[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"tool-datamllab--awesome-game-ai":3,"similar-datamllab--awesome-game-ai":48},{"id":4,"github_repo":5,"name":6,"description_en":7,"description_zh":8,"ai_summary_zh":8,"readme_en":9,"readme_zh":10,"quickstart_zh":11,"use_case_zh":12,"hero_image_url":13,"owner_login":14,"owner_name":15,"owner_avatar_url":16,"owner_bio":17,"owner_company":18,"owner_location":18,"owner_email":18,"owner_twitter":18,"owner_website":19,"owner_url":20,"languages":18,"stars":21,"forks":22,"last_commit_at":23,"license":24,"difficulty_score":25,"env_os":26,"env_gpu":27,"env_ram":27,"env_deps":28,"category_tags":31,"github_topics":35,"view_count":42,"oss_zip_url":18,"oss_zip_packed_at":18,"status":43,"created_at":44,"updated_at":45,"faqs":46,"releases":47},6236,"datamllab\u002Fawesome-game-ai","awesome-game-ai","Awesome Game AI materials of Multi-Agent Reinforcement Learning","awesome-game-ai 是一个专注于多智能体强化学习（Multi-Agent Reinforcement Learning）的游戏人工智能资源合集。它旨在解决游戏 AI 开发中从单智能体向复杂多智能体环境过渡的难题，特别是在处理玩家间策略博弈、完全信息（如围棋、象棋）与不完全信息（如德州扑克、斗地主）等不同场景下的技术挑战。\n\n该资源库系统性地整理了开源项目工具包、综述论文、前沿研究文献以及相关行业竞赛资料，并按游戏类型和发表年份进行了细致分类。无论是希望复现 AlphaGo、AlphaZero 或 OpenAI Five 等经典案例的研究人员，还是致力于开发具备高阶决策能力游戏 NPC 的开发者，都能从中快速找到所需的代码实现与理论支撑。\n\n其独特亮点在于不仅涵盖了传统的棋类游戏，还深入收录了针对即时战略游戏（如《星际争霸》、《Dota 2》）及各类扑克牌游戏的最新突破成果，为探索复杂动态环境下的协同与对抗机制提供了宝贵的一站式入口。如果你正投身于游戏 AI 的前沿探索，awesome-game-ai 将是不可或缺的参考指南。","# Awesome-Game-AI\n[![Awesome](https:\u002F\u002Fcdn.rawgit.com\u002Fsindresorhus\u002Fawesome\u002Fd7305f38d29fed78fa85652e3a63e154dd8e8829\u002Fmedia\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fsindresorhus\u002Fawesome)\n\nA curated, but incomplete, list of game AI resources on **multi-agent** learning.\n\nIf you want to contribute to this list, please feel free to send a pull request. Also you can contact [daochen.zha@rice.edu](mailto:daochen.zha@rice.edu), or [khlai@rice.edu](mailto:khlai@rice.edu).\n\n:loudspeaker: News: Please check out our open-sourced [Large Time Series Model (LTSM)](https:\u002F\u002Fgithub.com\u002Fdaochenzha\u002Fltsm)!\n\n:loudspeaker: Have you heard of data-centric AI? Please check out our [data-centric AI survey](https:\u002F\u002Farxiv.org\u002Fabs\u002F2303.10158) and [awesome data-centric AI resources](https:\u002F\u002Fgithub.com\u002Fdaochenzha\u002Fdata-centric-AI)!\n\n## What is Game AI?\n\nGame AI is focusing on predicting which actions should be taken, based on the current conditions. Generally, most games incorporate some sort of AI, which are usually characters or players in the game. For some popular games such as [Starcraft](https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FStarCraft) and [Dota 2](https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FDota_2), developers have spent years to design and refine the AI to enhance the experience.\n\n## Single-Agent vs. Multi-Agent\nNumerous studies and achievements have been made to game AI in single-agent environments, where there is a single player in the games. For instance, [Deep Q-learning](https:\u002F\u002Fwww.nature.com\u002Farticles\u002Fnature14236) is successfully applied to Atari Games. Other examples include [Super Mario](https:\u002F\u002Fgithub.com\u002Faleju\u002Fmario-ai), [Minecraft](https:\u002F\u002Fwww.aaai.org\u002Focs\u002Findex.php\u002FAAAI\u002FAAAI17\u002Fpaper\u002FviewPaper\u002F14630), and [Flappy Bird](https:\u002F\u002Fgithub.com\u002Fyenchenlin\u002FDeepLearningFlappyBird).\n\nMulti-agent environments are more challenging since each player has to reason about the other players' moves. Modern reinforcement learning techniques have boosted multi-agent game AI. In 2015, [AlphaGo](https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FAlphaGo), for the first time beat a human professional Go player on a full-sized 19×19 board. In 2017, [AlphaZero](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Farticle\u002Falphazero-shedding-new-light-grand-games-chess-shogi-and-go) taught itself from scratch and learned to master the games of chess, shogi, and Go. In more recent years, researchers have made efforts to poker games, such as [Libratus](https:\u002F\u002Fscience.sciencemag.org\u002Fcontent\u002F359\u002F6374\u002F418), [DeepStack](https:\u002F\u002Fscience.sciencemag.org\u002Fcontent\u002F356\u002F6337\u002F508) and [DouZero](https:\u002F\u002Fgithub.com\u002Fkwai\u002FDouZero), achieving expert-level performance in Texas Hold'em and Chinese Poker game Dou Dizhu. Now researchers keep progressing and achieve human-level AI on [Dota 2](https:\u002F\u002Fopenai.com\u002Ffive\u002F) and [Starcraft 2](https:\u002F\u002Fdeepmind.com\u002Fblog\u002Farticle\u002Falphastar-mastering-real-time-strategy-game-starcraft-ii) with deep reinforcement learning.