Difference between revisions of "NNUE"

From Chessprogramming wiki
Jump to: navigation, search
(46 intermediate revisions by 3 users not shown)
Line 1: Line 1:
 
'''[[Main Page|Home]] * [[Learning]] * [[Neural Networks]] * NNUE'''
 
'''[[Main Page|Home]] * [[Learning]] * [[Neural Networks]] * NNUE'''
 +
 +
[[FILE:SekienNue.jpg|border|right|thumb|250px| [[:Category:Toriyama Sekien|Toriyama Sekien]] - Nue (鵺) <ref>[https://en.wikipedia.org/wiki/Nue Nue] (鵺) from the [https://en.wikipedia.org/wiki/Konjaku_Gazu_Zoku_Hyakki Konjaku Gazu Zoku Hyakki] (今昔画図続百鬼) by [[:Category:Toriyama Sekien|Toriyama Sekien]], circa 1779, [https://en.wikipedia.org/wiki/Wikimedia_Commons Wikimedia Commons]</ref> <ref>[http://www.talkchess.com/forum3/viewtopic.php?f=2&t=74611&start=2 Re: What does NNUE actually mean] by [[ Henk Drost]], [[CCC]], July 29, 2020</ref> ]]
  
 
'''NNUE''', (&#398;U&#1048;&#1048; Efficiently Updatable Neural Networks)<br/>  
 
'''NNUE''', (&#398;U&#1048;&#1048; Efficiently Updatable Neural Networks)<br/>  
Line 5: Line 7:
 
NNUE was introduced in 2018 by [[Yu Nasu]] <ref>[[Yu Nasu]] ('''2018'''). ''&#398;U&#1048;&#1048; Efficiently Updatable Neural-Network based Evaluation Functions for Computer Shogi''.  Ziosoft Computer Shogi Club, [https://github.com/ynasu87/nnue/blob/master/docs/nnue.pdf pdf] (Japanese with English abstract)</ref>,
 
NNUE was introduced in 2018 by [[Yu Nasu]] <ref>[[Yu Nasu]] ('''2018'''). ''&#398;U&#1048;&#1048; Efficiently Updatable Neural-Network based Evaluation Functions for Computer Shogi''.  Ziosoft Computer Shogi Club, [https://github.com/ynasu87/nnue/blob/master/docs/nnue.pdf pdf] (Japanese with English abstract)</ref>,
 
and was used in Shogi adaptations of [[Stockfish]] such as [[YaneuraOu]] <ref>[https://github.com/yaneurao/YaneuraOu GitHub - yaneurao/YaneuraOu: YaneuraOu is the World's Strongest Shogi engine(AI player), WCSC29 1st winner, educational and USI compliant engine]</ref>,
 
and was used in Shogi adaptations of [[Stockfish]] such as [[YaneuraOu]] <ref>[https://github.com/yaneurao/YaneuraOu GitHub - yaneurao/YaneuraOu: YaneuraOu is the World's Strongest Shogi engine(AI player), WCSC29 1st winner, educational and USI compliant engine]</ref>,
and [[Kristallweizen-kai]] <ref>[https://github.com/Tama4649/Kristallweizen/ GitHub - Tama4649/Kristallweizen: 第29回世界コンピュータ将棋選手権 準優勝のKristallweizenです。]</ref>, apparently with [[AlphaZero]] strength <ref>[http://www.talkchess.com/forum3/viewtopic.php?f=2&t=72754 The Stockfish of shogi] by [[Larry Kaufman]], [[CCC]], January 07, 2020</ref>. NNUE chess engines could be considered as hybrids between [[Neural Networks]] engines  and [[AlphaBeta]] engines.
+
and [[Kristallweizen]] <ref>[https://github.com/Tama4649/Kristallweizen/ GitHub - Tama4649/Kristallweizen: 第29回世界コンピュータ将棋選手権 準優勝のKristallweizenです。]</ref>, apparently with [[AlphaZero]] strength <ref>[http://www.talkchess.com/forum3/viewtopic.php?f=2&t=72754 The Stockfish of shogi] by [[Larry Kaufman]], [[CCC]], January 07, 2020</ref>.  
  
