[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-instadeepai--jumanji":3,"tool-instadeepai--jumanji":64},[4,17,27,35,43,56],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":16},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,3,"2026-04-05T11:01:52",[13,14,15],"开发框架","图像","Agent","ready",{"id":18,"name":19,"github_repo":20,"description_zh":21,"stars":22,"difficulty_score":23,"last_commit_at":24,"category_tags":25,"status":16},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",138956,2,"2026-04-05T11:33:21",[13,15,26],"语言模型",{"id":28,"name":29,"github_repo":30,"description_zh":31,"stars":32,"difficulty_score":23,"last_commit_at":33,"category_tags":34,"status":16},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",107662,"2026-04-03T11:11:01",[13,14,15],{"id":36,"name":37,"github_repo":38,"description_zh":39,"stars":40,"difficulty_score":23,"last_commit_at":41,"category_tags":42,"status":16},3704,"NextChat","ChatGPTNextWeb\u002FNextChat","NextChat 是一款轻量且极速的 AI 助手，旨在为用户提供流畅、跨平台的大模型交互体验。它完美解决了用户在多设备间切换时难以保持对话连续性，以及面对众多 AI 模型不知如何统一管理的痛点。无论是日常办公、学习辅助还是创意激发，NextChat 都能让用户随时随地通过网页、iOS、Android、Windows、MacOS 或 Linux 端无缝接入智能服务。\n\n这款工具非常适合普通用户、学生、职场人士以及需要私有化部署的企业团队使用。对于开发者而言，它也提供了便捷的自托管方案，支持一键部署到 Vercel 或 Zeabur 等平台。\n\nNextChat 的核心亮点在于其广泛的模型兼容性，原生支持 Claude、DeepSeek、GPT-4 及 Gemini Pro 等主流大模型，让用户在一个界面即可自由切换不同 AI 能力。此外，它还率先支持 MCP（Model Context Protocol）协议，增强了上下文处理能力。针对企业用户，NextChat 提供专业版解决方案，具备品牌定制、细粒度权限控制、内部知识库整合及安全审计等功能，满足公司对数据隐私和个性化管理的高标准要求。",87618,"2026-04-05T07:20:52",[13,26],{"id":44,"name":45,"github_repo":46,"description_zh":47,"stars":48,"difficulty_score":23,"last_commit_at":49,"category_tags":50,"status":16},2268,"ML-For-Beginners","microsoft\u002FML-For-Beginners","ML-For-Beginners 是由微软推出的一套系统化机器学习入门课程，旨在帮助零基础用户轻松掌握经典机器学习知识。这套课程将学习路径规划为 12 周，包含 26 节精炼课程和 52 道配套测验，内容涵盖从基础概念到实际应用的完整流程，有效解决了初学者面对庞大知识体系时无从下手、缺乏结构化指导的痛点。\n\n无论是希望转型的开发者、需要补充算法背景的研究人员，还是对人工智能充满好奇的普通爱好者，都能从中受益。课程不仅提供了清晰的理论讲解，还强调动手实践，让用户在循序渐进中建立扎实的技能基础。其独特的亮点在于强大的多语言支持，通过自动化机制提供了包括简体中文在内的 50 多种语言版本，极大地降低了全球不同背景用户的学习门槛。此外，项目采用开源协作模式，社区活跃且内容持续更新，确保学习者能获取前沿且准确的技术资讯。如果你正寻找一条清晰、友好且专业的机器学习入门之路，ML-For-Beginners 将是理想的起点。",84991,"2026-04-05T10:45:23",[14,51,52,53,15,54,26,13,55],"数据工具","视频","插件","其他","音频",{"id":57,"name":58,"github_repo":59,"description_zh":60,"stars":61,"difficulty_score":10,"last_commit_at":62,"category_tags":63,"status":16},3128,"ragflow","infiniflow\u002Fragflow","RAGFlow 是一款领先的开源检索增强生成（RAG）引擎，旨在为大语言模型构建更精准、可靠的上下文层。它巧妙地将前沿的 RAG 技术与智能体（Agent）能力相结合，不仅支持从各类文档中高效提取知识，还能让模型基于这些知识进行逻辑推理和任务执行。\n\n在大模型应用中，幻觉问题和知识滞后是常见痛点。RAGFlow 通过深度解析复杂文档结构（如表格、图表及混合排版），显著提升了信息检索的准确度，从而有效减少模型“胡编乱造”的现象，确保回答既有据可依又具备时效性。其内置的智能体机制更进一步，使系统不仅能回答问题，还能自主规划步骤解决复杂问题。\n\n这款工具特别适合开发者、企业技术团队以及 AI 研究人员使用。无论是希望快速搭建私有知识库问答系统，还是致力于探索大模型在垂直领域落地的创新者，都能从中受益。RAGFlow 提供了可视化的工作流编排界面和灵活的 API 接口，既降低了非算法背景用户的上手门槛，也满足了专业开发者对系统深度定制的需求。作为基于 Apache 2.0 协议开源的项目，它正成为连接通用大模型与行业专有知识之间的重要桥梁。",77062,"2026-04-04T04:44:48",[15,14,13,26,54],{"id":65,"github_repo":66,"name":67,"description_en":68,"description_zh":69,"ai_summary_zh":69,"readme_en":70,"readme_zh":71,"quickstart_zh":72,"use_case_zh":73,"hero_image_url":74,"owner_login":75,"owner_name":76,"owner_avatar_url":77,"owner_bio":78,"owner_company":79,"owner_location":79,"owner_email":80,"owner_twitter":75,"owner_website":81,"owner_url":82,"languages":83,"stars":92,"forks":93,"last_commit_at":94,"license":95,"difficulty_score":23,"env_os":96,"env_gpu":97,"env_ram":98,"env_deps":99,"category_tags":112,"github_topics":113,"view_count":23,"oss_zip_url":79,"oss_zip_packed_at":79,"status":16,"created_at":117,"updated_at":118,"faqs":119,"releases":148},1233,"instadeepai\u002Fjumanji","jumanji","🕹️ A diverse suite of scalable reinforcement learning environments in JAX","jumanji 是一个基于 JAX 的强化学习环境套件，提供多种可扩展的模拟场景，帮助开发者和研究人员训练和测试强化学习算法。它包含从经典游戏（如 2048、贪吃蛇）到现实问题（如路径规划、装箱、图着色）的各种环境，适用于不同领域的研究与应用。\n\n这个工具解决了强化学习研究中环境多样性不足的问题，为算法开发提供了丰富的测试场景。用户可以通过这些环境快速验证新算法的性能，或用于教学演示。其模块化设计使得自定义环境变得简单，适合需要灵活实验的研究人员和开发者。\n\njumanji 特别适合从事强化学习研究的开发者和学术人员使用，也适合对 AI 模拟环境感兴趣的技术爱好者。它基于 JAX 构建，支持 GPU\u002FTPU 加速，具备良好的可扩展性和高性能计算能力，是进行大规模强化学习实验的理想选择。","\u003Cp align=\"center\">\n    \u003Ca href=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_01629a499edc.png\">\n        \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_01629a499edc.png\" alt=\"Jumanji logo\" width=\"50%\"\u002F>\n    \u003C\u002Fa>\n\u003C\u002Fp>\n\n[![Python Versions](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fpyversions\u002Fjumanji.svg?style=flat-square)](https:\u002F\u002Fwww.python.org\u002Fdoc\u002Fversions\u002F)\n[![PyPI Version](https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Fjumanji.svg)](https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Fjumanji)\n[![Tests](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Factions\u002Fworkflows\u002Ftests_linters.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Factions\u002Fworkflows\u002Ftests_linters.yml)\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[![MyPy](http:\u002F\u002Fwww.mypy-lang.org\u002Fstatic\u002Fmypy_badge.svg)](http:\u002F\u002Fmypy-lang.org\u002F)\n[![License](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-Apache%202.0-orange.svg)](https:\u002F\u002Fopensource.org\u002Flicenses\u002FApache-2.0)\n[![Hugging Face](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97-Hugging%20Face-F8D521)](https:\u002F\u002Fhuggingface.co\u002FInstaDeepAI)\n\n[**Environments**](#environments)\n| [**Installation**](#install)\n| [**Quickstart**](#quickstart)\n| [**Training**](#training)\n| [**Citation**](#citing)\n| [**Docs**](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji)\n---\n\n\u003Cdiv class=\"collage\">\n  \u003Cdiv class=\"row\" align=\"center\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_eed5a632d248.gif\" alt=\"BinPack\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_be426323c0ec.gif\" alt=\"Cleaner\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_62a67bb360bf.gif\" alt=\"Connector\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_e71aa016927b.gif\" alt=\"CVRP\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_1a812078b9bb.gif\" alt=\"FlatPack\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_92f6821f0070.gif\" alt=\"Game2048\" width=\"16%\">\n  \u003C\u002Fdiv>\n  \u003Cdiv class=\"row\" align=\"center\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_7a38e5cf915e.gif\" alt=\"GraphColoring\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_85e66ddc4cf6.gif\" alt=\"JobShop\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_aa2ceac83473.gif\" alt=\"Knapsack\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_7f9b9d587069.gif\" alt=\"Maze\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_0fa16a9b94aa.gif\" alt=\"Minesweeper\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_86749c3e4e03.gif\" alt=\"MMST\" width=\"16%\">\n  \u003C\u002Fdiv>\n  \u003Cdiv class=\"row\" align=\"center\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_36d61e99bb64.gif\" alt=\"MultiCVRP\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_08a921d8e311.gif\" alt=\"PacMan\" width=\"12.9%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_de324e8b20f3.gif\" alt=\"RobotWarehouse\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_dd50832b2a68.gif\" alt=\"RubiksCube\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_b97c2a9123e0.gif\" alt=\"SlidingTilePuzzle\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_700e007acabc.gif\" alt=\"Snake\" width=\"16%\">\n  \u003C\u002Fdiv>\n    \u003Cdiv class=\"row\" align=\"center\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_fa6ebb72bf4e.gif\" alt=\"RobotWarehouse\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_0469e26a3ffb.gif\" alt=\"Sudoku\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_9e04a2069a0b.gif\" alt=\"Tetris\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_ddcf186c4524.gif\" alt=\"Tetris\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_4c13ad67909d.gif\" alt=\"Level-Based Foraging\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_4428cfda825d.gif\" alt=\"Search and Rescue\" width=\"16%\">\n  \u003C\u002Fdiv>\n\u003C\u002Fdiv>\n\n## Jumanji @ ICLR 2024\n\nJumanji has been accepted at [ICLR 2024](https:\u002F\u002Ficlr.cc\u002F), check out our [research paper](https:\u002F\u002Farxiv.org\u002Fabs\u002F2306.09884).\n\n## Welcome to the Jungle! 🌴\n\nJumanji is a diverse suite of scalable reinforcement learning environments written in JAX. It now features 22 environments!\n\nJumanji is helping pioneer a new wave of hardware-accelerated research and development in the\nfield of RL. Jumanji's high-speed environments enable faster iteration and large-scale\nexperimentation while simultaneously reducing complexity. Originating in the research team at\n[InstaDeep](https:\u002F\u002Fwww.instadeep.com\u002F), Jumanji is now developed jointly with the open-source\ncommunity. To join us in these efforts, reach out, raise issues and read our\n[contribution guidelines](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fblob\u002Fmain\u002FCONTRIBUTING.md) or just\n[star](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji) 🌟 to stay up to date with the latest developments!