\n\n## Perfect Information vs. Imperfect Information\nPerfect information means that each player has access to the same information of the game, e.g., Go, Chess, and Gomoku. Imperfect information refers to the situation where players can not observe the full state of the game. For example, in card games, a player can not observe the hands of the other players. Imperfect information games are usually considered more challenging with more possibilities.\n\n## What is included?\nThis repository gathers some awesome resources for Game AI on multi-agent learning for both perfect and imperfect information games, including but not limited to, open-source projects, review papers, research papers, conferences, and competitions. The resources are categorized by games, and the papers are sorted by years.\n\n\n## Table of Contents\n\n* [Open-Source Projects](#open-source-projects)\n  * [Unified Toolkits](#unified-toolkits)\n  * [Texas Hold'em](#texas-holdem-projects)\n  * [Dou Dizhu](#dou-dizhu-projects)\n  * [Starcraft](#starcraft-projects)\n  * [Go](#go-projects)\n  * [Gomoku](#gomoku-projects)\n  * [Chess](#chess-projects)\n  * [Chinese Chess](#chinese-chess-projects)\n* [Review and General Papers](#review-and-general-papers)\n* [Research Papers](#research-papers)\n  * [Betting Games](#betting-games)\n  * [Dou Dizhu](#dou-dizhu)\n  * [Mahjong](#mahjong)\n  * [Bridge](#bridge)\n  * [Go](#go)\n  * [Starcraft](#starcraft)\n* [Conferences and Workshops](#conferences-and-workshops)\n* [Competitions](#competitions)\n* [Related Lists](#related-lists)\n\n## Open-Source Projects\n\n### Unified Toolkits\n  * RLCard: A Toolkit for Reinforcement Learning in Card Games [[paper](https:\u002F\u002Farxiv.org\u002Fabs\u002F1910.04376)] [[code](https:\u002F\u002Fgithub.com\u002Fdatamllab\u002Frlcard)].\n  * OpenSpiel: A Framework for Reinforcement Learning in Games [[paper](https:\u002F\u002Farxiv.org\u002Fabs\u002F1908.09453)] [[code](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fopen_spiel)].\n  * Unity ML-Agents Toolkit [[paper](https:\u002F\u002Farxiv.org\u002Fabs\u002F1809.02627)] [[code](https:\u002F\u002Fgithub.com\u002FUnity-Technologies\u002Fml-agents)].\n  * Alpha Zero General [[code](https:\u002F\u002Fgithub.com\u002Fsuragnair\u002Falpha-zero-general)].\n\n### Texas Hold'em Projects\n  * DeepStack-Leduc [[paper](https:\u002F\u002Farxiv.org\u002Fabs\u002F1701.01724)] [[code](https:\u002F\u002Fgithub.com\u002Flifrordi\u002FDeepStack-Leduc)].\n  * DeepHoldem [[code](https:\u002F\u002Fgithub.com\u002Fhappypepper\u002FDeepHoldem)].\n  * OpenAI Gym No Limit Texas Hold 'em Environment for Reinforcement Learning [[code](https:\u002F\u002Fgithub.com\u002Fwenkesj\u002Fholdem)].\n  * PyPokerEngine [[code](https:\u002F\u002Fgithub.com\u002Fishikota\u002FPyPokerEngine)].\n  * Deep mind pokerbot for pokerstars and partypoker [[code](https:\u002F\u002Fgithub.com\u002Fdickreuter\u002FPoker)].\n\n  \n### Dou Dizhu Projects\n  * PerfectDou: Dominating DouDizhu with Perfect Information Distillation [[code](https:\u002F\u002Fgithub.com\u002FNetease-Games-AI-Lab-Guangzhou\u002FPerfectDou)].\n  * DouZero: Mastering DouDizhu with Self-Play Deep Reinforcement Learning [[code](https:\u002F\u002Fgithub.com\u002Fkwai\u002FDouZero)].\n  * Doudizhu AI using reinforcement learning [[code](https:\u002F\u002Fgithub.com\u002Fskyduy\u002Fdoudizhu-rl)].\n  * Dou Di Zhu with Combinational Q-Learning [[paper](https:\u002F\u002Fgithub.com\u002Fqq456cvb\u002Fdoudizhu-C)] [[code](https:\u002F\u002Fgithub.com\u002Fqq456cvb\u002Fdoudizhu-C)].\n  * DouDiZhu [[code](https:\u002F\u002Fgithub.com\u002Fsongbaoming\u002FDouDiZhu)].\n  * 斗地主AI设计与实现 [[code](https:\u002F\u002Fgithub.com\u002FZhouWeikuan\u002FDouDiZhu)].\n\n### Starcraft Projects\n* StarCraft II Learning Environment [[paper](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.04782)] [[code](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fpysc2)].\n* Gym StarCraft [[code](https:\u002F\u002Fgithub.com\u002Falibaba\u002Fgym-starcraft)].\n* StartCraft II Reinforcement Learning Examples [[code](https:\u002F\u002Fgithub.com\u002Fchris-chris\u002Fpysc2-examples)].