==Stockfish NNUE==
+
=[[Stockfish NNUE]]=
As reported by [[Henk Drost]] <ref>[http://www.talkchess.com/forum3/viewtopic.php?f=7&t=74058 Stockfish NNUE] by [[Henk Drost]], [[CCC]], May 31, 2020</ref>,  
+
As reported by [[Henk Drost]] <ref>[http://www.talkchess.com/forum3/viewtopic.php?f=2&t=74059 Stockfish NN release (NNUE)] by [[Henk Drost]], [[CCC]], May 31, 2020</ref>,  
[[Nodchip]] incorporated NNUE into the chess playing Stockfish 10 as a proof of concept.
+
[[Hisayori Noda|Nodchip]] incorporated NNUE into the chess playing Stockfish 10 as a proof of concept.
 
[[Stockfish NNUE]] was born, and in summer 2020 the computer chess community bursts out enthusiastically due to its rapidly raising [[Playing Strength|playing strength]] with different networks trained using a mixture of [[Supervised Learning|supervised]] and [[Reinforcement Learning|reinforcement learning]] methods -
 
[[Stockfish NNUE]] was born, and in summer 2020 the computer chess community bursts out enthusiastically due to its rapidly raising [[Playing Strength|playing strength]] with different networks trained using a mixture of [[Supervised Learning|supervised]] and [[Reinforcement Learning|reinforcement learning]] methods -
despite the approximately halved search speed, seemingly becoming stronger than its original <ref>[http://www.talkchess.com/forum3/viewtopic.php?f=2&t=74484 Can the sardine! NNUE clobbers SF] by [[Henk Drost]], [[CCC]], July 16, 2020</ref>. In July 2020, the playing code of NNUE is put into the official Stockfish repository as a branch for further development and examination. However, the training code still remains in [[Nodchip]]'s repo <ref>[https://github.com/official-stockfish/Stockfish/issues/2823 NNUE merge · Issue #2823 · official-stockfish/Stockfish · GitHub] by [[Joost VandeVondele]], July 25, 2020</ref>.
+
despite the approximately halved search speed, becoming stronger than its original <ref>[http://www.talkchess.com/forum3/viewtopic.php?f=2&t=74484 Can the sardine! NNUE clobbers SF] by [[Henk Drost]], [[CCC]], July 16, 2020</ref>.  
 +
 
 +
=NNUE Engines=
 +
''see [[:Category:NNUE]]''
  
===Strong points of Stockfish NNUE===
+
Being tempted by the success of [[Stockfish NNUE]] and attracted by how easy the method and small the code is, many engine developers have started testing and applying [[NNUE]]. For quick trials and evaluating before going into serious development, some of them borrowed and/or rewrote NNUE code and use networks from Stockfish NNUE. Most of them reported positive results, such as [[David Carteau]] with [[Orion]] <ref>[http://www.talkchess.com/forum3/viewtopic.php?f=2&t=74828 Orion 0.7 : NNUE experiment] by [[David Carteau]], [[CCC]], August 19, 2020</ref>, [[Ehsan Rashid]] with [[DON]] <ref>[http://www.talkchess.com/forum3/viewtopic.php?f=2&t=72613&start=320#p856640 Re: New engine releases 2020...Don NNUE 2020?] by supersharp77, [[CCC]], August 19, 2020</ref>, various [[Stockfish#Derivatives|Stockfish derivatives]] by [[Michael Byrne]] <ref>[http://talkchess.com/forum3/viewtopic.php?f=2&t=74825 ... the last shall be first ...] by [[Michael Byrne|MikeB]], [[CCC]], 19 Aug 2020</ref>, and [[Volodymyr Shcherbyna]] with [[Igel]] <ref>[http://talkchess.com/forum3/viewtopic.php?f=2&t=67890&start=10#p856742 Introducing Igel chess engine] by [[Volodymyr Shcherbyna]], [[CCC]], 20 Aug 2020</ref> using the ''Night Nurse'' NNUE net by [[Dietrich Kappe]] <ref>[http://talkchess.com/forum3/viewtopic.php?f=2&t=74837 Night Nurse 0.2] by [[Dietrich Kappe]], [[CCC]], August 19, 2020</ref>. In October 2020, [[Daniel Shawul]] added NNUE support à la [[CFish]] into his [[Scorpio#Bitbases|egbbdll]] probing library of [[Scorpio]] <ref>[http://www.talkchess.com/forum3/viewtopic.php?f=7&t=75400&start=22 Re: Hacking around CFish NNUE] by [[Daniel Shawul]], [[CCC]], October 15, 2020</ref> <ref>[http://www.talkchess.com/forum3/viewtopic.php?f=7&t=75415&start=3 Re: How to scale stockfish NNUE score?] by [[Daniel Shawul]], [[CCC]], October 17, 2020</ref>.
* Reuses and gets benefits from the very optimized search function of Stockfish as well as almost all Stockfish's code
 