\n\n### Goals 🚀\n\n1. Provide a simple, well-tested API for JAX-based environments.\n2. Make research in RL more accessible.\n3. Facilitate the research on RL for problems in the industry and help close the gap between\nresearch and industrial applications.\n4. Provide environments whose difficulty can be scaled to be arbitrarily hard.\n\n### Overview 🦜\n\n- 🥑 **Environment API**: core abstractions for JAX-based environments.\n- 🕹️ **Environment Suite**: a collection of RL environments ranging from simple games to NP-hard\ncombinatorial problems.\n- 🍬 **Wrappers**: easily connect to your favourite RL frameworks and libraries such as\n[Acme](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Facme),\n[Stable Baselines3](https:\u002F\u002Fgithub.com\u002FDLR-RM\u002Fstable-baselines3),\n[RLlib](https:\u002F\u002Fdocs.ray.io\u002Fen\u002Flatest\u002Frllib\u002Findex.html), [Gymnasium](https:\u002F\u002Fgithub.com\u002FFarama-Foundation\u002FGymnasium)\nand [DeepMind-Env](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fdm_env) through our `dm_env` and `gym` wrappers.\n- 🎓 **Examples**: guides to facilitate Jumanji's adoption and highlight the added value of\nJAX-based environments.\n- 🏎️ **Training:** example agents that can be used as inspiration for the agents one may implement\nin their research.\n\n\u003Ch2 name=\"environments\" id=\"environments\">Environments 🌍\u003C\u002Fh2>\n\nJumanji provides a diverse range of environments ranging from simple games to NP-hard combinatorial\nproblems.\n\n| Environment                              | Category | Registered Version(s)                                | Source                                                                                           | Description                                                            |\n|------------------------------------------|----------|------------------------------------------------------|--------------------------------------------------------------------------------------------------|------------------------------------------------------------------------|\n| 🔢 Game2048                              | Logic  | `Game2048-v1`                                        | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Flogic\u002Fgame_2048\u002F)   | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fgame_2048\u002F)   |\n| 🎨 GraphColoring                              | Logic  | `GraphColoring-v1`                                   | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Flogic\u002Fgraph_coloring\u002F)   | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fgraph_coloring\u002F)   |\n| 💣 Minesweeper                           | Logic    | `Minesweeper-v0`                                     | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Flogic\u002Fminesweeper\u002F) | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fminesweeper\u002F) |\n| 🎲 RubiksCube                            | Logic    | `RubiksCube-v0`\u003Cbr\u002F>`RubiksCube-partly-scrambled-v0` | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Flogic\u002Frubiks_cube\u002F) | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Frubiks_cube\u002F) |\n| 🔀 SlidingTilePuzzle                       | Logic    | `SlidingTilePuzzle-v0` | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Flogic\u002Fsliding_tile_puzzle\u002F) | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fsliding_tile_puzzle\u002F) |\n| ✏️ Sudoku                       | Logic    | `Sudoku-v0` \u003Cbr\u002F>`Sudoku-very-easy-v0`| [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Flogic\u002Fsudoku\u002F) | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fsudoku\u002F) |\n| 📦 BinPack (3D BinPacking Problem)       | Packing  | `BinPack-v1`                                         | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Fpacking\u002Fbin_pack\u002F)  | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fbin_pack\u002F)    |\n| 🧩 FlatPack (2D Grid Filling Problem) | Packing  | `FlatPack-v0`                                         | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Fpacking\u002Fflat_pack\u002F)  | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fflat_pack\u002F)    |\n| 🏭 JobShop (Job Shop Scheduling Problem) | Packing  | `JobShop-v0`                                         | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Fpacking\u002Fjob_shop\u002F)  | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fjob_shop\u002F)    |\n| 🎒 Knapsack                              | Packing  | `Knapsack-v1`                                        | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Fpacking\u002Fknapsack\u002F)  | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fknapsack\u002F)    |\n| ▒ Tetris                              | Packing  | `Tetris-v0`                                        | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Fpacking\u002Ftetris\u002F)  | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Ftetris\u002F)    |\n| 🧹 Cleaner                               | Routing  | `Cleaner-v0`                                         | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Frouting\u002Fcleaner\u002F)   | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fcleaner\u002F)     |\n| :link: Connector                         | Routing  | `Connector-v3`                                       | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Frouting\u002Fconnector\u002F) | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fconnector\u002F)   |\n| 🚚 CVRP (Capacitated Vehicle Routing Problem)  | Routing  | `CVRP-v1`                                            | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Frouting\u002Fcvrp\u002F)      | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fcvrp\u002F)        |\n| 🚚 MultiCVRP (Multi-Agent Capacitated Vehicle Routing Problem)  | Routing  | `MultiCVRP-v0`                                            | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Frouting\u002Fmulti_cvrp\u002F)      | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fmulti_cvrp\u002F)        |\n| :mag: Maze   | Routing  | `Maze-v0`                                            | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Frouting\u002Fmaze\u002F)      | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fmaze\u002F)        |\n| :robot: RobotWarehouse  | Routing  | `RobotWarehouse-v0`                                  | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Frouting\u002Frobot_warehouse\u002F)      | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Frobot_warehouse\u002F)        |\n| 🐍 Snake                                       | Routing  | `Snake-v1`                                           | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Frouting\u002Fsnake\u002F)     | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fsnake\u002F)       |\n| 📬 TSP (Travelling Salesman Problem)           | Routing  | `TSP-v1`                                             | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Frouting\u002Ftsp\u002F)       | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Ftsp\u002F)         |\n| Multi Minimum Spanning Tree Problem | Routing  | `MMST-v0`                                | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Frouting\u002Fmmst)    | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fmmst\u002F)    |\n| ᗧ•••ᗣ•• PacMan   | Routing  | `PacMan-v1`                                            | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Frouting\u002Fpac_man\u002F)      | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fpac_man\u002F)\n| 👾 Sokoban                                                     | Routing  | `Sokoban-v0`                                         | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Frouting\u002Fsokoban\u002F)          | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fsokoban\u002F)         |\n| 🍎 Level-Based Foraging                                                     | Routing  | `LevelBasedForaging-v0`                                         | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Frouting\u002Flbf\u002F)          | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Flbf\u002F)         |\n| 🚁 Search and Rescue                                                     | Swarms  | `SearchAndRescue-v0`                                         | [code](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Fswarms\u002Fsearch_and_rescue\u002F)          | [doc](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fsearch_and_rescue\u002F)         |\n\n\u003Ch2 name=\"install\" id=\"install\">Installation 🎬\u003C\u002Fh2>\n\nYou can install the latest release of Jumanji from PyPI:\n\n```bash\npip install -U jumanji\n```\n\nAlternatively, you can install the latest development version directly from GitHub:\n\n```bash\npip install git+https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji.git\n```\n\nJumanji has been tested on Python 3.10, 3.11 and 3.12.