔记中，无法系统化地转化为可迭代的逻辑，导致重复造轮子。\n- **创新瓶颈明显**：面对复杂的市场非线性关系，人类直觉容易陷入局部最优，难以自主发现反直觉但高收益的数据组合模式。\n- **工程维护成本高**：大量的临时脚本缺乏统一标准，随着实验数量增加，代码库变得混乱且难以复现结果。\n\n### 使用 RD-Agent 后\n- **自动化闭环研发**：RD-Agent 自主完成从数据加载、特征工程、模型训练到回测评估的全流程，将单个因子的探索时间压缩至小时级。\n- **自我进化能力**：它能自动分析失败案例，提取通用规律并生成新的假设，像资深研究员一样不断迭代优化策略逻辑。\n- **突破人类认知局限**：通过大规模并行搜索与深度推理，RD-Agent 成功挖掘出多个传统方法忽略的高夏普比率因子，显著提升策略收益。\n- **标准化产出**：所有实验过程自动生成文档与结构化代码，确保每一步决策可追溯、可复现，大幅降低团队协作成本。\n\nRD-Agent 让 AI 自主驱动数据驱动的 AI 研发，将量化团队从繁琐的工程实现中解放出来，专注于更高维度的战略决策。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmicrosoft_RD-Agent_b9664e2a.png","microsoft","Microsoft","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fmicrosoft_4900709c.png","Open source projects and samples from Microsoft",null,"opensource@microsoft.com","OpenAtMicrosoft","https:\u002F\u002Fopensource.microsoft.com","https:\u002F\u002Fgithub.com\u002Fmicrosoft",[83,87,91,95,98,102,106,109,113,116],{"name":84,"color":85,"percentage":86},"Python","#3572A5",87,{"name":88,"color":89,"percentage":90},"Vue","#41b883",8.8,{"name":92,"color":93,"percentage":94},"Jupyter Notebook","#DA5B0B",1.7,{"name":96,"color":97,"percentage":42},"Shell","#89e051",{"name":99,"color":100,"percentage":101},"JavaScript","#f1e05a",0.6,{"name":103,"color":104,"percentage":105},"CSS","#663399",0.3,{"name":107,"color":108,"percentage":105},"Makefile","#427819",{"name":110,"color":111,"percentage":112},"Dockerfile","#384d54",0.1,{"name":114,"color":115,"percentage":112},"TypeScript","#3178c6",{"name":117,"color":118,"percentage":119},"HTML","#e34c26",0,12541,1496,"2026-04-16T21:35:38","MIT","Linux","未说明",{"notes":127,"python":128,"dependencies":129},"目前仅支持 Linux 操作系统。必须安装 Docker，且当前用户需能够无需 sudo 权限直接运行 Docker 命令。默认使用 LiteLLM 作为后端以集成多种大模型提供商。建议使用 Conda 创建虚拟环境进行安装。","3.10, 3.11",[130,131],"Docker","LiteLLM",[133,15,35,14,16,13],"其他",[135,136,137,138,139,140,141,142],"agent","ai","automation","data-mining","data-science","development","llm","research","2026-03-27T02:49:30.150509","2026-04-19T23:04:13.089140",[146,151,156,161,166,171],{"id":147,"question_zh":148,"answer_zh":149,"source_url":150},37607,"在 Windows 系统上运行本地可视化 UI 界面时，选择日志后没有任何信息显示，如何解决？","这是因为 Windows 系统的日志文件路径分隔符与 Linux 不同，导致路径被识别为标签时出错。该问题已在 PR#437 中修复。如果仍有部分未显示，请参考 PR#440 的更新。建议升级到包含这些修复的最新版本。","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F436",{"id":152,"question_zh":153,"answer_zh":154,"source_url":155},37608,"使用 SiliconFlow (deepseek 方法) 进行 Embedding 测试时报错 'The parameter is invalid' (Error code: 400)，如何解决？","这是因为缺少必要的参数 `encoding_format`。可以通过以下两种方式解决：\n1. **修改源代码（不推荐）**：\n   - 找到 `rdagent\u002Fapp\u002Futils\u002Fhealth_check.py`，在 `embedding` 函数调用中添加参数 `encoding_format=\"float\"`。\n   - 找到 `rdagent\u002Foai\u002Fbackend\u002Flitellm.py`，同样在 `embedding` 函数调用中添加参数 `encoding_format=\"float\"`。\n2. **配置文件方式（推荐）**：检查相关配置是否已正确支持该参数传递。","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1371",{"id":157,"question_zh":158,"answer_zh":159,"source_url":160},37609,"RD-Agent 是否支持结合 Backtrader 等回测库来发现和优化量化交易策略？","目前 Qlib 场景下，因子和模型的调优是相对分离的，但你可以通过修改框架代码将两者连接起来。Qlib 的回测框架对应真实世界的反馈，目前已封装在 Docker 中以方便使用和切换。虽然 Backtrader 专注于执行模拟，但 Qlib 提供了更全栈的方法（从数据入库到风险评估）。你可以将 Docker 中的回测框架替换为更适合你需求的框架（如 Backtrader），社区也欢迎此类贡献。","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F381",{"id":162,"question_zh":163,"answer_zh":164,"source_url":165},37610,"运行 rdagent 时遇到 'Read-only file system: \u002Froot\u002F.qlib\u002Fqlib_data' 错误，如何解决？","这是由于 `rdagent.utils.env` 文件中第 610 行和第 612 行的 `extra_volumes` 模式被错误地设置为只读 ('read only')，而程序试图创建目录时需要写入权限。\n**临时解决方法**：\n找到 `rdagent\u002Futils\u002Fenv.py` 文件，在第 610 行和第 612 行，将 `self.conf.extra_volume_mode` 的值修改为 `\"rw\"` (读写模式)。","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F794",{"id":167,"question_zh":168,"answer_zh":169,"source_url":170},37611,"是否可以基于 Langchain 复现或重构此项目？","目前项目并非基于 Langchain 构建，但理论上可以尝试复现。开发者表示不确定每个组件是否都能完美映射，鼓励用户在熟悉项目代码后尝试使用 Langchain 进行复现实验。","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F716",{"id":172,"question_zh":173,"answer_zh":174,"source_url":160},37612,"RD-Agent 与 Qlib 和 Backtrader 的关系是什么？应该如何选择？","Qlib 是专为策略研究人员设计的，涵盖从数据、模型训练到回测、风险评估的全流程，与 RD-Agent 的结合更为紧密和全栈。Backtrader 则更侧重于执行模拟和投资组合分析。两者通常不是互补关系，不建议同时运行。