\n* A Guide to DeepMind's StarCraft AI Environment [[code](https:\u002F\u002Fgithub.com\u002FllSourcell\u002FA-Guide-to-DeepMinds-StarCraft-AI-Environment)].\n* A reimplementation of Alphastar based on DI-engine with trained models [[code](https:\u002F\u002Fgithub.com\u002Fopendilab\u002FDI-star)].\n\n### Go Projects\n* ELF: a platform for game research with AlphaGoZero\u002FAlphaZero reimplementation [[code](https:\u002F\u002Fgithub.com\u002Fpytorch\u002FELF)] [[paper](https:\u002F\u002Farxiv.org\u002Fpdf\u002F1902.04522.pdf)].\n\n### Gomoku Projects\n* AlphaZero-Gomoku [[code](https:\u002F\u002Fgithub.com\u002Fjunxiaosong\u002FAlphaZero_Gomoku)].\n* gobang [[code](https:\u002F\u002Fgithub.com\u002Flihongxun945\u002Fgobang)].\n\n### Chess Projects\n* Chess-Alpha-Zero [[code](https:\u002F\u002Fgithub.com\u002FZeta36\u002Fchess-alpha-zero)].\n* Deep Pink [[code](https:\u002F\u002Fgithub.com\u002Ferikbern\u002Fdeep-pink)].\n* Simple chess AI [[code](https:\u002F\u002Fgithub.com\u002Flhartikk\u002Fsimple-chess-ai)].\n\n### Chinese Chess Projects\n* CCZero (中国象棋Zero) [[code](https:\u002F\u002Fgithub.com\u002FNeymarL\u002FChineseChess-AlphaZero)].\n\n### Mahjong Projects\n* pymahjong (Japanese Riichi Mahjong) [[code](https:\u002F\u002Fgithub.com\u002FAgony5757\u002Fmahjong\u002Ftree\u002Fmaster\u002Fpymahjong)].\n* Mortal [[code](https:\u002F\u002Fgithub.com\u002FEquim-chan\u002FMortal)].\n\n## Review and General Papers\n* Deep reinforcement learning from self-play in imperfect-information games, arXiv 2016 [[paper](https:\u002F\u002Farxiv.org\u002Fpdf\u002F1603.01121.pdf)].\n* Multi-agent Reinforcement Learning: An Overview, 2010 [[paper](https:\u002F\u002Flink.springer.com\u002Fchapter\u002F10.1007\u002F978-3-642-14435-6_7)].\n* An overview of cooperative and competitive multiagent learning, LAMAS 2005 [[paper](https:\u002F\u002Fdl.acm.org\u002Fcitation.cfm?id=2180576)].\n* Multi-agent reinforcement learning: a critical survey, 2003 [[paper](http:\u002F\u002Fjmvidal.cse.sc.edu\u002Flibrary\u002Fshoham03a.pdf)].\n\n## Research Papers\n### Betting Games\nBetting games are one of the most popular form of Poker games. The list includes [Goofspiel](https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FGoofspiel), [Kuhn Poker](https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FKuhn_poker), [Leduc Poker](http:\u002F\u002Fpoker.cs.ualberta.ca\u002Fpublications\u002FUAI05.pdf), and [Texas Hold'em](https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FTexas_hold_%27em).\n\n* Neural Replicator Dynamics, arXiv 2019 [[paper](https:\u002F\u002Farxiv.org\u002Fabs\u002F1906.00190)].\n* Computing Approximate Equilibria in Sequential Adversarial Games by Exploitability Descent, IJCAI 2019 [[paper](https:\u002F\u002Fwww.ijcai.org\u002Fproceedings\u002F2019\u002F0066.pdf)].\n* Solving Imperfect-Information Games via Discounted Regret Minimization, AAAI 2019 [[paper](https:\u002F\u002Faaai.org\u002Fojs\u002Findex.php\u002FAAAI\u002Farticle\u002Fview\u002F4007)].\n* Deep Counterfactual Regret Minimization, ICML, 2019 [[paper](http:\u002F\u002Fproceedings.mlr.press\u002Fv97\u002Fbrown19b\u002Fbrown19b.pdf)].\n* Actor-Critic Policy Optimization in Partially Observable Multiagent Environments, NeurIPS 2018 [[paper](https:\u002F\u002Fpapers.nips.cc\u002Fpaper\u002F7602-actor-critic-policy-optimization-in-partially-observable-multiagent-environments.pdf)].\n* Safe and Nested Subgame Solving for Imperfect-Information Games, NeurIPS, 2018 [[paper](http:\u002F\u002Fpapers.nips.cc\u002Fpaper\u002F6671-safe-and-nested-subgame-solving-for-imperfect-information-games.pdf)].\n* DeepStack: Expert-Level Artificial Intelligence in Heads-Up No-Limit Poker, Science 2017 [[paper](https:\u002F\u002Farxiv.org\u002Fpdf\u002F1701.01724.pdf)].\n* A Unified Game-Theoretic Approach to Multiagent Reinforcement Learning, NeurIPS 2017 [[paper](https:\u002F\u002Fpdfs.semanticscholar.org\u002Ffbe9\u002F950202a7fcc756369a38cb1ef4b9b994ae88.pdf)].\n* Poker-CNN: A pattern learning strategy for making draws and bets in poker games using convolutional networks [[paper](https:\u002F\u002Fwww.aaai.org\u002Focs\u002Findex.php\u002FAAAI\u002FAAAI16\u002Fpaper\u002Fview\u002F12172\u002F11606)].\n* Deep Reinforcement Learning from Self-Play in Imperfect-Information Games, arXiv 2016 [[paper](https:\u002F\u002Farxiv.org\u002Fabs\u002F1603.01121)].\n* Fictitious Self-Play in Extensive-Form Games, ICML 2015 [[paper](http:\u002F\u002Fproceedings.mlr.press\u002Fv37\u002Fheinrich15.pdf)].