\nNote that because the installation of JAX differs depending on your hardware accelerator,\nwe advise users to explicitly install the correct JAX version (see the\n[official installation guide](https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fjax#installation)).\n\n**Rendering:** Matplotlib is used for rendering all the environments. To visualize the environments\nyou will need a GUI backend. For example, on Linux, you can install Tk via:\n`apt-get install python3-tk`, or using conda: `conda install tk`. Check out\n[Matplotlib backends](https:\u002F\u002Fmatplotlib.org\u002Fstable\u002Fusers\u002Fexplain\u002Fbackends.html) for a list of\nbackends you can use.\n\n\u003Ch2 name=\"quickstart\" id=\"quickstart\">Quickstart ⚡\u003C\u002Fh2>\n\nRL practitioners will find Jumanji's interface familiar as it combines the widely adopted\n[OpenAI Gym](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fgym) and\n[DeepMind Environment](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fdm_env) interfaces. From OpenAI Gym, we adopted\nthe idea of a `registry` and the `render` method, while our `TimeStep` structure is inspired by\nDeepMind Environment.\n\n### Basic Usage 🧑‍💻\n\n```python\nimport jax\nimport jumanji\n\n# Instantiate a Jumanji environment using the registry\nenv = jumanji.make('Snake-v1')\n\n# Reset your (jit-able) environment\nkey = jax.random.PRNGKey(0)\nstate, timestep = jax.jit(env.reset)(key)\n\n# (Optional) Render the env state\nenv.render(state)\n\n# Interact with the (jit-able) environment\naction = env.action_spec.generate_value()          # Action selection (dummy value here)\nstate, timestep = jax.jit(env.step)(state, action)   # Take a step and observe the next state and time step\n```\n\n- `state` represents the internal state of the environment: it contains all the information required\nto take a step when executing an action. This should **not** be confused with the `observation`\ncontained in the `timestep`, which is the information perceived by the agent.\n- `timestep` is a dataclass containing `step_type`, `reward`, `discount`, `observation` and\n`extras`. This structure is similar to\n[`dm_env.TimeStep`](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fdm_env\u002Fblob\u002Fmaster\u002Fdocs\u002Findex.md) except for the\n`extras` field that was added to allow users to log environments metrics that are neither part of\nthe agent's observation nor part of the environment's internal state.\n\n### Advanced Usage 🧑‍🔬\n\nBeing written in JAX, Jumanji's environments benefit from many of its features including\nautomatic vectorization\u002Fparallelization (`jax.vmap`, `jax.pmap`) and JIT-compilation (`jax.jit`),\nwhich can be composed arbitrarily.\nWe provide an example of a more advanced usage in the\n[advanced usage guide](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fguides\u002Fadvanced_usage\u002F).\n\n### Registry and Versioning 📖\n\nLike OpenAI Gym, Jumanji keeps a strict versioning of its environments for reproducibility reasons.\nWe maintain a registry of standard environments with their configuration.\nFor each environment, a version suffix is appended, e.g. `Snake-v1`.\nWhen changes are made to environments that might impact learning results,\nthe version number is incremented by one to prevent potential confusion.\nFor a full list of registered versions of each environment, check out\n[the documentation](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Ftsp\u002F).\n\n\u003Ch2 name=\"training\" id=\"training\">Training 🏎️\u003C\u002Fh2>\n\nTo showcase how to train RL agents on Jumanji environments, we provide a random agent and a vanilla\nactor-critic (A2C) agent. These agents can be found in\n[jumanji\u002Ftraining\u002F](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Ftraining\u002F).\n\nBecause the environment framework in Jumanji is so flexible, it allows pretty much any problem to\nbe implemented as a Jumanji environment, giving rise to very diverse observations. For this reason,\nenvironment-specific networks are required to capture the symmetries of each environment.\nAlongside the A2C agent implementation, we provide examples of such environment-specific\nactor-critic networks in\n[jumanji\u002Ftraining\u002Fnetworks](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Ftraining\u002Fnetworks\u002F).\n\n> ⚠️ The example agents in `jumanji\u002Ftraining` are **only** meant to serve as inspiration for how one\n> can implement an agent. Jumanji is first and foremost a library of environments - as such, the\n> agents and networks will **not** be maintained to a production standard.\n\nFor more information on how to use the example agents, see the\n[training guide](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fguides\u002Ftraining\u002F).\n\n## Contributing 🤝\n\nContributions are welcome! See our issue tracker for\n[good first issues](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Flabels\u002Fgood%20first%20issue). Please read\nour [contributing guidelines](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fblob\u002Fmain\u002FCONTRIBUTING.md) for\ndetails on how to submit pull requests, our Contributor License Agreement, and community guidelines.\n\n\u003Ch2 name=\"citing\" id=\"citing\">Citing Jumanji ✏️\u003C\u002Fh2>\n\nIf you use Jumanji in your work, please cite the library using:\n\n```\n@misc{bonnet2024jumanji,\n    title={Jumanji: a Diverse Suite of Scalable Reinforcement Learning Environments in JAX},\n    author={Clément Bonnet and Daniel Luo and Donal Byrne and Shikha Surana and Sasha Abramowitz and Paul Duckworth and Vincent Coyette and Laurence I. Midgley and Elshadai Tegegn and Tristan Kalloniatis and Omayma Mahjoub and Matthew Macfarlane and Andries P. Smit and Nathan Grinsztajn and Raphael Boige and Cemlyn N. Waters and Mohamed A. Mimouni and Ulrich A. Mbou Sob and Ruan de Kock and Siddarth Singh and Daniel Furelos-Blanco and Victor Le and Arnu Pretorius and Alexandre Laterre},\n    year={2024},\n    eprint={2306.09884},\n    url={https:\u002F\u002Farxiv.org\u002Fabs\u002F2306.09884},\n    archivePrefix={arXiv},\n    primaryClass={cs.LG}\n}\n```\n\n## See Also 🔎\n\nOther works have embraced the approach of writing RL environments in JAX.\nIn particular, we suggest users check out the following sister repositories:\n\n- 🤖 [Qdax](https:\u002F\u002Fgithub.com\u002Fadaptive-intelligent-robotics\u002FQDax) is a library to accelerate\nQuality-Diversity and neuro-evolution algorithms through hardware accelerators and parallelization.\n- 🌳 [Evojax](https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fevojax) provides tools to enable neuroevolution algorithms\nto work with neural networks running across multiple TPU\u002FGPUs.\n- 🦾 [Brax](https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fbrax) is a differentiable physics engine that simulates\nenvironments made up of rigid bodies, joints, and actuators.\n- 🏋️‍ [Gymnax](https:\u002F\u002Fgithub.com\u002FRobertTLange\u002Fgymnax) implements classic environments including\nclassic control, bsuite, MinAtar and a collection of meta RL tasks.\n- 🎲 [Pgx](https:\u002F\u002Fgithub.com\u002Fsotetsuk\u002Fpgx) provides classic board game environments like\nBackgammon, Shogi, and Go.\n\n## Acknowledgements 🙏\n\nThe development of this library was supported with Cloud TPUs\nfrom Google's [TPU Research Cloud](https:\u002F\u002Fsites.research.google\u002Ftrc\u002Fabout\u002F) (TRC) 🌤.\n","\u003Cp align=\"center\">\n    \u003Ca href=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_01629a499edc.png\">\n        \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_01629a499edc.png\" alt=\"Jumanji logo\" width=\"50%\"\u002F>\n    \u003C\u002Fa>\n\u003C\u002Fp>\n\n[![Python版本](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fpyversions\u002Fjumanji.svg?style=flat-square)](https:\u002F\u002Fwww.python.org\u002Fdoc\u002Fversions\u002F)\n[![PyPI版本](https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Fjumanji.svg)](https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Fjumanji)\n[![测试](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Factions\u002Fworkflows\u002Ftests_linters.