如果你需要机构级的策略权衡（如杠杆、资金流控制），可能需要基于 Backtrader 开发额外代码，但在 RD-Agent 生态中，推荐使用 Qlib 以获得更完整的体验。",[176,181,186,191,196,201,206,211,216,221,226],{"id":177,"version":178,"summary_zh":179,"released_at":180},305700,"v0.8.0","## [0.8.0](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcompare\u002Fv0.7.0...v0.8.0) (2025-11-03)\n\n\n### 功能特性\n\n* 在提案中添加 RAG MCP（[#1267](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1267)）（[a0cd102](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fa0cd1025c141aee6d4e6cb10286c77d827b89379)）\n* 添加编码器检查并给予更多时间（[#1127](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1127)）（[e32d229](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fe32d229f2b722acac53f4e2f7d8a98e29cb19dc1)）\n* 为 UI 数据缓存添加 enable_cache 切换开关（[#1075](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1075)）（[0c9f193](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F0c9f1930e8d5df1c00bfb32ee578da2dc53db1ec)）\n* 添加 extra_eval 配置和 import_class，用于自定义评估器（[#1097](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1097)）（[5accec3](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F5accec37c8828ac42005c2d12b815bef599b547e)）\n* 在提案中添加 hypo_critic 和 hypo_rewrite（[#1106](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1106)）（[71440f6](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F71440f643fc9d952dfa064359c1945b729dbfd9f)）\n* 向 MultiProcessEvolvingStrategy 添加 improve_mode，用于选择性地执行任务（[#1273](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1273)）（[9344635](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F93446356952803d8b1f1eb0c39da825c19274cb6)）\n* 将循环 ID 映射添加到跟踪节点，并更新 UI 标签（[#1098](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1098)）（[5437851](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F54378518dadd6c38496eceda8ef5b33b375a5c97)）\n* 在调试模式中添加掩码推理功能（[#1154](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1154)）（[ef749ab](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fef749ab744fb6fbafd1a8e6a3642cce20ce96069)）\n* 为跟踪添加仅显示成功项的过滤器切换开关（[#1047](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1047)）（[5e582cc](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F5e582cc71d5c153666c465cb2d797dc71e43c501)）\n* 添加选项，允许仅在第一次评估循环中启用超参数调优（[#1211](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1211)）（[bc3fa17](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fbc3fa170b029f50c8f7b1828cdf4ffd024e64b8b)）\n* 将之前的运行循环添加到运行历史记录中（[#1142](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1142)）（[8de9f75](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F8de9f757ea134b04cde0622c6225678d85a87862)）\n* 为 DSRunnerFeedback 添加 reasoning 属性，以增强评估上下文（[#1162](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1162)）（[4e41c97](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F4e41c9797cbafd35cc0d883fede4226398c573e1)）\n* 添加样本提交文件检查（[#1053](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1053)）（[6a840d8](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F6a840d819251e64d98daa40289592a05ac5fb369)）\n* 添加 show_hard_limit 选项，并更新 DataScience 设置中的时间限制处理（[#1144](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1144)) (","2025-11-03T14:01:36",{"id":182,"version":183,"summary_zh":184,"released_at":185},305701,"v0.7.0","## [0.7.0](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcompare\u002Fv0.6.1...v0.7.0) (2025-07-08)\n\n\n### 功能特性\n\n* 添加代码变更摘要 ([#1000](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1000)) ([937ec26](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F937ec263b215928633822c4d76ad4e47442c8198))\n* 新增 hide_base_name 选项，并更新数据文件夹提示信息 ([#1004](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1004)) ([2f61fa8](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F2f61fa8cd90c91ad29f320ce9ea6c49f49ac9111))\n* 为 DS 场景实验添加运行时间统计功能 ([#1007](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1007)) ([030abd8](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F030abd87191377641a678c80852f5ecad84e7a6e))\n* 合并代码摘要功能，并支持更多追踪记录 ([#1025](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1025)) ([48201e7](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F48201e79b55ff5a98dad51702a7d0ac6b1ddc9eb))\n* 显示首次进化轮次的代码差异 ([#1009](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1009)) ([4844622](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F4844622e5fd28d7cbaabd9d7888f8204c60b76b3))\n* 