\n* Solving Heads-up Limit Texas Hold’em, IJCAI 2015 [[paper](https:\u002F\u002Fpoker.cs.ualberta.ca\u002Fpublications\u002F2015-ijcai-cfrplus.pdf)].\n* Regret Minimization in Games with Incomplete Information, NeurIPS 2007 [[paper](https:\u002F\u002Fpoker.cs.ualberta.ca\u002Fpublications\u002FNIPS07-cfr.pdf)].\n\n### Dou Dizhu\n\n* PerfectDou: Dominating DouDizhu with Perfect Information Distillation, NeurIPS 2022 [[paper](https:\u002F\u002Farxiv.org\u002Fabs\u002F2203.16406)] [[code](https:\u002F\u002Fgithub.com\u002FNetease-Games-AI-Lab-Guangzhou\u002FPerfectDou)].\n* DouZero: Mastering DouDizhu with Self-Play Deep Reinforcement Learning, ICML 2021 [[paper](https:\u002F\u002Farxiv.org\u002Fabs\u002F2106.06135)] [[code](https:\u002F\u002Fgithub.com\u002Fkwai\u002FDouZero)].\n* DeltaDou: Expert-level Doudizhu AI through Self-play, IJCAI 2019 [[paper](https:\u002F\u002Fwww.ijcai.org\u002Fproceedings\u002F2019\u002F0176.pdf)].\n* Combinational Q-Learning for Dou Di Zhu, arXiv 2019 [[paper](https:\u002F\u002Farxiv.org\u002Fabs\u002F1901.08925)] [[code](https:\u002F\u002Fgithub.com\u002Fqq456cvb\u002Fdoudizhu-C)].\n* Determinization and information set Monte Carlo Tree Search for the card game Dou Di Zhu, CIG 2011 [[paper](https:\u002F\u002Fieeexplore.ieee.org\u002Fdocument\u002F6031993)].\n\n### Mahjong\n\n* Variational oracle guiding for reinforcement learning, ICLR 2022 [[paper](https:\u002F\u002Fopenreview.net\u002Fforum?id=pjqqxepwoMy)]\n* Suphx: Mastering Mahjong with Deep Reinforcement Learning, arXiv 2020 [[paper](https:\u002F\u002Farxiv.org\u002Fabs\u002F2003.13590)].\n* Method for Constructing Artificial Intelligence Player with Abstraction to Markov Decision Processes in Multiplayer Game of Mahjong, arXiv 2019 [[paper](https:\u002F\u002Farxiv.org\u002Fabs\u002F1904.07491)].\n* Building a Computer Mahjong Player Based on Monte Carlo Simulation and Opponent Models, IEEE CIG 2017 [[paper](https:\u002F\u002Fieeexplore.ieee.org\u002Fdocument\u002F7317929)].\n\n### Bridge\n\n* Boosting a Bridge Artificial Intelligence, ICTAI 2017 [[paper](https:\u002F\u002Fieeexplore.ieee.org\u002Fdocument\u002F8372096)].\n\n### Go\n* Mastering the game of Go without human knowledge, Nature 2017 [[paper](https:\u002F\u002Fwww.nature.com\u002Farticles\u002Fnature24270)].\n* Mastering the game of Go with deep neural networks and tree search, Nature 2016 [[paper](https:\u002F\u002Fwww.nature.com\u002Farticles\u002Fnature16961)].\n* Temporal-difference search in computer Go, Machine Learning, 2012 [[paper](https:\u002F\u002Flink.springer.com\u002Farticle\u002F10.1007\u002Fs10994-012-5280-0)].\n* Monte-Carlo tree search and rapid action value estimation in computer Go, Artificial Intelligence, 2011 [[paper](https:\u002F\u002Fwww.ics.uci.edu\u002F~dechter\u002Fcourses\u002Fics-295\u002Fwinter-2018\u002Fpapers\u002Fmcts-gelly-silver.pdf)].\n* Computing “elo ratings” of move patterns in the game of go, ICGA Journal, 2007 [[paper](https:\u002F\u002Fhal.inria.fr\u002Ffile\u002Findex\u002Fdocid\u002F149859\u002Ffilename\u002FMMGoPatterns.pdf)].\n\n### Starcraft\n* Grandmaster level in StarCraft II using multi-agent reinforcement learning, Nature 2019 [[paper](https:\u002F\u002Fwww.nature.com\u002Farticles\u002Fs41586-019-1724-z)].\n* On Reinforcement Learning for Full-length Game of StarCraft, AAAI 2019 [[paper](https:\u002F\u002Faaai.org\u002Fojs\u002Findex.php\u002FAAAI\u002Farticle\u002Fview\u002F4394)].\n* Stabilising experience replay for deep multi-agent reinforcement learning, ICML 2017 [[paper](https:\u002F\u002Fdl.acm.org\u002Fcitation.cfm?id=3305500)].\n* Cooperative reinforcement learning for multiple units combat in starCraft, SSCI 2017 [[paper](https:\u002F\u002Fieeexplore.ieee.org\u002Fabstract\u002Fdocument\u002F8280949)].\n* Learning macromanagement in starcraft from replays using deep learning, CIG 2017 [[paper](https:\u002F\u002Fieeexplore.ieee.org\u002Fabstract\u002Fdocument\u002F8080430)].\n* Applying reinforcement learning to small scale combat in the real-time strategy game StarCraft: Broodwar, CIG 2012 [[paper](https:\u002F\u002Fieeexplore.ieee.org\u002Fabstract\u002Fdocument\u002F6374183)].