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Factions\u002Fworkflows\u002Ftests_linters.yml)\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[![MyPy](http:\u002F\u002Fwww.mypy-lang.org\u002Fstatic\u002Fmypy_badge.svg)](http:\u002F\u002Fmypy-lang.org\u002F)\n[![许可证](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-Apache%202.0-orange.svg)](https:\u002F\u002Fopensource.org\u002Flicenses\u002FApache-2.0)\n[![Hugging Face](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97-Hugging%20Face-F8D521)](https:\u002F\u002Fhuggingface.co\u002FInstaDeepAI)\n\n[**环境**](#environments)\n| [**安装**](#install)\n| [**快速入门**](#quickstart)\n| [**训练**](#training)\n| [**引用**](#citing)\n| [**文档**](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji)\n---\n\n\u003Cdiv class=\"collage\">\n  \u003Cdiv class=\"row\" align=\"center\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_eed5a632d248.gif\" alt=\"BinPack\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_be426323c0ec.gif\" alt=\"Cleaner\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_62a67bb360bf.gif\" alt=\"Connector\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_e71aa016927b.gif\" alt=\"CVRP\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_1a812078b9bb.gif\" alt=\"FlatPack\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_92f6821f0070.gif\" alt=\"Game2048\" width=\"16%\">\n  \u003C\u002Fdiv>\n  \u003Cdiv class=\"row\" align=\"center\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_7a38e5cf915e.gif\" alt=\"GraphColoring\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_85e66ddc4cf6.gif\" alt=\"JobShop\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_aa2ceac83473.gif\" alt=\"Knapsack\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_7f9b9d587069.gif\" alt=\"Maze\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_0fa16a9b94aa.gif\" alt=\"Minesweeper\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_86749c3e4e03.gif\" alt=\"MMST\" width=\"16%\">\n  \u003C\u002Fdiv>\n  \u003Cdiv class=\"row\" align=\"center\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_36d61e99bb64.gif\" alt=\"MultiCVRP\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_08a921d8e311.gif\" alt=\"PacMan\" width=\"12.9%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_de324e8b20f3.gif\" alt=\"RobotWarehouse\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_dd50832b2a68.gif\" alt=\"RubiksCube\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_readme_b97c2a9123e0.gif\" alt=\"SlidingTilePuzzle\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Finstadeepai jumanji main docs env anim snake.gif\" alt=\"Snake\" width=\"16%\">\n  \u003C\u002Fdiv>\n    \u003Cdiv class=\"row\" align=\"center\">\n    \u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Finstadeepai jumanji main docs env anim sokoban.gif\" alt=\"RobotWarehouse\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Finstadeepai jumanji main docs env anim sudoku.gif\" alt=\"Sudoku\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Finstadeepai jumanji main docs env anim tetris.gif\" alt=\"Tetris\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Finstadeepai jumanji main docs env anim tsp.gif\" alt=\"Tetris\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Finstadeepai jumanji main docs env anim lbf.gif\" alt=\"Level-Based Foraging\" width=\"16%\">\n    \u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Finstadeepai jumanji main docs env anim search and rescue.gif\" alt=\"Search and Rescue\" width=\"16%\">\n  \u003C\u002Fdiv>\n\u003C\u002Fdiv>\n\n## Jumanji @ ICLR 2024\n\nJumanji已被接受至[ICLR 2024](https:\u002F\u002Ficlr.cc\u002F)，请查看我们的[研究论文](https:\u002F\u002Farxiv.org\u002Fabs\u002F2306.09884)。\n\n## 欢迎来到丛林！🌴\n\nJumanji是一套用JAX编写的多样化、可扩展的强化学习环境。目前已有22个环境！\n\nJumanji正助力开创RL领域硬件加速研发的新浪潮。Jumanji的高速环境不仅能够加快迭代速度、支持大规模实验，还能同时降低复杂度。Jumanji起源于[InstaDeep](https:\u002F\u002Fwww.instadeep.com\u002F)的研究团队，如今已与开源社区共同开发。如您希望加入我们的行列，请随时联系我们、提交问题并阅读我们的[贡献指南](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fblob\u002Fmain\u002FCONTRIBUTING.md)，或直接[加星](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji) 🌟，以及时了解最新进展！\n\n### 目标🚀\n\n1. 为基于JAX的环境提供简单、经过充分测试的API。\n2. 让RL研究更加普及。\n3. 促进工业界RL问题的研究，帮助缩小研究与工业应用之间的差距。\n4. 提供难度可任意调整的环境。\n\n### 概览🦜\n\n- 🥑 **环境API**：基于JAX环境的核心抽象。\n- 🕹️ **环境套件**：涵盖从简单游戏到NP难组合问题的各类RL环境。\n- 🍬 **封装器**：通过我们的`dm_env`和`gym`封装器，轻松对接您喜爱的RL框架与库，如[Acme](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Facme)、\n[Stable Baselines3](https:\u002F\u002Fgithub.com\u002FDLR-RM\u002Fstable-baselines3)、\n[RLlib](https:\u002F\u002Fdocs.ray.io\u002Fen\u002Flatest\u002Frllib\u002Findex.html)、[Gymnasium](https:\u002F\u002Fgithub.com\u002FFarama-Foundation\u002FGymnasium)\n以及[DeepMind-Env](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fdm_env)。\n- 🎓 **示例**：指导手册，便于Jumanji的采用，并突出基于JAX环境的附加价值。\n- 🏎️ **训练**：示例智能体，可作为您在研究中实现智能体的灵感来源。\n\n\u003Ch2 name=\"environments\" id=\"environments\">环境🌍\u003C\u002Fh2>\n\nJumanji提供了多样化的环境，涵盖从简单游戏到NP难组合问题。\n\n| 环境                              | 类别 | 注册版本                                | 源代码                                                                                           | 描述                                                            |\n|------------------------------------------|----------|------------------------------------------------------|--------------------------------------------------------------------------------------------------|------------------------------------------------------------------------|\n| 🔢 Game2048                              | 逻辑  | `Game2048-v1`                                        | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Flogic\u002Fgame_2048\u002F)   | [文档](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fgame_2048\u002F)   |\n| 🎨 GraphColoring                              | 逻辑  | `GraphColoring-v1`                                   | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Flogic\u002Fgraph_coloring\u002F)   | [文档](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fgraph_coloring\u002F)   |\n| 💣 Minesweeper                           | 逻辑    | `Minesweeper-v0`                                     | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Flogic\u002Fminesweeper\u002F) | [文档](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fminesweeper\u002F) |\n| 🎲 RubiksCube                            | 逻辑    | `RubiksCube-v0`\u003Cbr\u002F>`RubiksCube-partly-scrambled-v0` | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Flogic\u002Frubiks_cube\u002F) | [文档](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Frubiks_cube\u002F) |\n| 🔀 SlidingTilePuzzle                       | 逻辑    | `SlidingTilePuzzle-v0` | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Flogic\u002Fsliding_tile_puzzle\u002F) | [文档](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fsliding_tile_puzzle\u002F) |\n| ✏️ Sudoku                       | 逻辑    | `Sudoku-v0` \u003Cbr\u002F>`Sudoku-very-easy-v0`| [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Flogic\u002Fsudoku\u002F) | [文档](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fsudoku\u002F) |\n| 📦 BinPack (3D BinPacking Problem)       | 包装  | `BinPack-v1`                                         | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Fenvironments\u002Fpacking\u002Fbin_pack\u002F)  | [文档](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Fbin_pack\u002F)    |\n| 🧩 FlatPack (2D Grid Filling Problem) | 包装  | `FlatPack-v0`                                         | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain jumanji en environments packing flat_pack\u002F)  | [文档](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji en environments flat_pack\u002F)    |\n| 🏭 JobShop (Job Shop Scheduling Problem) | 包装  | `JobShop-v0`                                         | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji tree main jumanji en environments packing job_shop\u002F)  | [文档](https:\u002F\u002Finstadeepai.github.io jumanji en environments job_shop\u002F)    |\n| 🎒 Knapsack                              | 包装  | `Knapsack-v1`                                        | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai jumanji tree main jumanji en environments packing knapsack\u002F)  | [文档](https:\u002F\u002Finstadeepai.github.io jumanji en environments knapsack\u002F)    |\n| ▒ Tetris                              | 包装  | `Tetris-v0`                                        | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai jumanji tree main jumanji en environments packing tetris\u002F)  | [文档](https:\u002F\u002Finstadeepai.github.io jumanji en environments tetris\u002F)    |\n| 🧹 Cleaner                               | 路由  | `Cleaner-v0`                                         | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai jumanji tree main jumanji en environments routing cleaner\u002F)   | [文档](https:\u002F\u002Finstadeepai.github.io jumanji en environments cleaner\u002F)     |\n| :link: Connector                         | 路由  | `Connector-v3`                                       | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai jumanji tree main jumanji en environments routing connector\u002F) | [文档](https:\u002F\u002Finstadeepai.github.io jumanji en environments connector\u002F)   |\n| 🚚 CVRP (Capacitated Vehicle Routing Problem)  | 路由  | `CVRP-v1`                                            | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai jumanji tree main jumanji en environments routing cvrp\u002F)      | [文档](https:\u002F\u002Finstadeepai.github.io jumanji en environments cvrp\u002F)        |\n| 🚚 MultiCVRP (Multi-Agent Capacitated Vehicle Routing Problem)  | 路由  | `MultiCVRP-v0`                                            | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai jumanji tree main jumanji en environments routing multi_cvrp\u002F)      | [文档](https:\u002F\u002Finstadeepai.github.io jumanji en environments multi_cvrp\u002F)        |\n| :mag: Maze   | 路由  | `Maze-v0`                                            | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai jumanji tree main jumanji en environments routing maze\u002F)      | [文档](https:\u002F\u002Finstadeepai.github.io jumanji en environments maze\u002F)        |\n| :robot: RobotWarehouse  | 路由  | `RobotWarehouse-v0`                                  | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai jumanji tree main jumanji en environments routing robot_warehouse\u002F)      | [文档](https:\u002F\u002Finstadeepai.github.io jumanji en environments robot_warehouse\u002F)        |\n| 🐍 Snake                                       | 路由  | `Snake-v1`                                           | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai jumanji tree main jumanji en environments routing snake\u002F)     | [文档](https:\u002F\u002Finstadeepai.github.io jumanji en environments snake\u002F)       |\n| 📬 TSP (Travelling Salesman Problem)           | 路由  | `TSP-v1`                                             | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai jumanji tree main jumanji en environments routing tsp\u002F)       | [文档](https:\u002F\u002Finstadeepai.github.io jumanji en environments tsp\u002F)         |\n| 多重最小生成树问题 | 路由  | `MMST-v0`                                | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai jumanji tree main jumanji en environments routing mmst)    | [文档](https:\u002F\u002Finstadeepai.github.io jumanji en environments mmst\u002F)    |\n| ᗧ•••ᗣ•• PacMan   | 路由  | `PacMan-v1`                                            | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai jumanji tree main jumanji en environments routing pac_man\u002F)      | [文档](https:\u002F\u002Finstadeepai.github.io jumanji en environments pac_man\u002F)\n| 👾 Sokoban                                                     | 路由  | `Sokoban-v0`                                         | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai jumanji tree main jumanji en environments routing sokoban\u002F)          | [文档](https:\u002F\u002Finstadeepai.github.io jumanji en environments sokoban\u002F)         |\n| 🍎 Level-Based Foraging                                                     | 路由  | `LevelBasedForaging-v0`                                         | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai jumanji tree main jumanji en environments routing lbf\u002F)          | [文档](https:\u002F\u002Finstadeepai.github.io jumanji en environments lbf\u002F)         |\n| 🚁 Search and Rescue                                                     | 群体  | `SearchAndRescue-v0`                                         | [代码](https:\u002F\u002Fgithub.com\u002Finstadeepai jumanji tree main jumanji en environments swarms search_and_rescue\u002F)          | [文档](https:\u002F\u002Finstadeepai.github.io jumanji en environments search_and_rescue\u002F)         |\n\n\u003Ch2 name=\"install\" id=\"install\">安装 🎬\u003C\u002Fh2>\n\n您可以通过 PyPI 安装 Jumanji 的最新版本：\n\n```bash\npip install -U jumanji\n```\n\n或者，您也可以直接从 GitHub 安装最新的开发版本：\n\n```bash\npip install git+https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji.git\n```\n\nJumanji 已在 Python 3.10、3.11 和 3.12 上进行了测试。\n请注意，由于 JAX 的安装因您的硬件加速器而异，\n我们建议用户显式安装正确的 JAX 版本（参见\n[官方安装指南](https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fjax#installation)）。\n\n**渲染：** 所有环境的渲染均使用 Matplotlib。要可视化这些环境，\n您需要一个 GUI 后端。例如，在 Linux 上，您可以使用以下命令安装 Tk：\n`apt-get install python3-tk`，或通过 conda 安装：`conda install tk`。有关可使用的后端列表，请参阅\n[Matplotlib 后端](https:\u002F\u002Fmatplotlib.org\u002Fstable\u002Fusers\u002Fexplain\u002Fbackends.html)。\n\n\u003Ch2 name=\"quickstart\" id=\"quickstart\">快速入门 ⚡\u003C\u002Fh2>\n\n强化学习从业者会发现 Jumanji 的接口非常熟悉，因为它结合了广泛采用的\n[OpenAI Gym](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fgym) 和\n[DeepMind Environment](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fdm_env) 接口。我们从 OpenAI Gym 中借鉴了\n“注册表”和“渲染”方法的概念，而我们的 `TimeStep` 结构则受到 DeepMind Environment 的启发。\n\n\n\n### 基本用法 🧑‍💻\n\n```python\nimport jax\nimport jumanji\n\n# 使用注册表实例化一个 Jumanji 环境\nenv = jumanji.make('Snake-v1')\n\n# 重置您的（可 JIT 编译的）环境\nkey = jax.random.PRNGKey(0)\nstate, timestep = jax.jit(env.reset)(key)\n\n# （可选）渲染环境状态\nenv.render(state)\n\n# 与（可 JIT 编译的）环境交互\naction = env.action_spec.generate_value()          # 动作选择（此处为虚拟值）\nstate, timestep = jax.jit(env.step)(state, action)   # 执行一步并观察下一个状态和时间步\n```\n\n- `state` 表示环境的内部状态：它包含执行动作时所需的所有信息。\n这不应与 `timestep` 中的 `observation` 混淆，后者是智能体所感知的信息。\n- `timestep` 是一个数据类，包含 `step_type`、`reward`、`discount`、`observation` 和\n`extras`。该结构类似于\n[`dm_env.TimeStep`](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fdm_env\u002Fblob\u002Fmaster\u002Fdocs\u002Findex.md)，只是增加了 `extras` 字段，\n以便用户可以记录既不属于智能体观测也不属于环境内部状态的环境指标。\n\n### 高级用法 🧑‍🔬\n\n由于 Jumanji 是用 JAX 编写的，其环境受益于 JAX 的许多特性，包括自动向量化\u002F并行化（`jax.vmap`、`jax.pmap`）以及 JIT 编译（`jax.jit`），\n这些功能可以任意组合使用。\n我们在\n[高级用法指南](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fguides\u002Fadvanced_usage\u002F) 中提供了一个更高级用法的示例。\n\n### 注册表与版本管理 📖\n\n与 OpenAI Gym 类似，Jumanji 为了保证可重复性，对环境进行了严格的版本管理。\n我们维护了一个标准环境及其配置的注册表。\n对于每个环境，都会附加一个版本后缀，例如 `Snake-v1`。\n当对可能影响学习结果的环境进行更改时，版本号会递增一位，以避免潜在的混淆。\n有关每个环境的完整注册版本列表，请参阅\n[文档](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fenvironments\u002Ftsp\u002F)。\n\n\u003Ch2 name=\"training\" id=\"training\">训练 🏎️\u003C\u002Fh2>\n\n为了展示如何在 Jumanji 环境上训练强化学习智能体，我们提供了随机智能体和经典的演员-评论家（A2C）智能体。这些智能体可以在\n[jumanji\u002Ftraining\u002F](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Ftraining\u002F) 中找到。\n\n由于 Jumanji 中的环境框架非常灵活，几乎任何问题都可以实现为 Jumanji 环境，从而产生非常多样化的观测。因此，需要针对特定环境的网络来捕捉每个环境的对称性。\n除了 A2C 智能体的实现外，我们还在\n[jumanji\u002Ftraining\u002Fnetworks](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Ftree\u002Fmain\u002Fjumanji\u002Ftraining\u002Fnetworks\u002F) 中提供了此类环境特定的演员-评论家网络示例。\n\n> ⚠️ `jumanji\u002Ftraining` 中的示例智能体**仅**旨在为如何实现智能体提供灵感。Jumanji 首先且最重要的是一个环境库——因此，这些智能体和网络**不会**按照生产标准进行维护。\n\n有关如何使用示例智能体的更多信息，请参阅\n[训练指南](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji\u002Fguides\u002Ftraining\u002F)。\n\n## 贡献 🤝\n\n欢迎贡献！请查看我们的问题跟踪器，了解\n[优质首次问题](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Flabels\u002Fgood%20first%20issue)。请阅读我们的\n[贡献指南](https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fblob\u002Fmain\u002FCONTRIBUTING.md)，\n以了解如何提交拉取请求、我们的贡献者许可协议以及社区准则。\n\n\u003Ch2 name=\"citing\" id=\"citing\">引用 Jumanji ✏️\u003C\u002Fh2>\n\n如果您在工作中使用 Jumanji，请按如下方式引用该库：\n\n```\n@misc{bonnet2024jumanji,\n    title={Jumanji：一套基于 JAX 的多样化、可扩展强化学习环境},\n    author={Clément Bonnet、Daniel Luo、Donal Byrne、Shikha Surana、Sasha Abramowitz、Paul Duckworth、Vincent Coyette、Laurence I. Midgley、Elshadai Tegegn、Tristan Kalloniatis、Omayma Mahjoub、Matthew Macfarlane、Andries P. Smit、Nathan Grinsztajn、Raphael Boige、Cemlyn N. Waters、Mohamed A. Mimouni、Ulrich A. Mbou Sob、Ruan de Kock、Siddarth Singh、Daniel Furelos-Blanco、Victor Le、Arnu Pretorius、Alexandre Laterre},\n    year={2024},\n    eprint={2306.09884},\n    url={https:\u002F\u002Farxiv.org\u002Fabs\u002F2306.09884},\n    archivePrefix={arXiv},\n    primaryClass={cs.LG}\n}\n```\n\n## 参见 🔎\n\n其他项目也采用了在 JAX 中编写强化学习环境的方法。\n特别是，我们建议用户查看以下姊妹仓库：\n\n- 🤖 [Qdax](https:\u002F\u002Fgithub.com\u002Fadaptive-intelligent-robotics\u002FQDax) 是一个通过硬件加速器和并行化来加速质量-多样性及神经进化算法的库。\n- 🌳 [Evojax](https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fevojax) 提供工具，使神经进化算法能够与运行在多个 TPU\u002FGPU 上的神经网络协同工作。\n- 🦾 [Brax](https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fbrax) 是一个可微分物理引擎，用于模拟由刚体、关节和执行器组成的环境。\n- 🏋️‍ [Gymnax](https:\u002F\u002Fgithub.com\u002FRobertTLange\u002Fgymnax) 实现了经典环境，包括经典控制、bsuite、MinAtar 以及一系列元强化学习任务。\n- 🎲 [Pgx](https:\u002F\u002Fgithub.com\u002Fsotetsuk\u002Fpgx) 提供经典棋类游戏环境，如西洋双陆棋、将棋和围棋。\n\n## 致谢 🙏\n\n本库的开发得到了谷歌 [TPU 研究云](https:\u002F\u002Fsites.research.google\u002Ftrc\u002Fabout\u002F)（TRC）提供的 Cloud TPU 支持 🌤。","# Jumanji 快速上手指南\n\n---\n\n## 环境准备\n\n### 系统要求\n\n- Python 3.8 或更高版本\n- 支持 JAX 的硬件环境（如 NVIDIA GPU）\n\n### 前置依赖\n\n在安装 Jumanji 之前，请确保已安装以下依赖：\n\n- [JAX](https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fjax)\n- [Optax](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Foptax)\n\n建议使用 `pip` 安装这些依赖，也可以通过 Conda 或其他方式安装。\n\n---\n\n## 安装步骤\n\n你可以通过 `pip` 安装 Jumanji：\n\n```bash\npip install jumanji\n```\n\n如果你在中国，可以使用国内镜像源加速安装：\n\n```bash\npip install jumanji -i https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple\n```\n\n---\n\n## 基本使用\n\n以下是一个最简单的使用示例，展示如何加载并运行一个 Jumanji 环境（以 `Game2048` 为例）：\n\n```python\nimport jumanji\nfrom jumanji.environments.logic.game_2048 import Game2048\n\n# 创建环境实例\nenv = Game2048()\n\n# 重置环境以获取初始状态\nobservation, _ = env.reset()\n\n# 打印初始观察值\nprint(observation)\n\n# 执行一个动作（例如：向右移动）\naction = 1\nnext_observation, reward, done, _ = env.step(action)\n\n# 打印下一步的观察值和奖励\nprint(next_observation)\nprint(reward)\nprint(done)\n```\n\n你也可以使用 `gym` 接口来与 Jumanji 环境交互，只需导入 `gym` 并注册环境即可：\n\n```python\nimport gym\nenv = gym.make(\"Game2048-v1\")\nobservation, _ = env.reset()\n```\n\n---\n\n以上就是 Jumanji 的快速上手指南。更多高级用法和环境信息，可参考 [官方文档](https:\u002F\u002Finstadeepai.github.io\u002Fjumanji)。","某物流科技公司正在研发一款基于强化学习的智能路径规划系统，用于优化配送车辆的路线安排，以降低运输成本并提高配送效率。开发团队需要一个高效、可扩展的强化学习环境来训练和测试他们的算法。\n\n### 没有 jumanji 时  \n- 开发人员需要手动构建复杂的模拟环境，如 CVRP（带容量约束的车辆路径问题），导致开发周期长且容易出错。  \n- 环境缺乏标准化接口，不同任务之间难以复用代码，增加了维护成本。  \n- 训练过程中无法充分利用 GPU 或 TPU 的并行计算能力，导致训练速度缓慢，影响迭代效率。  \n- 缺乏多样化的任务场景，限制了算法的泛化能力和实际应用范围。  \n\n### 使用 jumanji 后  \n- 团队可以直接使用 jumanji 提供的 CVRP 等预定义环境，快速搭建实验平台，节省大量开发时间。  \n- jumanji 提供统一的 API 接口，使得不同任务之间的代码复用率显著提升，降低了维护复杂度。  \n- 借助 JAX 的自动微分和并行计算能力，训练过程加速数倍，提升了模型迭代效率。  \n- jumanji 内置多种任务场景（如 Maze、Game2048、RobotWarehouse 等），帮助团队验证算法在不同领域的适应性。  \n\n核心价值：jumanji 为强化学习研究和应用提供了高效、灵活且可扩展的环境支持，显著提升了开发效率与算法性能。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Finstadeepai_jumanji_be426323.gif","instadeepai","InstaDeep Ltd","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Finstadeepai_f2efb2ea.png","We productionise innovation for the benefit of everyone",null,"hello@instadeep.com","https:\u002F\u002Finstadeep.com","https:\u002F\u002Fgithub.com\u002Finstadeepai",[84,88],{"name":85,"color":86,"percentage":87},"Python","#3572A5",100,{"name":89,"color":90,"percentage":91},"JavaScript","#f1e05a",0,820,95,"2026-03-25T05:05:41","Apache-2.0","Linux, macOS, Windows","需要 NVIDIA GPU，显存 8GB+，CUDA 11.7+","16GB+",{"notes":100,"python":101,"dependencies":102},"建议使用 conda 管理环境，首次运行需下载约 5GB 模型文件","3.8+",[103,104,105,106,107,108,109,110,111],"jax","jaxlib","gymnasium","dm_env","numpy","scipy","pandas","matplotlib","tqdm",[13,54],[103,114,115,116],"python","reinforcement-learning","research","2026-03-27T02:49:30.150509","2026-04-06T07:13:48.159017",[120,125,130,135,139,144],{"id":121,"question_zh":122,"answer_zh":123,"source_url":124},5604,"如何解决 Jumanji 与 chex 的依赖冲突问题？","Jumanji 1.1.0 与 distrax 0.1.5 和 esquilax 2.0.0 存在 chex 版本冲突。建议升级 esquilax 到 2.0.1，该版本放宽了对 chex 的依赖要求。如果仍然遇到问题，可以尝试不固定 jax 和 chex 的版本，仅设置最低版本要求。","https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fissues\u002F276",{"id":126,"question_zh":127,"answer_zh":128,"source_url":129},5605,"如何在 Jumanji 中支持 PyTree 类型的 specs？","Jumanji 当前的 `jumanji_specs_to_dm_env_specs` 函数无法直接处理嵌套的 PyTree 类型 specs。可以通过使用 `jax.tree_map` 对 specs 进行递归转换。但需要注意，Jumanji 使用的是 `NamedTuple` 而不是字典结构，因此需要确保 specs 的结构与环境观察的结构一致。","https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fissues\u002F75",{"id":131,"question_zh":132,"answer_zh":133,"source_url":134},5606,"如何为 Jumanji 的 step 函数添加自定义数据类型（dtype）支持？","可以在 Jumanji 的 step 函数中添加一个 `dtype` 参数，并将其默认值设为 `jnp.float32`。例如，在 `truncation` 函数中添加 `dtype: jnp.dtype = jnp.float32`，并用于初始化 discount 数组。此修改可提高内存效率，适用于需要使用 float16 等其他数据类型的场景。","https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fissues\u002F256",{"id":136,"question_zh":137,"answer_zh":138,"source_url":124},5607,"如何解决 CUDA 和 JAX 版本不匹配导致的 ptxas 错误？","ptxas 错误通常由 CUDA 工具包和 JAX 版本不兼容引起。建议检查当前安装的 CUDA 版本，并确保 JAX 和 JAXLIB 的版本与之匹配。例如，CUDA 12.1 应搭配对应的 JAX 版本。避免手动固定 JAX 或 chex 的版本，而是使用最低版本限制以保持兼容性。",{"id":140,"question_zh":141,"answer_zh":142,"source_url":143},5608,"如何将第三方环境（如 Predator-Prey Flock）集成到 Jumanji？","可以将第三方环境（如 Esquilax 提供的 Predator-Prey Flock）提交给 Jumanji 团队进行集成。维护者会评估其适用性，并可能协助完成代码迁移。确保环境符合 Jumanji 的接口规范，包括使用 JAX 实现和兼容的 specs 结构。","https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fissues\u002F247",{"id":145,"question_zh":146,"answer_zh":147,"source_url":134},5609,"如何处理 Jumanji 中的多智能体环境？","在 Jumanji 中处理多智能体环境时，可以通过指定 `shape` 参数来适配多个智能体的奖励和折扣值。例如，在 `truncation` 函数中使用 `shape` 来定义奖励和折扣的形状，从而支持多智能体的输出格式。",[149,154,159,164,169,174,179,184,189,194,199,204,209,214,219],{"id":150,"version":151,"summary_zh":152,"released_at":153},105237,"v1.1.1","The main addition in this release is the new search and rescue environment, thanks [@zombie-einstein](https:\u002F\u002Fgithub.com\u002Fzombie-einstein)!! :mag: :rescue_worker_helmet: \r\n\r\n## What's Changed\r\n* fix: termination condition by @WiemKhlifi in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F266\r\n* fix: Require jax below 0.4.36 by @zombie-einstein in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F269\r\n* fix(docs): pypi images by @sash-a in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F268\r\n* Implement Search & Rescue Multi-Agent Environment by @zombie-einstein in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F259\r\n* chore: fixes to tests for JAX 0.5 upgrade by @sash-a in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F278\r\n* Matplotlib upgrade by @zombie-einstein in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F270\r\n* Fix bug in Graph Coloring by @VitamintK in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F281\r\n* chore: bump version to 1.1.1 by @sash-a in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F282\r\n\r\n## New Contributors\r\n* @zombie-einstein made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F269\r\n* @VitamintK made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F281\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fcompare\u002Fv1.1.0...v1.1.1","2025-06-18T13:48:58",{"id":155,"version":156,"summary_zh":157,"released_at":158},105238,"v1.1.0","## Release Notes\r\nMany small fixes and upgrades. Notably fixed rendering in certain environments, upgraded gym wrapper to gymnasium and upgraded from python 3.8-3.9 to python 3.10-3.12. Additionally added a fully jittable version of the [Level Based Foraging environment](https:\u002F\u002Fgithub.com\u002Fsemitable\u002Flb-foraging). \r\n\r\n## What's Changed\r\n* fix: pacman ghost valid action calculations result in NaNs by @taodav in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F241\r\n* fix: remove num agents from cleaner init in docs by @sash-a in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F249\r\n* fix: binpack docs by @sash-a in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F250\r\n* chore: fix setup.py for windows by @sash-a in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F252\r\n* Refactor: Minor code style improvements by @helpingstar in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F246\r\n* Feat: Full Level-Based Foraging(LBF) environment by @WiemKhlifi in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F218\r\n* fix: rendering in cvrp and tsp by @sash-a in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F251\r\n* chore: upgrade python by @sash-a in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F254\r\n* feat: switch to ruff and upgrade pre-commit hooks by @sash-a in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F260\r\n* feat: return individual rewards in Connector env by @WiemKhlifi in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F263\r\n* Add dtype choice in step type\u002Ffunctions by @thomashirtz in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F262\r\n* feat: upgrade gym wrapper to gymnasium by @sash-a in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F264\r\n\r\n## New Contributors\r\n* @taodav made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F241\r\n* @helpingstar made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F246\r\n* @WiemKhlifi made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F218\r\n* @thomashirtz made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F262\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fcompare\u002Fv1.0.1...v1.1.0","2024-11-22T16:29:07",{"id":160,"version":161,"summary_zh":162,"released_at":163},105239,"v1.0.1","## Release Notes\r\nEnvironment specs are now attributes instead of methods with no arguments. Additionally, FlatPack training was broken due to a type-casting issue, which is now fixed.\r\n\r\n## What's Changed\r\n* fix: flatpack was training with ints by @sash-a in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F234\r\n* docs: update readme with paper and missing GIFs by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F235\r\n* feat(specs): make environment specs managed attributes by @aar65537 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F220\r\n* ci: increment version to 1.0.1 by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F236\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fcompare\u002Fv1.0.0...v1.0.1","2024-03-29T11:23:48",{"id":165,"version":166,"summary_zh":167,"released_at":168},105240,"v1.0.0","# Jumanji v1\r\nThis release accompanies the [ICLR 2024 paper](https:\u002F\u002Fopenreview.net\u002Fforum?id=C4CxQmp9wc). The library now includes 22 environments!