尝试在全部数据上运行 coder 模型 ([#1017](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1017)) ([4973e05](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F4973e0532248c6172eec3bb70dffda052af2d14f))\n\n\n### 错误修复\n\n* 修复 DS 评估中的一个小 bug ([#1012](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1012)) ([5a520e9](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F5a520e9d44899d44fddc0f2e5571596223161b71))\n* 修复量化场景中的一些问题 ([#1026](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1026)) ([7b34d41](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F7b34d418642d1c0c2986db9ecf6a5d9bc22cc3da))\n* 增加对 Deepseek 模型的实验性支持，并更新配置相关文档 ([#1024](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1024)) ([35cfc19](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F35cfc193f9b35d786aeb7585334427ad358c982f))","2025-07-08T03:19:35",{"id":187,"version":188,"summary_zh":189,"released_at":190},305702,"v0.6.1","## [0.6.1](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcompare\u002Fv0.6.0...v0.6.1) (2025-06-28)\n\n\n### Bug修复\n\n* 修复挂载问题 ([#1001](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F1001)) ([4ae2f13](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F4ae2f1303dfcbaea53d459be7c8e85bf85ce5f4f))\n* 处理 dag_parent 索引错误的bug ([#996](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F996)) ([bda12ff](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fbda12ffecf9ae116e0d04eece0c6a1b61413d916))\n* 改进日志文件夹排序及选择的用户体验 ([#993](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F993)) ([b116807](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fb11680777f116b6c40f9e535e0da10c186c95050))","2025-06-28T12:07:17",{"id":192,"version":193,"summary_zh":194,"released_at":195},305703,"v0.6.0","## [0.6.0](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcompare\u002Fv0.5.0...v0.6.0) (2025-06-26)\n\n\n### 功能特性\n\n* 用于多轨迹的异步机制 ([#981](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F981)) ([9e60c32](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F9e60c32cf348481eb55617809c059c359d7603b8))\n\n\n### 错误修复\n\n* 将 `direct_exp_gen` 改为异步，以避免死循环 ([#992](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F992)) ([78c203d](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F78c203d8eefbba67fc120b35cb25e85b2200ac49))\n* 清理 Docker 容器，防止资源堆积和系统性能下降 ([#975](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F975)) ([05cf094](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F05cf094913e48c903c8a4476d6c609d8bfa10681))\n* 修复一个 bug 并更新文档 ([#978](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F978)) ([d1ae9e1](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fd1ae9e1dcc2ccd1ffe05cb1c6db3e905fa70425c))\n* 合并数据科学 v3 和 v2 版本 ([#974](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F974)) ([1ba7548](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F1ba754853ce2010ce1cb0bbd217b67689fa1ebdf))\n* 优化细节 ([#979](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F979)) ([25caa3d](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F25caa3d00c255286dce27915b9355987b87ed2e8))\n* 优化提示词 ([#987](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F987)) ([76df96e](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F76df96ee88212a8aee7f518b9cacf80591dc2939))","2025-06-26T14:32:37",{"id":197,"version":198,"summary_zh":199,"released_at":200},305704,"v0.5.0","## [0.5.0](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcompare\u002Fv0.4.0...v0.5.0) (2025-06-18)\n\n\n### 功能特性\n\n* 添加对 score_df 中值是否为 NaN 的检查 ([#756](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F756)) ([d9cc780](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fd9cc78098beb27f3a1bf2f2d461302db177b7d41))\n* 添加竞赛级别过滤器，并将常量提取到工具函数中 ([#869](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F869)) ([b40b605](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fb40b6055368e6c72d8435352104b1c281b06da7f))\n* 添加 DocDev 用于自动生成工作空间文档 ([#781](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F781)) ([bcba6ea](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fbcba6eac32684ebb267c93b4e85dbfa9561d15d1))\n* 添加起草流程 ([#832](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F832)) ([efedddf](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fefedddf39bc19221fdffc2e39ee0a09097fc82b0))\n* 在 