imported>Ohtashinichiro 来自 2026年3月30日 (一) 08:33

2026-03-30T08:33:29Z

新页面

{{NoteTA
|G1 = IT
}}
{{Infobox software
| name = PhoenixGo
| title = PhoenixGo
| logo = Image:PhoenixGo Logo.jpg
| logo caption = PhoenixGo
| logo_size =
| logo_alt =
| screenshot = 
| caption =
| screenshot_size =
| screenshot_alt =
| collapsible =
| author = [[腾讯]]公司[[微信]]翻譯團隊<ref name="qqnews20180511">{{Cite web |url=http://tech.qq.com/a/20180511/024785.htm |title=腾讯围棋AI技术PhoenixGo正式开源 |accessdate=2018-05-12 |date=2018-05-11 |publisher=[[腾讯公司]] |archive-date=2019-02-17 |archive-url=https://web.archive.org/web/20190217142216/http://tech.qq.com/a/20180511/024785.htm |dead-url=no }}</ref>
| developer =
| released = {{Start date and age|2018|05|11|df=yes}}
| discontinued =
| programming language = [[C++]]、[[Python]]、[[C语言]]
| operating system = [[Linux]]、[[Windows]]、[[MacOS]]
| platform =
| size =
| language =
| language count = 
| language footnote =
| genre = [[電腦圍棋]]
| license = [[BSD许可证]]<ref name="license">{{Cite web |url=https://github.com/Tencent/PhoenixGo/blob/master/LICENSE.TXT |title=PhoenixGo/LICENSE.TXT at master · Tencent/PhoenixGo |accessdate=2018-05-11 |archive-date=2019-02-17 |archive-url=https://web.archive.org/web/20190217142216/https://github.com/Tencent/PhoenixGo/blob/master/LICENSE.TXT |dead-url=no }}</ref>
| repo = {{URL|https://github.com/Tencent/PhoenixGo}}
| website = {{URL|https://github.com/Tencent/PhoenixGo}}
| standard =
| AsOf =
}}
'''PhoenixGo'''，即'''凤凰围棋'''，為[[腾讯]]公司[[微信]]翻譯團隊所開發的[[電腦圍棋]]軟體<ref name="qqnews20180511"/>。

== 簡介 ==
{{See also|AlphaGo Zero}}
PhoenixGo是由[[騰訊]]公司的[[微信]]翻譯團隊<ref name="qqnews20180511"/>依照[[Google DeepMind]]在科學期刊《[[自然 (期刊)|自然]]》上對於[[AlphaGo Zero]]所發表的論文《{{lang|en|Mastering the game of Go without human knowledge}}<ref name="agz-nature">{{Cite journal |last=Silver |first=David |last2=Schrittwieser |first2=Julian |last3=Simonyan |first3=Karen |last4=Antonoglou |first4=Ioannis |last5=Huang |first5=Aja |last6=Guez |first6=Arthur |last7=Hubert |first7=Thomas |last8=Baker |first8=Lucas |last9=Lai |first9=Matthew |last10=Bolton |first10=Adrian |last11=Chen |first11=Yutian |title=Mastering the game of Go without human knowledge |url=https://www.nature.com/articles/nature24270 |journal=Nature |language=en |date=2017-10 |volume=550 |issue=7676 |doi=10.1038/nature24270 |issn=1476-4687 |access-date=2018-05-11 |archive-date=2018-04-12 |archive-url=https://web.archive.org/web/20180412102056/https://www.nature.com/articles/nature24270 |dead-url=no }}</ref>》所實做出的[[開源]]電腦圍棋程式<ref name="weiqi20180511">{{Cite web |url=http://weiqi.qq.com/news/9353.html |title=微信开源PhoenixGo：业界领先的人工智能围棋 |accessdate=2018-05-12 |date=2018-05-11 |publisher=[[野狐围棋]] |archive-date=2018-05-12 |archive-url=https://web.archive.org/web/20180512044359/http://weiqi.qq.com/news/9353.html |dead-url=no }}</ref><ref name="github">{{Cite web |url=https://github.com/Tencent/PhoenixGo |title=Tencent/PhoenixGo: Go AI program which implement the AlphaGo Zero paper |accessdate=2018-05-12 |archive-date=2018-05-16 |archive-url=https://web.archive.org/web/20180516063137/https://github.com/Tencent/PhoenixGo |dead-url=no }}</ref>。也就是不使用人類棋譜與累積的圍棋知識，僅實做[[圍棋規則]]，使用單一[[類神經網路]]從自我對弈中學習（不像[[AlphaGo]]以人類角度思考，設計了Policy Network與Value Network）。