\n\n## Conferences and Workshops\n* [IEEE Conference on Computational Intelligence and Games (CIG)](http:\u002F\u002Fwww.ieee-cig.org\u002F)\n* [AAAI Workshop on Reinforcement Learning in Games](http:\u002F\u002Faaai-rlg.mlanctot.info\u002F)\n* [Bridging Game Theory and Deep Learning](https:\u002F\u002Fnips.cc\u002FConferences\u002F2019\u002FSchedule?showEvent=13158)\n* [IJCAI 2018 Computer Games Workshop](https:\u002F\u002Fwww.lamsade.dauphine.fr\u002F~cazenave\u002Fcgw2018\u002Fcgw2018.html)\n* [IEEE Conference on Games (CoG)](http:\u002F\u002Fieee-cog.org\u002F2020\u002F)\n\n## Competitions\n* [International Computer Games Association (ICGA)](http:\u002F\u002Ficga.org\u002F)\n* [Annual Computer Poker Competition](http:\u002F\u002Fwww.computerpokercompetition.org\u002F)\n\n## Related Lists\n* [Awesome StarCraft AI](https:\u002F\u002Fgithub.com\u002FSKTBrain\u002Fawesome-starcraftAI)\n* [Awesome Deep Reinforcement Learning](https:\u002F\u002Fgithub.com\u002Ftigerneil\u002Fawesome-deep-rl)\n","# 令人惊叹的游戏AI\n[![Awesome](https:\u002F\u002Fcdn.rawgit.com\u002Fsindresorhus\u002Fawesome\u002Fd7305f38d29fed78fa85652e3a63e154dd8e8829\u002Fmedia\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fsindresorhus\u002Fawesome)\n\n这是一份精心整理但并不完整的、关于**多智能体**学习的游戏AI资源列表。\n\n如果您希望为本列表贡献力量，欢迎随时提交拉取请求。您也可以联系 [daochen.zha@rice.edu](mailto:daochen.zha@rice.edu) 或 [khlai@rice.edu](mailto:khlai@rice.edu)。\n\n:loudspeaker: 最新消息：请查看我们开源的【大型时间序列模型（LTSM）】！[GitHub链接](https:\u002F\u002Fgithub.com\u002Fdaochenzha\u002Fltsm)\n\n:loudspeaker: 您听说过以数据为中心的人工智能吗？请参阅我们的【以数据为中心的人工智能综述】和【以数据为中心的人工智能优秀资源】！\n\n## 什么是游戏AI？\n\n游戏AI专注于根据当前的游戏状态预测应采取的动作。通常，大多数游戏都会内置某种形式的AI，这些AI表现为游戏中的角色或玩家。对于一些热门游戏，如《星际争霸》和《Dota 2》，开发者花费了多年时间设计和优化AI，以提升玩家的游戏体验。\n\n## 单智能体与多智能体\n\n在单智能体环境中，针对游戏AI的研究和成果已经非常丰富，例如，在Atari游戏中成功应用了深度Q学习。其他例子还包括超级马里奥、我的世界以及Flappy Bird等。\n\n相比之下，多智能体环境更具挑战性，因为每个玩家都需要考虑其他玩家的行动。近年来，随着强化学习技术的发展，多智能体游戏AI取得了显著进展。2015年，AlphaGo首次在标准19×19围棋盘上战胜了职业棋手；2017年，AlphaZero从零开始自学，掌握了国际象棋、将棋和围棋。近年来，研究人员又将目光投向扑克类游戏，如Libratus、DeepStack和DouZero等，这些项目在德州扑克和中国斗地主中都达到了专家级水平。如今，借助深度强化学习，研究人员还在《Dota 2》和《星际争霸2》中实现了接近人类水平的AI表现。\n\n## 完全信息与不完全信息\n\n完全信息博弈是指所有玩家都能获取相同的信息，例如围棋、国际象棋和五子棋。而不完全信息博弈则意味着玩家无法观察到游戏的全部状态。比如在纸牌游戏中，玩家无法看到其他玩家的手牌。因此，不完全信息博弈往往具有更高的复杂性和更多的可能性。\n\n## 本资源包含哪些内容？\n\n本仓库汇集了大量关于多智能体学习的游戏AI资源，涵盖完全信息和不完全信息两类游戏，包括但不限于开源项目、综述论文、研究论文、会议及竞赛等。资源按游戏分类，并按年份排序。\n\n## 目录\n\n* [开源项目](#开源项目)\n  * [通用工具包](#通用工具包)\n  * [德州扑克](#德州扑克项目)\n  * [斗地主](#斗地主项目)\n  * [星际争霸](#星际争霸项目)\n  * [围棋](#围棋项目)\n  * [五子棋](#五子棋项目)\n  * [国际象棋](#国际象棋项目)\n  * [中国象棋](#中国象棋项目)\n* [综述与通用论文](#综述与通用论文)\n* [研究论文](#研究论文)\n  * [博彩类游戏](#博彩类游戏)\n  * [斗地主](#斗地主)\n  * [麻将](#麻将)\n  * [桥牌](#桥牌)\n  * [围棋](#围棋)\n  * [星际争霸](#星际争霸)\n* [会议与研讨会](#会议与研讨会)\n* [比赛](#比赛)\n* [相关列表](#相关列表)\n\n## 开源项目\n\n### 通用工具包\n  * RLCard：用于卡牌游戏强化学习的工具包 [[论文](https:\u002F\u002Farxiv.org\u002Fabs\u002F1910.04376)] [[代码](https:\u002F\u002Fgithub.com\u002Fdatamllab\u002Frlcard)]。\n  * OpenSpiel：适用于各类游戏的强化学习框架 [[论文](https:\u002F\u002Farxiv.org\u002Fabs\u002F1908.09453)] [[代码](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fopen_spiel)]。\n  * Unity ML-Agents 工具包 [[论文](https:\u002F\u002Farxiv.org\u002Fabs\u002F1809.02627)] [[代码](https:\u002F\u002Fgithub.com\u002FUnity-Technologies\u002Fml-agents)]。\n  * Alpha Zero General [[代码](https:\u002F\u002Fgithub.com\u002Fsuragnair\u002Falpha-zero-general)]。\n\n### 德州扑克项目\n  * DeepStack-Leduc [[论文](https:\u002F\u002Farxiv.org\u002Fabs\u002F1701.01724)] [[代码](https:\u002F\u002Fgithub.com\u002Flifrordi\u002FDeepStack-Leduc)]。\n  * DeepHoldem [[代码](https:\u002F\u002Fgithub.com\u002Fhappypepper\u002FDeepHoldem)]。\n  * OpenAI Gym 无限制德州扑克环境，专为强化学习设计 [[代码](https:\u002F\u002Fgithub.com\u002Fwenkesj\u002Fholdem)]。\n  * PyPokerEngine [[代码](https:\u002F\u002Fgithub.com\u002Fishikota\u002FPyPokerEngine)]。\n  * DeepMind开发的扑克机器人，可用于PokerStars和Partypoker [[代码](https:\u002F\u002Fgithub.com\u002Fdickreuter\u002FPoker)]。\n\n### 斗地主项目\n  * PerfectDou：通过完美信息蒸馏技术称霸斗地主 [[代码](https:\u002F\u002Fgithub.com\u002FNetease-Games-AI-Lab-Guangzhou\u002FPerfectDou)]。\n  * DouZero：利用自我对弈的深度强化学习掌握斗地主 [[代码](https:\u002F\u002Fgithub.com\u002Fkwai\u002FDouZero)]。\n  * 基于强化学习的斗地主AI [[代码](https:\u002F\u002Fgithub.com\u002Fskyduy\u002Fdoudizhu-rl)]。\n  * 结合组合Q学习的斗地主AI [[论文](https:\u002F\u002Fgithub.com\u002Fqq456cvb\u002Fdoudizhu-C)] [[代码](https:\u002F\u002Fgithub.com\u002Fqq456cvb\u002Fdoudizhu-C)]。\n  * 斗地主AI [[代码](https:\u002F\u002Fgithub.com\u002Fsongbaoming\u002FDouDiZhu)]。\n  * 斗地主AI的设计与实现 [[代码](https:\u002F\u002Fgithub.com\u002FZhouWeikuan\u002FDouDiZhu)]。\n\n### 星际争霸项目\n  * StarCraft II 学习环境 [[论文](https:\u002F\u002Farxiv.org\u002Fabs\u002F1708.04782)] [[代码](https:\u002F\u002Fgithub.com\u002Fdeepmind\u002Fpysc2)]。\n  * Gym StarCraft [[代码](https:\u002F\u002Fgithub.com\u002Falibaba\u002Fgym-starcraft)]。