DSTrace 中添加 last_exp_fb，并更新反馈检索的使用方式 ([#910](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F910)) ([10531fd](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F10531fda9438c6915b26d5013bd2413e1333ceb9))\n* 在 RD 循环中添加 mlflow 日志记录器以进行日志记录 ([#815](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F815)) ([b91b54f](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fb91b54f355c26b751087d0c14774f466e82866de))\n* 添加朴素实验生成器，并更新提案配置 ([#759](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F759)) ([75494f4](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F75494f4fed5bc845acfd7f7bacef385f0f96c514))\n* 添加 RD-Agent-Quant 场景 ([#838](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F838)) ([6e42d52](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F6e42d523a85df67aa13927abbf0894564c71880e))\n* 向 LiteLLMAPIBackend 和 LLMSett… 添加 reasoning_effort 参数 ([#754](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F754)) ([113889f](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F113889fefe9b09aaea1b564704c81664b8f77ec5))\n* 在反馈中添加审稿人信息 ([#765](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F765)) ([1a95bee](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F1a95bee6aa6bc6f45fdeb484f3a6f81caa273038))\n* 改进检查点选择器 ([#790](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F790)) ([50ea033](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F50ea0336e93d8cb39fb871e81a3f61abdf293bc7))\n* 将工作空间中的 Python 和 CSV 文件归档，以保存结果 ([#814](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F814)) ([67d0e01](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F67d0e01e7c9237da1371d93cbf9d86f5f46faac4))\n* 检查点选择 ([#744](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F744)) ([a15a06a](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fa15a06ad643977db59d7cac9da52e637cf80395a))\n* 自定义数据 ([#810](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F810)) ([6322916](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F632291608cf605bd8bcfcab0017824823bdecdb8))\n* 转储模型 ([#776](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fis","2025-06-18T07:08:29",{"id":202,"version":203,"summary_zh":204,"released_at":205},305705,"v0.4.0","## [0.4.0](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcompare\u002Fv0.3.0...v0.4.0) (2025-04-04)\n\n\n### 功能特性\n\n* （Kaggle）添加竞赛基础模板：tabular-playground-series-may-2022（[#481](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F481)）（[f3405ca](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Ff3405ca732eb0ddca8e18ea72f69cbd86055c4ab)）\n* 统一的 CoSTEER，以适应更多场景（[#491](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F491)）（[cddbd02](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fcddbd02e3ad3ccf6ad01443777319dc5c7eb08a7)）\n* 添加一项新竞赛（[#474](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F474)）（[2fc0d77](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F2fc0d77c485a31f647e21f4578e2e326f7032964)）\n* 添加一个工具，用于将工作区文件保存到指定文件夹（[#728](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F728)）（[bca864b](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fbca864b7edeafe3f88405efb695ca8acad6252f8)）\n* 添加基线分数统计（[#590](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F590)）（[2948026](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F2948026c390d067b643f8c8247c1447f1dc023e4)）\n* 在 env.py 中为 Docker 卷添加可配置的卷模式（[#537](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F537)）（[642a022](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F642a02239431411b91959f23e69b454997ca75d5)）\n* 为语义搜索添加约束标签（[#680](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F680)）（[0584cfc](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F0584cfcd13ca1a62c85390ea2ee7574370748d31)）\n* 将交叉验证添加到工作流中（[#700](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F700)）（[82e9b00](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F82e9b00be62b01673353a7aaa3ab0e2e3ecaf3ca)）\n* 添加 describe_data_folder_v2（[#738](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F738)）（[bc8e846](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fbc8e8460e0246321792ff3347b1b8905416ad075)）\n* 为加载函数添加 do_truncate 控制（[#656](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F656)）（[2b960a5](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F2b960a58dfdeba69522a0f72ecf0975bb6ae87ee)）\n* 