團隊除了以[[BSD许可证]]釋出可供對弈的程式碼以外<ref name="license"/>，另外釋出了20 blocks的訓練資料供下載<ref name="qqnews20180511"/><ref name="github-release">{{Cite web |url=https://github.com/Tencent/PhoenixGo/releases |title=Releases · Tencent/PhoenixGo |accessdate=2018-05-12 |archive-date=2019-02-17 |archive-url=https://web.archive.org/web/20190217142216/https://github.com/Tencent/PhoenixGo/releases |dead-url=no }}</ref>，但由於未放出訓練程式碼<ref name="weiqi20180511"/>或完整的原始訓練資料（如棋譜）。但開放的20 blocks版本棋力相較於比賽版本弱不少，且使用不同的架構<ref>{{Cite web |url=https://github.com/gcp/leela-zero/issues/1405| title=Phoenix Go open sources · Issue #1405 · gcp/leela-zero |accessdate=2018-05-16}}</ref>，相較於同時間開源且開放訓練資料的其他軟體（主要是[[Facebook]]的[[ELF OpenGo]]），社群可重複利用的資源較少。

現今在[[野狐围棋]]所使用的帳號名稱<code>金毛測試</code>與<code>金毛陪練</code>，故暱稱為「金毛」<ref name="weiqi20180511"/>。

== 成績 ==
=== 野狐围棋 ===
PhoenixGo曾經以多個帳號在同集團的[[野狐围棋]]上對弈{{notetag|帳號包括<code>BensonDarr</code>、<code>时空奇点</code>、<code>金毛测试</code>、<code>金毛陪练</code>。}}：
* 帳號<code>金毛測試</code>以單片[[NVIDIA Tesla|NVIDIA Tesla P40]]進行對弈，勝率在90%以上<ref name="weiqi20180511"/>，並且列為10段帳號<ref>{{Cite web |url=http://www.foxwq.com/rank/monthsort |title=腾讯围棋 |accessdate=2018-05-12 |archive-date=2018-05-18 |archive-url=https://web.archive.org/web/20180518191400/http://www.foxwq.com/rank/monthsort |dead-url=no }}</ref>。
* 帳號<code>BensonDarr</code>勝率也在90%以上，並且列為10段帳號<ref>{{Cite web |url=http://www.foxwq.com/rank/rp |title=腾讯围棋 |accessdate=2018-05-12 |archive-date=2018-05-12 |archive-url=https://web.archive.org/web/20180512043643/http://www.foxwq.com/rank/rp |dead-url=no }}</ref>。

=== CGOS ===
PhoenixGo在2018年3月以<code>cronus</code>為名在[[CGOS]]上對弈<ref name="github"/>，其中<code>cronus_v0.1.3</code><ref>{{Cite web |url=http://www.yss-aya.com/cgos/19x19/cross/cronus_v0.1.3.html |title=Crosstable for cronus_v0.1.3 |accessdate=2018-05-12 |date=2018-03-02 |archive-date=2018-05-12 |archive-url=https://web.archive.org/web/20180512043755/http://www.yss-aya.com/cgos/19x19/cross/cronus_v0.1.3.html |dead-url=no }}</ref>是CGOS在2019年9月前排名最高的對弈程式（被<code>tesla_sp</code>超越<ref>{{Cite web |url=http://www.yss-aya.com/cgos/19x19/cross/tesla_sp.html |title=Crosstable for tesla_sp |accessdate=2020-05-07 |language=en |archive-date=2020-03-11 |archive-url=https://web.archive.org/web/20200311043416/http://www.yss-aya.com/cgos/19x19/cross/tesla_sp.html |dead-url=no }}</ref>），{{As of|2020|5|alt=於2020年5月}}，BayesElo約4435分<ref name="cgos-bayes-19x19">{{Cite web |url=http://www.yss-aya.com/cgos/19x19/bayes.html |title=19x19 All Time Ranks |accessdate=2020-05-07 |language=en |archive-date=2018-01-17 |archive-url=https://web.archive.org/web/20180117032048/http://www.yss-aya.com/cgos/19x19/bayes.html |dead-url=no }}</ref>。

=== 2018世界人工智能围棋大赛 ===
PhoenixGo在2018年4月以「凤凰围棋」身份參加自家[[騰訊]]主辦的2018世界人工智能围棋大赛，並以全勝戰績取得冠軍<ref>{{Cite web |url=http://www.sport.gov.cn/n316/n343/n1192/c856828/content.html |title=2018世界人工智能围棋大赛在福州市落幕 |accessdate=2018-05-12 |date=2018-05-02 |publisher=[[国家体育总局]] |archive-date=2018-05-12 |archive-url=https://web.archive.org/web/20180512112225/http://www.sport.gov.cn/n316/n343/n1192/c856828/content.html |dead-url=no }}</ref>。

== 相關連結 ==
* [[AlphaGo Zero]]，所參考論文的電腦圍棋軟體。
* [[计算机围棋]]
* [[圍棋軟體]]

== 參考資料 ==
{{Reflist|2}}

== 註解 ==
{{NoteFoot}}

== 外部連結 ==
* {{Official|https://github.com/Tencent/PhoenixGo}}

[[Category:围棋软件]]
[[Category:游戏人工智能]]
[[Category:人工智能应用]]
[[Category:使用BSD许可证的软件]]
[[Category:用C++編程的自由軟體]]
[[Category:用Python編程的自由軟體]]
[[Category:用C編程的自由軟體]]

PhoenixGo - 版本历史

imported>Ohtashinichiro 来自 2026年3月30日 (一) 08:33