\n  * StarCraft II 强化学习示例 [[代码](https:\u002F\u002Fgithub.com\u002Fchris-chris\u002Fpysc2-examples)]。\n  * 关于DeepMind星际争霸AI环境的指南 [[代码](https:\u002F\u002Fgithub.com\u002FllSourcell\u002FA-Guide-to-DeepMinds-StarCraft-AI-Environment)]。\n  * 基于DI-engine重新实现的AlphaStar，并附带训练好的模型 [[代码](https:\u002F\u002Fgithub.com\u002Fopendilab\u002FDI-star)]。\n\n### 围棋项目\n  * ELF：一个基于AlphaGoZero\u002FAlphaZero重写的平台，用于游戏研究 [[代码](https:\u002F\u002Fgithub.com\u002Fpytorch\u002FELF)] [[论文](https:\u002F\u002Farxiv.org\u002Fpdf\u002F1902.04522.pdf)]。\n\n### 五子棋项目\n* AlphaZero-Gomoku [[代码](https:\u002F\u002Fgithub.com\u002Fjunxiaosong\u002FAlphaZero_Gomoku)]。\n* gobang [[代码](https:\u002F\u002Fgithub.com\u002Flihongxun945\u002Fgobang)]。\n\n### 国际象棋项目\n* Chess-Alpha-Zero [[代码](https:\u002F\u002Fgithub.com\u002FZeta36\u002Fchess-alpha-zero)]。\n* Deep Pink [[代码](https:\u002F\u002Fgithub.com\u002Ferikbern\u002Fdeep-pink)]。\n* Simple chess AI [[代码](https:\u002F\u002Fgithub.com\u002Flhartikk\u002Fsimple-chess-ai)]。\n\n### 中国象棋项目\n* CCZero (中国象棋Zero) [[代码](https:\u002F\u002Fgithub.com\u002FNeymarL\u002FChineseChess-AlphaZero)]。\n\n### 麻将项目\n* pymahjong（日本立直麻将）[[代码](https:\u002F\u002Fgithub.com\u002FAgony5757\u002Fmahjong\u002Ftree\u002Fmaster\u002Fpymahjong)]。\n* Mortal [[代码](https:\u002F\u002Fgithub.com\u002FEquim-chan\u002FMortal)]。\n\n## 综述与通用论文\n* 不完全信息博弈中的自我对弈深度强化学习，arXiv 2016 [[论文](https:\u002F\u002Farxiv.org\u002Fpdf\u002F1603.01121.pdf)]。\n* 多智能体强化学习：综述，2010年 [[论文](https:\u002F\u002Flink.springer.com\u002Fchapter\u002F10.1007\u002F978-3-642-14435-6_7)]。\n* 合作与竞争性多智能体学习概述，LAMAS 2005 [[论文](https:\u002F\u002Fdl.acm.org\u002Fcitation.cfm?id=2180576)]。\n* 多智能体强化学习：批判性综述，2003年 [[论文](http:\u002F\u002Fjmvidal.cse.sc.edu\u002Flibrary\u002Fshoham03a.pdf)]。\n\n## 研究论文\n### 投注类游戏\n投注类游戏是扑克游戏中最受欢迎的形式之一。列表包括[Goofspiel](https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FGoofspiel)、[库恩扑克](https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FKuhn_poker)、[勒杜克扑克](http:\u002F\u002Fpoker.cs.ualberta.ca\u002Fpublications\u002FUAI05.pdf)和[德州扑克](https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FTexas_hold_%27em)。\n\n* 神经元复制动力学，arXiv 2019 [[论文](https:\u002F\u002Farxiv.org\u002Fabs\u002F1906.00190)]。\n* 基于可利用性下降的顺序对抗性博弈近似均衡计算，IJCAI 2019 [[论文](https:\u002F\u002Fwww.ijcai.org\u002Fproceedings\u002F2019\u002F0066.pdf)]。\n* 通过折扣后悔最小化求解不完全信息博弈，AAAI 2019 [[论文](https:\u002F\u002Faaai.org\u002Fojs\u002Findex.php\u002FAAAI\u002Farticle\u002Fview\u002F4007)]。\n* 深度反事实后悔最小化，ICML 2019 [[论文](http:\u002F\u002Fproceedings.mlr.press\u002Fv97\u002Fbrown19b\u002Fbrown19b.pdf)]。\n* 部分可观测多智能体环境中的演员—评论家策略优化，NeurIPS 2018 [[论文](https:\u002F\u002Fpapers.nips.cc\u002Fpaper\u002F7602-actor-critic-policy-optimization-in-partially-observable-multiagent-environments.pdf)]。\n* 不完全信息博弈的安全嵌套子博弈求解，NeurIPS 2018 [[论文](http:\u002F\u002Fpapers.nips.cc\u002Fpaper\u002F6671-safe-and-nested-subgame-solving-for-imperfect-information-games.pdf)]。\n* DeepStack：无限制头寸单挑扑克中的专家级人工智能，Science 2017 [[论文](https:\u002F\u002Farxiv.org\u002Fpdf\u002F1701.01724.pdf)]。\n* 多智能体强化学习的统一博弈论方法，NeurIPS 2017 [[论文](https:\u002F\u002Fpdfs.semanticscholar.org\u002Ffbe9\u002F950202a7fcc756369a38cb1ef4b9b994ae88.pdf)]。\n* 扑克-CNN：使用卷积网络在扑克游戏中进行跟注和下注的模式学习策略 [[论文](https:\u002F\u002Fwww.aaai.org\u002Focs\u002Findex.php\u002FAAAI\u002FAAAI16\u002Fpaper\u002Fview\u002F12172\u002F11606)]。\n* 不完全信息博弈中的自我对弈深度强化学习，arXiv 2016 [[论文](https:\u002F\u002Farxiv.org\u002Fabs\u002F1603.01121)]。\n* 广义形式博弈中的虚构自我对弈，ICML 2015 [[论文](http:\u002F\u002Fproceedings.mlr.press\u002Fv37\u002Fheinrich15.pdf)]。\n* 解决头寸限注德州扑克，IJCAI 2015 [[论文](https:\u002F\u002Fpoker.cs.ualberta.ca\u002Fpublications\u002F2015-ijcai-cfrplus.pdf)]。\n* 不完全信息博弈中的后悔最小化，NeurIPS 2007 [[论文](https:\u002F\u002Fpoker.cs.ualberta.ca\u002Fpublications\u002FNIPS07-cfr.pdf)]。\n\n### 斗地主\n\n* PerfectDou：通过完美信息蒸馏统治斗地主，NeurIPS 2022 [[论文](https:\u002F\u002Farxiv.org\u002Fabs\u002F2203.16406)] [[代码](https:\u002F\u002Fgithub.com\u002FNetease-Games-AI-Lab-Guangzhou\u002FPerfectDou)]。\n* DouZero：通过自我对弈深度强化学习掌握斗地主，ICML 2021 [[论文](https:\u002F\u002Farxiv.org\u002Fabs\u002F2106.06135)] [[代码](https:\u002F\u002Fgithub.com\u002Fkwai\u002FDouZero)]。\n* DeltaDou：通过自我对弈达到专家级斗地主AI，IJCAI 2019 [[论文](https:\u002F\u002Fwww.ijcai.org\u002Fproceedings\u002F2019\u002F0176.pdf)]。\n* 斗地主的组合Q学习，arXiv 2019 [[论文](https:\u002F\u002Farxiv.org\u002Fabs\u002F1901.08925)] [[代码](https:\u002F\u002Fgithub.com\u002Fqq456cvb\u002Fdoudizhu-C)]。\n* 斗地主牌戏中的确定化与信息集蒙特卡洛树搜索，CIG 2011 [[论文](https:\u002F\u002Fieeexplore.ieee.org\u002Fdocument\u002F6031993)]。