为加载函数添加 do_truncate 控制（[#656](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F656)）（[2b960a5](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F2b960a58dfdeba69522a0f72ecf0975bb6ae87ee)）\n* 在数据科学场景中添加 EDA（[#639](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F639)）（[35aa479](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F35aa479f00edf118d43ec228e0a84c155332957a)）\n* 添加假设指南和基于规则的排名（[#746](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F746)）（[c077b82](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fc077b8239cc72904c4bc450845ed2a11aa5445f0)）\n* 向 shrink_text 函数及设置中添加行长限制（[#715](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F715)）（[75ed5e1](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F75ed5e1c2ce1bf20bb55190c10a4134e04694d2b)）\n* 为 m 添加 loop_n 参数","2025-04-04T03:51:53",{"id":207,"version":208,"summary_zh":209,"released_at":210},305706,"v0.3.0","## [0.3.0](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcompare\u002Fv0.2.1...v0.3.0) (2024-10-21)\n\n\n### 功能特性\n\n* 添加 Kaggle 新模板 ([#289](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F289)) ([eee3ab5](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Feee3ab5b25198224826cb7a8a17eab28bd5d1f7d))\n* 为 Kaggle 场景添加下载 submission.csv 按钮 ([#317](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F317)) ([dcdcbe4](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fdcdcbe46b4858bfb133ae3cca056e7f602d5cf63))\n* 添加 Kaggle 命令 ([#271](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F271)) ([0938394](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F0938394b7084ffbf3294d8c23d2d34bf7322ca0b))\n* 添加 Kaggle 模板：feedback-prize ([#331](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F331)) ([a288e39](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fa288e399e6b0beec62729bd7d46b98a55de5ab79))\n* 为 Kaggle 添加更多模板 ([#291](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F291)) ([da752ec](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fda752ec806e6f5f5679bc27ac1c072ed9a319251))\n* 将普通 RAG 集成到框架中 ([#360](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F360)) ([91b0b1f](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F91b0b1f66c3c1bf757cb64c4cfbdcaafe59eab74))\n* 添加 qlib_factor_strategy ([#307](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F307)) ([f8f59ff](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Ff8f59ff0a1be4428a68c8c27f220aabad0b6c9f0))\n* 在 Kaggle 场景中添加排名功能 ([#401](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F401)) ([b16b4be](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fb16b4beb402e0c27dfb39ee9d2a120f1b56d447c))\n* 在 RDLoop 中为每一步和循环添加运行时测量功能。([#281](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F281)) ([83058c8](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F83058c864ceeec413dd29bf501030d5a7bd34679))\n* 添加 s3e11 Kaggle 模板 ([#324](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F324)) ([8c57524](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F8c57524bead1c8f655a08763d608eb7a6dd5975e))\n* 添加 RepoAnalyzer，以增强工作空间的自动摘要功能 ([#264](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F264)) ([0bd349a](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F0bd349af50b9b881ba1774bdeb4d723529ef2aa9))\n* 增加对 Kaggle 场景中 RAG 加载和存储的支持。([#269](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F269)) ([c4895de](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fc4895de83f1ed000e563d42b3468a6bd9e5a4965))\n* 宣布 Discord 和 WeChat 社区 ([#367](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F367)) ([acac507](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Facac5078a103b71afa6bd6c053b0766a6a7e609d))\n* 在一次 Kaggle RDLoop 结束后自动提交结果 ([#345](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F345)) ([ab55d70](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fab55d7052b53a928b84dc5d5d0d2999d90ca9056))\n* 改进反馈与评估 ([#346](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F346)) ([cc9a8c1](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002F","2024-10-21T09:37:35",{"id":212,"version":213,"summary_zh":214,"released_at":215},305707,"v0.2.1","## [0.2.1](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcompare\u002Fv0.2.0...v0.2.1) (2024-09-10)\n\n\n### 错误修复\n\n* 