* Run with CPU only, doesn't require expensive video cards, delete the need of installing video drivers and specific libraries, thus it becomes much easier to install (compare with other NN engines such as [[Leela Chess Zero]]) for users and can run in almost all modern computers
 
* Require much smaller training sets. Some high score networks can be built with the effort of one or a few people. It doesn't require the massive computing from a supercomputer and/or from community
 
  
 
=NN Structure=
 
=NN Structure=
 
The neural network consists of four layers, W1 through W4. The input layer W1 is heavily overparametrized, feeding in the [[Board Representation|board representation]] for various king configurations.
 
The neural network consists of four layers, W1 through W4. The input layer W1 is heavily overparametrized, feeding in the [[Board Representation|board representation]] for various king configurations.
 
The efficiency of the net is due to [[Incremental Updates|incremental update]] of W1 in [[Make Move|make]] and [[Unmake Move|unmake move]],
 
The efficiency of the net is due to [[Incremental Updates|incremental update]] of W1 in [[Make Move|make]] and [[Unmake Move|unmake move]],
where only a fraction of its neurons need to be recalculated. The remaining three layers with 256x2x32-32x32-32x1 neurons are computational less expensive, best calculated using appropriate [[SIMD and SWAR Techniques|SIMD instructions]] like [[AVX2]] on [[x86-64]], or if available, [[AVX-512]].
+
where only a fraction of its neurons need to be recalculated. The remaining three layers with 32x2x256, 32x32 and 32x1 weights are computational less expensive, best calculated using appropriate [[SIMD and SWAR Techniques|SIMD instructions]] like [[AVX2]] on [[x86-64]], or if available, [[AVX-512]].
  
 
[[FILE:NNUE.jpg|none|border|text-bottom]]  
 
[[FILE:NNUE.jpg|none|border|text-bottom]]  
Line 30: Line 32:
  
 
=Forum Posts=
 
=Forum Posts=
 +
==2020 ...==
 
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=72754 The Stockfish of shogi] by [[Larry Kaufman]], [[CCC]], January 07, 2020 » [[Stockfish]], [[Shogi]]
 
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=72754 The Stockfish of shogi] by [[Larry Kaufman]], [[CCC]], January 07, 2020 » [[Stockfish]], [[Shogi]]
 
: [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=72754&start=18 Re: The Stockfish of shogi] by [[Gian-Carlo Pascutto]], [[CCC]], January 18, 2020
 
: [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=72754&start=18 Re: The Stockfish of shogi] by [[Gian-Carlo Pascutto]], [[CCC]], January 18, 2020
 
* [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=74058 Stockfish NNUE] by [[Henk Drost]], [[CCC]], May 31, 2020 » [[Stockfish]]
 
* [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=74058 Stockfish NNUE] by [[Henk Drost]], [[CCC]], May 31, 2020 » [[Stockfish]]
 