\n\n### 麻将\n\n* 强化学习中的变分oracle引导，ICLR 2022 [[论文](https:\u002F\u002Fopenreview.net\u002Fforum?id=pjqqxepwoMy)]。\n* Suphx：用深度强化学习掌握麻将，arXiv 2020 [[论文](https:\u002F\u002Farxiv.org\u002Fabs\u002F2003.13590)]。\n* 在多人麻将游戏中构建基于马尔可夫决策过程抽象的人工智能玩家的方法，arXiv 2019 [[论文](https:\u002F\u002Farxiv.org\u002Fabs\u002F1904.07491)]。\n* 基于蒙特卡洛模拟和对手模型的计算机麻将玩家开发，IEEE CIG 2017 [[论文](https:\u002F\u002Fieeexplore.ieee.org\u002Fdocument\u002F7317929)]。\n\n### 桥牌\n\n* 提升桥牌人工智能，ICTAI 2017 [[论文](https:\u002F\u002Fieeexplore.ieee.org\u002Fdocument\u002F8372096)]。\n\n### 围棋\n\n* 不依赖人类知识掌握围棋，Nature 2017 [[论文](https:\u002F\u002Fwww.nature.com\u002Farticles\u002Fnature24270)]。\n* 使用深度神经网络和树搜索掌握围棋，Nature 2016 [[论文](https:\u002F\u002Fwww.nature.com\u002Farticles\u002Fnature16961)]。\n* 计算机围棋中的时序差分搜索，机器学习，2012年 [[论文](https:\u002F\u002Flink.springer.com\u002Farticle\u002F10.1007\u002Fs10994-012-5280-0)]。\n* 计算机围棋中的蒙特卡洛树搜索和快速动作价值估计，人工智能，2011年 [[论文](https:\u002F\u002Fwww.ics.uci.edu\u002F~dechter\u002Fcourses\u002Fics-295\u002Fwinter-2018\u002Fpapers\u002Fmcts-gelly-silver.pdf)]。\n* 计算围棋中走法模式的“elo评分”，ICGA期刊，2007年 [[论文](https:\u002F\u002Fhal.inria.fr\u002Ffile\u002Findex\u002Fdocid\u002F149859\u002Ffilename\u002FMMGoPatterns.pdf)]。\n\n### 星际争霸\n\n* 使用多智能体强化学习达到星际争霸II的大师级水平，Nature 2019 [[论文](https:\u002F\u002Fwww.nature.com\u002Farticles\u002Fs41586-019-1724-z)]。\n* 关于星际争霸完整游戏的强化学习，AAAI 2019 [[论文](https:\u002F\u002Faaai.org\u002Fojs\u002Findex.php\u002FAAAI\u002Farticle\u002Fview\u002F4394)]。\n* 稳定化深度多智能体强化学习的经验回放，ICML 2017 [[论文](https:\u002F\u002Fdl.acm.org\u002Fcitation.cfm?id=3305500)]。\n* 星际争霸中多单位作战的合作强化学习，SSCI 2017 [[论文](https:\u002F\u002Fieeexplore.ieee.org\u002Fabstract\u002Fdocument\u002F8280949)]。\n* 使用深度学习从重播中学习星际争霸中的宏观管理，CIG 2017 [[论文](https:\u002F\u002Fieeexplore.ieee.org\u002Fabstract\u002Fdocument\u002F8080430)]。\n* 将强化学习应用于即时战略游戏《星际争霸：母巢之战》中的小规模战斗，CIG 2012 [[论文](https:\u002F\u002Fieeexplore.ieee.org\u002Fabstract\u002Fdocument\u002F6374183)]。\n\n## 会议与研讨会\n* [IEEE计算智能与游戏会议（CIG）](http:\u002F\u002Fwww.ieee-cig.org\u002F)\n* [AAAI游戏强化学习研讨会](http:\u002F\u002Faaai-rlg.mlanctot.info\u002F)\n* [打通博弈论与深度学习](https:\u002F\u002Fnips.cc\u002FConferences\u002F2019\u002FSchedule?showEvent=13158)\n* [IJCAI 2018计算机游戏研讨会](https:\u002F\u002Fwww.lamsade.dauphine.fr\u002F~cazenave\u002Fcgw2018\u002Fcgw2018.html)\n* [IEEE游戏会议（CoG）](http:\u002F\u002Fieee-cog.org\u002F2020\u002F)\n\n## 竞赛\n* [国际计算机游戏协会（ICGA）](http:\u002F\u002Ficga.org\u002F)\n* [年度计算机扑克竞赛](http:\u002F\u002Fwww.computerpokercompetition.org\u002F)\n\n## 相关列表\n* [Awesome StarCraft AI](https:\u002F\u002Fgithub.com\u002FSKTBrain\u002Fawesome-starcraftAI)\n* [Awesome Deep Reinforcement Learning](https:\u002F\u002Fgithub.com\u002Ftigerneil\u002Fawesome-deep-rl)","# awesome-game-ai 快速上手指南\n\n`awesome-game-ai` 并非一个单一的可安装软件包，而是一个精选的**游戏人工智能（多智能体学习）资源列表**。它汇集了开源项目、研究论文、工具包及竞赛信息，涵盖德州扑克、斗地主、星际争霸、围棋等完美与非完美信息博弈场景。\n\n本指南将指导开发者如何利用该列表中的核心统一工具包（如 `RLCard` 或 `OpenSpiel`）快速搭建环境并运行第一个多智能体游戏 AI 示例。\n\n## 环境准备\n\n在开始之前，请确保您的开发环境满足以下要求：\n\n*   **操作系统**: Linux (推荐 Ubuntu 18.04+), macOS, 或 Windows (建议配合 WSL2 使用)\n*   **Python 版本**: 3.7 - 3.10 (多数深度学习库对最新 Python 版本支持尚在完善中)\n*   **前置依赖**:\n    *   `git`: 用于克隆代码仓库\n    *   `pip`: Python 包管理工具\n    *   `CMake` & `Build Essentials`: 部分底层库编译需要 (Linux: `sudo apt-get install build-essential cmake`)\n    *   `SWIG`: 部分接口绑定需要 (Linux: `sudo apt-get install swig`)\n\n> **国内加速建议**：\n> 建议使用清华源或阿里源加速 Python 包安装：\n> ```bash\n> pip config set global.index-url https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple\n> ```\n\n## 安装步骤\n\n由于该仓库是资源列表，我们推荐安装其中最具代表性的通用工具包 **RLCard** (专注于卡牌类游戏多智能体强化学习) 或 **OpenSpiel** (DeepMind 出品的通用博弈框架)。以下以 **RLCard** 为例，因为它对中文开发者友好的斗地主、麻将等项目支持较好且安装简便。\n\n### 1. 克隆资源列表（可选，用于查阅具体项目代码）\n如果您想浏览列表中特定游戏（如 DouZero, PerfectDou）的源码，可先克隆主仓库：\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fdatamllab\u002Fawesome-game-ai.git\ncd awesome-game-ai\n```\n\n### 2. 安装 RLCard 工具包\n直接通过 pip 安装稳定版：\n```bash\npip install rlcard\n```\n\n若需安装包含所有环境依赖的完整版本（推荐）：\n```bash\npip install 'rlcard[all]'\n```\n\n> **注意**：如果您想尝试列表中的 **OpenSpiel**，由于其编译依赖复杂，建议参考其官方文档或使用预编译包：\n> ```bash\n> pip install open-spiel\n> ```\n\n## 基本使用\n\n以下示例展示如何使用 `RLCard` 快速加载一个**斗地主 (Dou Dizhu)** 环境，并运行一个随机智能体进行交互。这是验证环境是否就绪的最快方式。\n\n### 示例：运行斗地主随机对战\n\n创建一个名为 `quick_start.py` 的文件，写入以下代码：\n\n```python\nimport rlcard\nfrom rlcard.agents import RandomAgent\n\n# 1. 初始化环境\n# 支持的游戏包括：'doudizhu', 'mahjong', 'texasholdem', 'goofspiel' 等\nenv = rlcard.make('doudizhu')\n\n# 2. 