\r\n\r\n## What's Changed\r\n* docs: fix typo in mmst.md by @eltociear in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F178\r\n* fix: latest chex and jax by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F181\r\n* fix bugs in specs test by @George-Ogden in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F184\r\n* docs: fix links at top of README by @George-Ogden in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F196\r\n* fix(maze): fixed row and col in _compute_action_mask function. by @danielpalen in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F194\r\n* fix(roboticWarehouse): goal conditional by @arnupretorius in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F191\r\n* Tetris docs by @George-Ogden in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F200\r\n* fix: mention to rware in rendering code by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F198\r\n* feat: Toggle logo in dark mode by @callumtilbury in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F205\r\n* fix: reward and discount spec not in wrapper by @sash-a in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F206\r\n* fix: timestep extras default value by @sash-a in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F207\r\n* chore: fix matplotlib and jax typing issues by @sash-a in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F216\r\n* Refactoring\u002Ftype hints by @dantp-ai in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F215\r\n* Revert \"feat: Toggle logo in dark mode\" by @callumtilbury in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F209\r\n* Fix: issue in dtype of grid in cleaner env by @raphaelavalos in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F217\r\n* feat: pacman environment by @siddarthsingh1 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F186\r\n* feat: adding sokoban environment by @mvmacfarlane in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F185\r\n* ci: fix tests in pipeline by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F225\r\n* fix: requirements by @sash-a in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F224\r\n* Fix\u002Fpackages data not included in sokoban by @coyettev in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F227\r\n* fix: autoreset wrappers by @sash-a in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F223\r\n* fix: default value for obs in extras by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F228\r\n* feat: FlatPack environment by @RuanJohn in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F188\r\n* feat: implement the sliding tile puzzle env by @ElshadaiK in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F189\r\n* ci: increment version by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F230\r\n* fix: pypi deployment by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F231\r\n* ci: pypi deployment replace setuptools with hatch by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F232\r\n\r\n## New Contributors\r\n* @eltociear made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F178\r\n* @George-Ogden made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F184\r\n* @danielpalen made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F194\r\n* @callumtilbury made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F205\r\n* @dantp-ai made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F215\r\n* @raphaelavalos made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F217\r\n* @siddarthsingh1 made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F186\r\n* @mvmacfarlane made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F185\r\n* @RuanJohn made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F188\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fcompare\u002Fv0.3.1...v1.0.0","2024-03-15T17:51:37",{"id":170,"version":171,"summary_zh":172,"released_at":173},105241,"v0.3.1","## Release Notes\r\nIn this release, we update the dependencies to support the latest `jax` (and `chex`) version and optimize the environment speed of `Game2048`.\r\n\r\n## What's Changed\r\n* fix(examples): port notebook to colab by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F169\r\n* ci: update to latest jax and chex by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F174\r\n* build: remove jaxlib by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F175\r\n* docs: update readme citation by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F176\r\n* feat(2048): environment performance improvements by @aar65537 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F172\r\n* build: bump version to 0.3.1 by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F177\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fcompare\u002Fv0.3.0...v0.3.1","2023-06-20T15:44:22",{"id":175,"version":176,"summary_zh":177,"released_at":178},105242,"v0.3.0","## Update\r\nWe release `v0.3.0` with a more standardized codebase that now includes 18 environments.\r\n\r\n## What's Changed\r\n* fix(generator_maze): correct definition of width and height by @coyettev in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F126\r\n* fix: make autoreset wrapper return 2 on reset by @rodSiry in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F123\r\n* feat(maze): update generator to return state by @PDuckworth in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F122\r\n* feat(connector): single agent by @sash-a in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F119\r\n* feat(tsp): generator by @surana01 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F137\r\n* refactor(snake): define viewer outside the env class by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F134\r\n* fix(training): connector num_agents in networks by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F145\r\n* feat(cvrp): generator by @surana01 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F138\r\n* feat(knapsack): generator by @surana01 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F139\r\n* fix(parametric_action_distribution): sum kl divergence over event_ndims in parametric action distribution by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F142\r\n* feat(connector): random walk board generator by @mwolinska in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F120\r\n* fix(2048): incorrect action mask by @aar65537 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F144\r\n* fix: put upper bounds on versions of jax and jaxlib by @dluo96 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F150\r\n* feat(robot_warehouse): full environment by @arnupretorius in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F140\r\n* feat(graph_coloring): implement graph_coloring environment by @ElshadaiK in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F130\r\n* feat(sudoku): implement environment by @Egiob in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F125\r\n* fix: data copy setup by @Egiob in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F151\r\n* docs: image fixes by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F154\r\n* feat(MMST): multi minimum spanning tree environment by @ulricharmel in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F135\r\n* feat(multi_cvrp): Implement MultiCVRP environment by @DriesSmit in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F133\r\n* feat(tetris): implement Tetris environment by @MedAliMimouni in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F132\r\n* fix: jax version constraint by @aar65537 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F160\r\n* feat: paper configs by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F156\r\n* docs: update online doc with new environments by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F158\r\n* feat(training): multi-worker training by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F164\r\n* feat(training): upload checkpoints by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F165\r\n* ci: increment version by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F167\r\n\r\n## New Contributors\r\n* @rodSiry made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F123\r\n* @PDuckworth made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F122\r\n* @sash-a made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F119\r\n* @mwolinska made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F120\r\n* @Uokoroafor made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F120\r\n* @RandyBrown1965 made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F120\r\n* @ojorgensen made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F120\r\n* @baubels made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F120\r\n* @aar65537 made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F144\r\n* @arnupretorius made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F140\r\n* @ElshadaiK made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F130\r\n* @Egiob made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F125\r\n* @ulricharmel made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F135\r\n* @DriesSmit made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F133\r\n* @MedAliMimouni made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F132\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fcompare\u002Fv0.2.2...v0.3.0","2023-06-09T16:53:41",{"id":180,"version":181,"summary_zh":182,"released_at":183},105243,"v0.2.2","## Release Notes\r\nThis release proposes a few fixes including a standardized viewer and generator interface for Minesweeper and RubiksCube.