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=74059 Stockfish NN release (NNUE)] by [[Henk Drost]], [[CCC]], May 31, 2020
 
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=74059 Stockfish NN release (NNUE)] by [[Henk Drost]], [[CCC]], May 31, 2020
 +
===July===
 
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=74484 Can the sardine! NNUE clobbers SF] by [[Henk Drost]], [[CCC]], July 16, 2020
 
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=74484 Can the sardine! NNUE clobbers SF] by [[Henk Drost]], [[CCC]], July 16, 2020
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=74520 Sergio Vieri second net is out] by Sylwy, [[CCC]], July 21, 2020
 
 
* [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=74531 NNUE accessible explanation] by [[Martin Fierz]], [[CCC]], July 21, 2020
 
* [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=74531 NNUE accessible explanation] by [[Martin Fierz]], [[CCC]], July 21, 2020
 
: [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=74531&start=1 Re: NNUE accessible explanation] by [[Jonathan Rosenthal]], [[CCC]], July 23, 2020
 
: [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=74531&start=1 Re: NNUE accessible explanation] by [[Jonathan Rosenthal]], [[CCC]], July 23, 2020
 
: [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=74531&start=5 Re: NNUE accessible explanation] by [[Jonathan Rosenthal]], [[CCC]], July 24, 2020
 