初始化智能体\n# 这里使用随机智能体作为示例，实际项目中可替换为 DQN, CFR 等算法智能体\nagents = [RandomAgent(num_actions=env.num_actions) for _ in range(env.num_players)]\n\n# 3. 设置智能体到环境\nenv.set_agents(agents)\n\n# 4. 运行一个回合 (Episode)\nprint(\"开始一局斗地主...\")\ntime_step = env.reset()\n\nwhile not time_step.is_done():\n    # 获取当前玩家\n    current_player = env.get_player_id()\n    \n    # 智能体根据当前状态选择动作\n    action = agents[current_player].step(time_step)\n    \n    # 环境执行动作并返回新状态\n    time_step = env.step(action)\n\n# 5. 输出结果\nrewards = env.get_payoffs()\nprint(f\"游戏结束。各玩家收益: {rewards}\")\n```\n\n### 运行代码\n\n在终端执行：\n```bash\npython quick_start.py\n```\n\n如果看到“游戏结束”及收益输出，说明环境配置成功。您可以前往 `awesome-game-ai` 列表中的 [Open-Source Projects](#open-source-projects) 部分，查找特定游戏（如 `DouZero` 或 `PerfectDou`）的仓库链接，克隆对应项目以复现顶尖的 AI 模型。","某游戏工作室的算法团队正致力于开发一款支持多人在线对战的卡牌游戏，急需构建具备高水平博弈能力的非完美信息多智能体 AI。\n\n### 没有 awesome-game-ai 时\n- **资源搜集如大海捞针**：团队成员需分散在 arXiv、GitHub 和各大学术会议网站中手动筛选“斗地主”或“德州扑克”相关的多智能体强化学习论文与代码，耗时数周仍难辨优劣。\n- **复现基准模型困难**：缺乏统一的开源项目参考，开发人员难以找到经过验证的基线代码（如 Libratus 或 DouZero 的实现细节），导致从零造轮子，试错成本极高。\n- **技术选型盲目**：面对完美信息与不完美信息博弈的理论差异，团队因缺少系统的综述文章和分类资源，难以判断哪种算法架构更适合当前游戏的隐藏手牌机制。\n- **前沿动态滞后**：无法及时获取如 AlphaStar 或 OpenAI Five 等最新多智能体协作技术的演进路径，导致技术方案可能起步即落后于行业顶尖水平。\n\n### 使用 awesome-game-ai 后\n- **一站式资源聚合**：团队直接利用该清单中按游戏类型（如 Texas Hold'em、Dou Dizhu）分类的精选库，半天内即可锁定多个高星开源项目和核心论文。\n- **快速搭建基线系统**：通过引用清单中成熟的统一工具包和特定游戏项目代码，开发人员迅速复现了专家级博弈模型，将原型开发周期从数月缩短至两周。\n- **精准匹配技术路线**：借助清单中关于“不完美信息博弈”的专项综述，团队清晰理解了对手建模与概率推理的关键点，迅速确定了适合隐藏信息机制的算法组合。\n- **紧跟学术前沿**：定期查阅该清单更新的会议论文与竞赛成果，确保团队始终掌握多智能体协作学习的最新 SOTA（状态最优）方法，保持技术竞争力。\n\nawesome-game-ai 通过系统化整理多智能体博弈资源，将研发团队从繁琐的信息检索中解放出来，使其能专注于核心算法的创新与落地。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdatamllab_awesome-game-ai_737f58de.png","datamllab","Data Analytics Lab at Rice University","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fdatamllab_e8922862.png","We develop automated and interpretable machine learning algorithms\u002Fsystems with understanding of their theoretical properties.",null,"https:\u002F\u002Fcs.rice.edu\u002F~xh37\u002Findex.html","https:\u002F\u002Fgithub.com\u002Fdatamllab",952,115,"2026-04-09T10:34:44","MIT",1,"","未说明",{"notes":29,"python":27,"dependencies":30},"该仓库是一个游戏 AI（特别是多智能体学习）资源的精选列表，而非单一的独立软件工具。它汇集了针对不同游戏（如德州扑克、斗地主、星际争霸、围棋等）的多个开源项目、论文和竞赛链接。每个列出的子项目（如 DouZero, PerfectDou, ELF 等）都有各自独立的运行环境要求和依赖库，需参考其对应的代码仓库链接获取具体安装说明。",[],[32,33,34],"图像","开发框架","Agent",[36,37,38,39,40,41],"reinforcement-learning","game-ai","imperfect-information-games","awesome","multi-agent","ai",2,"ready","2026-03-27T02:49:30.150509","2026-04-10T20:35:36.846550",[],[],[49,59,67,76,84,93],{"id":50,"name":51,"github_repo":52,"description_zh":53,"stars":54,"difficulty_score":55,"last_commit_at":56,"category_tags":57,"status":43},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",[34,33,32,58],"数据工具",{"id":60,"name":61,"github_repo":62,"description_zh":63,"stars":64,"difficulty_score":55,"last_commit_at":65,"category_tags":66,"status":43},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",[33,32,34],{"id":68,"name":69,"github_repo":70,"description_zh":71,"stars":72,"difficulty_score":42,"last_commit_at":73,"category_tags":74,"status":43},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",149489,"2026-04-10T11:32:46",[33,34,75],"语言模型",{"id":77,"name":78,"github_repo":79,"description_zh":80,"stars":81,"difficulty_score":42,"last_commit_at":82,"category_tags":83,"status":43},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",108322,"2026-04-10T11:39:34",[33,32,34],{"id":85,"name":86,"github_repo":87,"description_zh":88,"stars":89,"difficulty_score":42,"last_commit_at":90,"category_tags":91,"status":43},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",[92,34,32,33],"插件",{"id":94,"name":95,"github_repo":96,"description_zh":97,"stars":98,"difficulty_score":42,"last_commit_at":99,"category_tags":100,"status":43},4721,"markitdown","microsoft\u002Fmarkitdown","MarkItDown 是一款由微软 AutoGen 团队打造的轻量级 Python 工具，专为将各类文件高效转换为 Markdown 格式而设计。它支持 PDF、Word、Excel、PPT、图片（含 OCR）、音频（含语音转录）、HTML 乃至 YouTube 链接等多种格式的解析，能够精准提取文档中的标题、列表、表格和链接等关键结构信息。\n\n在人工智能应用日益普及的今天，大语言模型（LLM）虽擅长处理文本，却难以直接读取复杂的二进制办公文档。MarkItDown 恰好解决了这一痛点，它将非结构化或半结构化的文件转化为模型“原生理解”且 Token 效率极高的 Markdown 格式，成为连接本地文件与 AI 分析 pipeline 的理想桥梁。此外，它还提供了 MCP（模型上下文协议）服务器，可无缝集成到 Claude Desktop 等 LLM 应用中。\n\n这款工具特别适合开发者、数据科学家及 AI 研究人员使用，尤其是那些需要构建文档检索增强生成（RAG）系统、进行批量文本分析或希望让 AI 助手直接“阅读”本地文件的用户。虽然生成的内容也具备一定可读性，但其核心优势在于为机器",93400,"2026-04-06T19:52:38",[92,33]]