\r\n\r\n## What's Changed\r\n* docs(contributing): state to fork the repo to contribute by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F84\r\n* docs: typo in terminal logger by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F105\r\n* feat: create viewer interface for rendering of environments by @dluo96 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F99\r\n* docs: add pgx to readme by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F108\r\n* refactor(train): remove env factory by @dluo96 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F110\r\n* feat: separate rendering and instance generation for the RubiksCube environment by @TristanKalloniatis in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F82\r\n* test: consistent naming in Cleaner, Maze, Game2048, and Minesweeper by @dluo96 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F112\r\n* refactor(cleaner): return state from instance generator by @coyettev in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F87\r\n* feat: allow custom rendering and instance generation methods for Minesweeper by @TristanKalloniatis in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F85\r\n* ci: increment version by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F118\r\n\r\n## New Contributors\r\n* @TristanKalloniatis made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F82\r\n* @coyettev made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F87\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fcompare\u002Fv0.2.1...v0.2.2","2023-04-14T16:56:32",{"id":185,"version":186,"summary_zh":187,"released_at":188},105244,"v0.2.1","## Release Notes\r\nThis major release introduces a whole set of new environments, namely:\r\n- `Game2048`: the classic 4x4 2048 game\r\n- `Minesweeper`: the classic game from the video game\r\n- `RubiksCube`: the standard puzzle\r\n- `JobShop`: job-shop scheduling problem as another canonical CO problem\r\n- `Cleaner`: a maze with multiple _agents_ that spawn across the grid in order to clean all the tiles ASAP\r\n- `Connector`: replacement of the previous `Routing` environment reimplemented for better efficiency and modularity.\r\n- `Maze`: classic maze on a grid\r\n\r\nThis release also stabilizes the API with a cleaner definition of the environment specs, rendering implemented for all the environments (using matplotlib), and an actor-critic agent implement for training alongside environment-specific actor-critic networks for each environment.\r\n\r\n## What's Changed\r\n* fix: pygame requirements by @wang-r-j in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F72\r\n* feat: internal update for v0.2.0 by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F73\r\n* refactor: remove connect4 environment\r\n* feat: implement training and environment-specific networks in jumanji.training by @clement-bonnet \r\n* feat: replace routing environment with connector by @sash-a \r\n* feat: implement rubiks_cube environment by @TristanKalloniatis \r\n* feat: implement game_2048 environment by @OmaymaMahjoub \r\n* feat: implement minesweeper environment by @TristanKalloniatis \r\n* feat: implement job_shop environment by @dluo96 \r\n* refactor: knapsack, cvrp and tsp by @surana01 \r\n* feat: implement cleaner environment by @coyettev \r\n* feat: implement maze environment by @PDuckworth \r\n* feat: clean and augment snake with action mask by @clement-bonnet \r\n* refactor: implement arbitrarily nested specs by @dluo96 \r\n* ci: drop support for python 3.7\r\n\r\n## New Contributors\r\n* @wang-r-j made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F72\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fcompare\u002Fv0.1.6...v0.2.1","2023-03-21T17:11:26",{"id":190,"version":191,"summary_zh":192,"released_at":193},105245,"v0.1.6","## Release Notes\r\nThis release fixes the previous release (v0.1.5) that is broken on PyPi. Users may now run `pip install -U jumanji` or `pip install jumanji==0.1.6` in replacement of `pip install jumanji==0.1.5` (that is broken).\r\n\r\n## What's Changed\r\n* refactor: remove brax by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F69\r\n* build: increment semantic release by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F70\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fcompare\u002Fv0.1.5...v0.1.6","2023-02-22T16:01:16",{"id":195,"version":196,"summary_zh":197,"released_at":198},105246,"v0.1.5","## What's Changed\r\n* feat: deprecate Connect4 by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F67\r\n* feat: increment semantic release by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F68\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fcompare\u002Fv0.1.4...v0.1.5","2023-02-20T15:52:31",{"id":200,"version":201,"summary_zh":202,"released_at":203},105247,"v0.1.4","## What's Changed\r\n* fix: import protocol from typing extensions to ensure Python 3.7 compatibility by @dluo96 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F59\r\n* ci: added python 3.7 to github actions by @dluo96 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F58\r\n* fix: solve flake8 error causing ci to fail by @dluo96 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F64\r\n* refactor: binpack instantiation of instance generator by @cyprienc in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F61\r\n* build: increment semantic version to 0.1.4 by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F65\r\n\r\n## New Contributors\r\n* @cyprienc made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F61\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fcompare\u002Fv0.1.3...v0.1.4","2023-01-04T18:05:55",{"id":205,"version":206,"summary_zh":207,"released_at":208},105248,"v0.1.3","## Release Notes\r\nIn this release, we fixed the registration of the CVRP environment.\r\n\r\n## What's Changed\r\n* fix: import cvrp for environment registration by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F52\r\n* build: increment semantic version to 0.1.3 by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F53\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fcompare\u002Fv0.1.2...v0.1.3","2022-11-16T13:01:01",{"id":210,"version":211,"summary_zh":212,"released_at":213},105249,"v0.1.2","## Release Notes\r\nIn this release, we added the CVRP environment, fixed a few small bugs, and improved the documentation.\r\n\r\n## What's Changed\r\n* fix(docs): rendering of table in Examples section. by @BioGeek in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F16\r\n* bug: make pull requests trigger workflow by @cemlyn007 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F18\r\n* docs: hyperlink to registered environments in main README by @dluo96 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F20\r\n* fix: brax wrapper docstring by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F21\r\n* Fixed typo in env.py for snake game by @iamunr4v31 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F23\r\n* feat: badges by @alaterre in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F24\r\n* feat: implemented CVRP environment by @surana01 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F35\r\n* feat: explicit version number and testing by @alaterre in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F39\r\n* docs: state that we use the google style guide by @dluo96 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F42\r\n* refactor: improve consistency of extras by @dluo96 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F44\r\n* fix: change flake8 link from gitlab to github to resolve failing ci by @dluo96 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F48\r\n* fix: resolve jumanji import crash in non-terminal and non-notebook setting by @dluo96 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F46\r\n* feat: use protocol to force all environment states to have a key by @dluo96 in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F45\r\n* docs: update the contributing document with jumanji api 0.1.x by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F49\r\n* build: increment semantic version to 0.1.2 by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F51\r\n\r\n## New Contributors\r\n* @BioGeek made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F16\r\n* @cemlyn007 made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F18\r\n* @dluo96 made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F20\r\n* @iamunr4v31 made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F23\r\n* @surana01 made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F35\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fcompare\u002Fv0.1.1...v0.1.2","2022-11-16T11:35:05",{"id":215,"version":216,"summary_zh":217,"released_at":218},105250,"v0.1.1","## What's Changed\r\n* fix: colab link by @lollcat in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F8\r\n* fix: logo by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F9\r\n* fix: increase version of jax for control flow by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F10\r\n* feat: update version to 0.1.1 by @clement-bonnet in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F11\r\n\r\n## New Contributors\r\n* @lollcat made their first contribution in https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fpull\u002F8\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fcompare\u002Fv0.1.0...v0.1.1","2022-08-31T17:21:08",{"id":220,"version":221,"summary_zh":222,"released_at":223},105251,"v0.1.0","## Jumanji v0.1.0  💃 🚀 🎉  \r\n\r\nFirst release of Jumanji! 🌴 \r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Finstadeepai\u002Fjumanji\u002Fcommits\u002Fv0.1.0","2022-08-31T14:33:03"]