: [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=74531&start=5 Re: NNUE accessible explanation] by [[Jonathan Rosenthal]], [[CCC]], July 24, 2020
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=74565 Stockfish NNUE] by Lion, [[CCC]], July 25, 2020
+
: [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=74531&start=8 Re: NNUE accessible explanation] by [[Jonathan Rosenthal]], [[CCC]], August 03, 2020
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=74585 SF-NNUE going forward...] by Zenmastur, [[CCC]], July 27, 2020
+
* [https://groups.google.com/d/msg/fishcooking/Wpk-9COzk64/ez643VTkAAAJ BrainLearn NNUE 1.0] by [[Andrea Manzo]], [[Computer Chess Forums|FishCooking]], July 25, 2020 » [[BrainLearn]]
 +
* [https://groups.google.com/d/msg/fishcooking/yWtpz_FY5_Y/RMTG56fkAAAJ ShashChess NNUE 1.0] by [[Andrea Manzo]], [[Computer Chess Forums|FishCooking]], July 25, 2020 » [[ShashChess ]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=74607 LC0 vs. NNUE - some tech details...] by [[Srdja Matovic]], [[CCC]], July 29, 2020 » [[Leela Chess Zero#Lc0|Lc0]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=74611 What does NNUE actually mean] by Paloma, [[CCC]], July 29, 2020
 +
===August===
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=73521&start=59 Re: Minic version 2] by [[Vivien Clauzon]], [[CCC]], August 08, 2020 » [[Minic]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=74777 Neural Networks weights type] by [[Fabio Gobbato]], [[CCC]], August 13, 2020 » [[Stockfish NNUE]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=67890&start=17 Re: Introducing Igel chess engine - Igel and NNUE] by [[Volodymyr Shcherbyna]], [[CCC]], August 19, 2020 » [[Igel]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=74828 Orion 0.7 : NNUE experiment] by [[David Carteau]], [[CCC]], August 19, 2020 » [[Orion]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=74837 Night Nurse 0.2] by [[Dietrich Kappe]], [[CCC]], August 19, 2020 » [[A0lite]], [[Igel]]
 +
===September===
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=74955 Train a neural network evaluation] by [[Fabio Gobbato]], [[CCC]], September 01, 2020 » [[Automated Tuning]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=75016 RubiChess NNUE player implemented] by [[Andreas Matthies]], [[CCC]], September 06, 2020 » [[RubiChess]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=75042 Neural network quantization] by [[Fabio Gobbato]], [[CCC]], September 08, 2020 » [[Neural Networks]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=75049 AVX-512 and NNUE] by [[Gian-Carlo Pascutto]], [[CCC]], September 08, 2020 » [[AVX-512]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=75241 Nemorino 6 (NNUE)] by [[Christian Günther|Florentino]], [[CCC]], September 28, 2020 » [[Nemorino]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=75247 A Crossroad in Computer Chess; Or Desperate Flailing for Relevance] by [[Andrew Grant]], [[CCC]], September 29, 2020
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=75248 NNUE variation] by [[Ed Schroder|Ed Schröder]], [[CCC]], September 29, 2020
 +
===October ...===
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=75296 BONA_PIECE_ZERO] by [[Marco Belli|elcabesa]], [[CCC]], October 04, 2020
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=75335&start=91 Re: Final Release of Ethereal, V12.75] by [[Andrew Grant]], [[CCC]], October 09, 2020 » [[Ethereal]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=75345 Request for someone to train an NNUE for Ethereal] by [[Andrew Grant]], [[CCC]], October 09, 2020
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=75350 Ethereal Tuning - Data Dump] by [[Andrew Grant]], [[CCC]], October 10, 2020
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=75358 Dangerous turn] by [[Dann Corbit]], [[CCC]], October 10, 2020
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=75393 Black crushing white, weird ?] by [[Vivien Clauzon]], [[CCC]], October 14, 2020 » [[Minic#MinicNNUE|MinicNNUE]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=75400 Hacking around CFish NNUE] by [[Maksim Korzh]], [[CCC]], October 15, 2020 » [[CFish]]
 +
: [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=75400&start=22 Re: Hacking around CFish NNUE] by [[Daniel Shawul]], [[CCC]], October 15, 2020 » [[Scorpio]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=75415 How to scale stockfish NNUE score?] by [[Maksim Korzh]], [[CCC]], October 17, 2020
 +
: [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=75415&start=3 Re: How to scale stockfish NNUE score?] by [[Daniel Shawul]], [[CCC]], October 17, 2020
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=75418 Embedding Stockfish NNUE to ANY CHESS ENGINE: YouTube series] by [[Maksim Korzh]], [[CCC]], October 17, 2020 » [[BBC]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=75433 Seer] by [[Gerd Isenberg]], [[CCC]], October 18, 2020 » [[Seer]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=75482 BBC 1.3 + Stockfish NNUE released!] by [[Maksim Korzh]], [[CCC]], October 21, 2020 » [[BBC]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=75500 Mayhem NNUE - New NN engine] by [[Toni Helminen|JohnWoe]], [[CCC]], October 22, 2020 » [[Mayhem NNUE]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=75501 Centipawns vs Millipawns with NNUE] by Madeleine Birchfield, [[CCC]], October 23, 2020 » [[Centipawns]], [[Millipawns]]
 +
* [http://www.talkchess.com/forum3/viewtopic.php?f=7&t=75506 NNUE Question - King Placements] by [[Andrew Grant]], [[CCC]], October 23, 2020 » [[Stockfish NNUE#NNUE Structure|Stockfish NNUE - NNUE Structure]]
  
 
=External Links=
 
=External Links=
 +
==NNUE==
 +
* [https://en.wikipedia.org/wiki/Efficiently_updatable_neural_network Efficiently updatable neural network | Wikipedia]
 +
* [https://github.com/yaneurao/YaneuraOu GitHub - yaneurao/YaneuraOu: YaneuraOu is the World's Strongest Shogi engine(AI player), WCSC29 1st winner, educational and USI compliant engine]
 +
* [https://github.com/Tama4649/Kristallweizen/ GitHub - Tama4649/Kristallweizen: 第29回世界コンピュータ将棋選手権 準優勝のKristallweizenです。]
 +
* [http://qhapaq.hatenablog.com/entry/2018/06/02/221612 次世代の将棋思考エンジン、NNUE関数を学ぼう(その1.ネットワーク構造編) - コンピュータ将棋 Qhapaq], June 02, 2018 (Japanese)
 +
: Learn Next Generation Shogi Thinking Engine, NNUE Function (Part 1. Network Structure) - Computer Shogi
 +
* [http://qhapaq.hatenablog.com/entry/2018/07/08/193316 次世代の将棋思考エンジン、NNUE関数を学ぼう(その2.改造/学習編) - コンピュータ将棋 Qhapaq], July 08, 2018 (Japanese)
 +
: Let's Learn Next Generation Shogi Thinking Engine, NNUE Function (Part 2. Remodeling/Learning) - Computer Shogi
 