配置中的默认模型值 ([#256](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F256)) ([c097585](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fc097585f631f401c2c0966f6ad4c17286924f011))\n* 修复 `.env` 文件错误 ([#257](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F257)) ([923063c](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F923063c1fd957c4ed42e97272c72b5e9545451dc))\n* 更新 README 文件 ([#248](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F248)) ([8cede22](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F8cede2209922876490148459e1134da828e1fda0))","2024-09-10T11:44:11",{"id":217,"version":218,"summary_zh":219,"released_at":220},305708,"v0.2.0","## [0.2.0](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcompare\u002Fv0.1.0...v0.2.0) (2024-09-07)\n\n\n### 功能特性\n\n* 添加信息收集功能 ([#233](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F233)) ([89f4af9](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F89f4af90fb4d95a0689bf9efc8ffd9326469c0aa))\n* 为 Kaggle 场景添加交叉验证功能 ([#236](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F236)) ([e0b03ba](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fe0b03ba6b5c3d9aa552b99d470e106d4e348e64d))\n* 为 Docker 环境添加进度状态显示 ([#215](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F215)) ([538d4ef](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F538d4ef2e52de795b90d3f75b2e1e877ab85c18d))\n* 为 Kaggle 场景添加循环代码。([#211](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F211)) ([975c327](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F975c32715e51aec6b49537401f5fc59115e04a01))\n* 演示展示效果及使用方法 ([#162](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F162)) ([8cf122a](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F8cf122a0155f434fa4477ae7a6d616b5caecd3e0))\n* 框架的试点运行 ([#227](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F227)) ([e9b103e](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fe9b103e684fdd2b98cd1a89971a3fce2d6e884a1))\n* 为 Kaggle 场景支持更多模型 ([#223](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F223)) ([e3a9659](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fe3a96598c0720fe092ec86d7ca8c195c7d6bcc72))\n* 更新 model_experiment.py 以支持基础的 EDA ([#220](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F220)) ([bf2684c](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fbf2684c4d55ab8e1048ac0291695475ad53b0cd6))\n\n\n### Bug 修复\n\n* 修复 LLM 调用中的一些 bug ([#217](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F217)) ([7b010f8](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F7b010f8b5940aba65a58f1d78192aa80bcd0e654))\n* 处理软件包依赖问题。([#234](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F234)) ([46be295](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F46be2952952af534fd8d98a656c704c688d7cbdd))\n* 移除无用行 ([#177](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F177)) ([64e9a8e](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F64e9a8e39a2072a962111db18f5b9565df5b0176))","2024-09-07T04:58:41",{"id":222,"version":223,"summary_zh":224,"released_at":225},305709,"v0.1.0","## [0.1.0](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcompare\u002Fv0.0.1...v0.1.0) (2024-08-09)\n\n\n### 功能\n\n* 添加 rdagent 入口。([#187](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F187)) ([121b6d9](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F121b6d98de38cd03be30cbee47b40baf39a2b60b))\n* 更改 UI 入口。([#197](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F197)) ([fa5d335](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Ffa5d3354d22240888f4fc4007d9834f7424632aa))\n* 移除 PDF 文件并启用在线 PDF 阅读功能。([#183](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F183)) ([18c0501](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F18c05016a23d694c7b12759cf1322562dcffc56a))\n\n\n### 错误修复\n\n* 修复 README 中失效的链接。([#189](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F189)) ([1b89218](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F1b89218f6bc697494f4a1b8a42ad18963002714f))\n* 修复快速入门问题。([#191](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F191)) ([44f61bf](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F44f61bfa1058a8efb59ca48b7f1417765aeea33e))\n* 更新 readme.md 中的命令行示例。