* [http://yaneuraou.yaneu.com/2020/06/19/stockfish-nnue-the-complete-guide/ Stockfish NNUE – The Complete Guide], June 19, 2020 (Japanese and English)
 
* [http://yaneuraou.yaneu.com/2020/06/19/stockfish-nnue-the-complete-guide/ Stockfish NNUE – The Complete Guide], June 19, 2020 (Japanese and English)
* [https://github.com/nodchip/Stockfish GitHub - nodchip/Stockfish: UCI chess engine] ([[Stockfish NNUE]] by [[Nodchip]])
+
* [http://yaneuraou.yaneu.com/2020/08/21/3-technologies-in-shogi-ai-that-could-be-used-for-chess-ai/ 3 technologies in shogi AI that could be used for chess AI] by [[Motohiro Isozaki]], August 21, 2020 » [[Stockfish NNUE]]
 +
* [https://github.com/nodchip/Stockfish GitHub - nodchip/Stockfish: UCI chess engine] ([[Stockfish NNUE]] by [[Hisayori Noda|Nodchip]])
 
* [https://www.qhapaq.org/shogi/shogiwiki/stockfish-nnue/ Stockfish NNUE Wiki]
 
* [https://www.qhapaq.org/shogi/shogiwiki/stockfish-nnue/ Stockfish NNUE Wiki]
* [https://github.com/official-stockfish/Stockfish/issues/2823 NNUE merge · Issue #2823 · official-stockfish/Stockfish · GitHub] by [[Joost VandeVondele]], July 25, 2020 <ref>[http://www.talkchess.com/forum3/viewtopic.php?f=2&t=74560 An info] by Sylwy, [[CCC]], July 25, 2020</ref>
+
==Misc==
* [https://www.comp.nus.edu.sg/~sergio-v/nnue/ Index of /~sergio-v/nnue] by [[Sergio Vieri]]
+
* [https://en.wikipedia.org/wiki/Nue Nue from Wikipedia]
  
 
=References=  
 
=References=  
 
<references />
 
<references />
 
'''[[Neural Networks|Up one Level]]'''
 
'''[[Neural Networks|Up one Level]]'''
 +
[[Category:Toriyama Sekien]]

Revision as of 14:16, 23 October 2020

Home * Learning * Neural Networks * NNUE

Toriyama Sekien - Nue (鵺) [1] [2]

NNUE, (ƎUИИ Efficiently Updatable Neural Networks)
a Neural Network architecture intended to replace the evaluation of Shogi, chess and other board game playing alpha-beta searchers running on a CPU. NNUE was introduced in 2018 by Yu Nasu [3], and was used in Shogi adaptations of Stockfish such as YaneuraOu [4], and Kristallweizen [5], apparently with AlphaZero strength [6].

Stockfish NNUE

As reported by Henk Drost [7], Nodchip incorporated NNUE into the chess playing Stockfish 10 as a proof of concept. Stockfish NNUE was born, and in summer 2020 the computer chess community bursts out enthusiastically due to its rapidly raising playing strength with different networks trained using a mixture of supervised and reinforcement learning methods - despite the approximately halved search speed, becoming stronger than its original [8].

NNUE Engines

see Category:NNUE

Being tempted by the success of Stockfish NNUE and attracted by how easy the method and small the code is, many engine developers have started testing and applying NNUE. For quick trials and evaluating before going into serious development, some of them borrowed and/or rewrote NNUE code and use networks from Stockfish NNUE. Most of them reported positive results, such as David Carteau with Orion [9], Ehsan Rashid with DON [10], various Stockfish derivatives by Michael Byrne [11], and Volodymyr Shcherbyna with Igel [12] using the Night Nurse NNUE net by Dietrich Kappe [13]. In October 2020, Daniel Shawul added NNUE support à la CFish into his egbbdll probing library of Scorpio [14] [15].