([#192](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F192)) ([9c45d24](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F9c45d24a192da02f7d9765cb001097da1bc36c61))","2024-08-09T12:06:50",{"id":227,"version":228,"summary_zh":229,"released_at":230},305710,"v0.0.1","## 0.0.1 (2024-08-08)\n\n\n### Features\n\n* Add description for scenario experiments. ([#174](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F174)) ([fbd8c6d](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Ffbd8c6d87e1424c08997103b8e8fbf264858c4ed))\n* Added QlibFactorFromReportScenario and improved the report-factor loop. ([#161](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F161)) ([882c79b](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F882c79bf11583980e646b130f71cfa20201ffc7b))\n* filter feature which is high correlation to former implemented features ([#145](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F145)) ([e818326](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fe818326422740e04a4863f7c3c18744dde2ad98f))\n* Remove redundant 'key steps' section in frontend scene display. ([#169](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F169)) ([e767005](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fe76700513bee29232c93b97414419df330d9be8d))\n* streamlit webapp demo for different scenarios ([#135](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F135)) ([d8da7db](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fd8da7db865e6653fc4740efee9a843b69bd79699))\n* Uploaded Documentation, Updated Prompts & Some Code for model demo ([#144](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F144)) ([529f935](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F529f935aa98623f0dc1dda29eecee3ef738dd446))\n\n\n### Bug Fixes\n\n* Add framework handling for task coding failure. ([#176](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F176)) ([5e14fa5](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F5e14fa54a9dd30a94aebe2643b8c9a3b85517a11))\n* Comprehensive update to factor extraction. ([#143](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F143)) ([b5ea040](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fb5ea04019fd5fa15c0f8b9a7e4f18f490f7057d4))\n* first round app folder cleaning ([#166](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F166)) ([6a5a750](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F6a5a75021912927deb5e8e4c7ad3ec4b51bfc788))\n* fix pickle problem ([#140](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F140)) ([7ee4258](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F7ee42587b60d94417f34332cee395cf210dc8a0e))\n* fix release CI ([#165](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F165)) ([85d6a5e](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F85d6a5ed91113fda34ae079b23c89aa24acd2cb2))\n* fix release CI error ([#160](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F160)) ([1c9f8ef](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F1c9f8ef287961731944acc9008496b4dddeddca7))\n* fix several bugs in data mining scenario ([#147](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F147)) ([b233380](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fb233380e2c66fb030db39424f0f040c86e37f5c4))\n* fix some small bugs in report-factor loop ([#152](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F152)) ([a79f9f9](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fa79f9f93406aff6305a76e6a6abd3852642e4c62))\n* fix_release_ci_error ([#150](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F150)) ([4f82e99](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F4f82e9960a2638af9d831581185ddd3bac5711fc))\n* Fixed some bugs introduced during refactoring. ([#167](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F167)) ([f8f1445](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Ff8f1445283fb89aefeb2918243c35a219a51a56c))\n* optimize some prompts in factor loop. ([#158](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fissues\u002F158)) ([c2c1330](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002Fc2c13300b9ad315a663ec2d0eada414e56c6f54f))\n\n\n### Miscellaneous Chores\n\n* release 0.0.1 ([1feacd3](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002FRD-Agent\u002Fcommit\u002F1feacd39b21193de11e9bbecf880ddf96d7c261c))","2024-08-08T05:09:15"]