NN Structure

The neural network consists of four layers, W1 through W4. The input layer W1 is heavily overparametrized, feeding in the board representation for various king configurations. The efficiency of the net is due to incremental update of W1 in make and unmake move, where only a fraction of its neurons need to be recalculated. The remaining three layers with 32x2x256, 32x32 and 32x1 weights are computational less expensive, best calculated using appropriate SIMD instructions like AVX2 on x86-64, or if available, AVX-512.

NNUE.jpg

NNUE structure with incremental update [16]

Publications

  • Yu Nasu (2018). ƎUИИ Efficiently Updatable Neural-Network based Evaluation Functions for Computer Shogi. Ziosoft Computer Shogi Club, pdf (Japanese with English abstract)

Forum Posts

2020 ...

Re: The Stockfish of shogi by Gian-Carlo Pascutto, CCC, January 18, 2020

July

Re: NNUE accessible explanation by Jonathan Rosenthal, CCC, July 23, 2020
Re: NNUE accessible explanation by Jonathan Rosenthal, CCC, July 24, 2020
Re: NNUE accessible explanation by Jonathan Rosenthal, CCC, August 03, 2020

August

September

October ...

Re: Hacking around CFish NNUE by Daniel Shawul, CCC, October 15, 2020 » Scorpio
Re: How to scale stockfish NNUE score? by Daniel Shawul, CCC, October 17, 2020

External Links

NNUE

Learn Next Generation Shogi Thinking Engine, NNUE Function (Part 1. Network Structure) - Computer Shogi
Let's Learn Next Generation Shogi Thinking Engine, NNUE Function (Part 2. Remodeling/Learning) - Computer Shogi

Misc

References

  1. Nue (鵺) from the Konjaku Gazu Zoku Hyakki (今昔画図続百鬼) by Toriyama Sekien, circa 1779, Wikimedia Commons
  2. Re: What does NNUE actually mean by Henk Drost, CCC, July 29, 2020
  3. Yu Nasu (2018). ƎUИИ Efficiently Updatable Neural-Network based Evaluation Functions for Computer Shogi. Ziosoft Computer Shogi Club, pdf (Japanese with English abstract)
  4. GitHub - yaneurao/YaneuraOu: YaneuraOu is the World's Strongest Shogi engine(AI player), WCSC29 1st winner, educational and USI compliant engine
  5. GitHub - Tama4649/Kristallweizen: 第29回世界コンピュータ将棋選手権 準優勝のKristallweizenです。
  6. The Stockfish of shogi by Larry Kaufman, CCC, January 07, 2020
  7. Stockfish NN release (NNUE) by Henk Drost, CCC, May 31, 2020
  8. Can the sardine! NNUE clobbers SF by Henk Drost, CCC, July 16, 2020
  9. Orion 0.7 : NNUE experiment by David Carteau, CCC, August 19, 2020
  10. Re: New engine releases 2020...Don NNUE 2020? by supersharp77, CCC, August 19, 2020
  11. ... the last shall be first ... by MikeB, CCC, 19 Aug 2020
  12. Introducing Igel chess engine by Volodymyr Shcherbyna, CCC, 20 Aug 2020
  13. Night Nurse 0.2 by Dietrich Kappe, CCC, August 19, 2020
  14. Re: Hacking around CFish NNUE by Daniel Shawul, CCC, October 15, 2020
  15. Re: How to scale stockfish NNUE score? by Daniel Shawul, CCC, October 17, 2020
  16. Image from Yu Nasu (2018). ƎUИИ Efficiently Updatable Neural-Network based Evaluation Functions for Computer Shogi. Ziosoft Computer Shogi Club, pdf (Japanese with English abstract)

Up one Level