kth.sePublications KTH
Operational message
There are currently operational disruptions. Troubleshooting is in progress.
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Chessformer: A Chess-Playing Transformer Model
KTH, School of Electrical Engineering and Computer Science (EECS).
KTH, School of Electrical Engineering and Computer Science (EECS).
2025 (English)Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
Abstract [en]

We introduce Chessformer: a chess-playing transformer model, trained to play by predicting the next move in human chess games given all the previous moves. The purpose is to show that transformers can learn chess when trained in this way and to train a model with a more human-like playing style than other chess-bots.

Chessformer is trained on a dataset of $16$ million games, and through this learns to play both legal and strategic moves. In random positions outside the training data the best version of Chessformer plays a legal move $99.2\% \pm 0.1\%$ of the time. Playing against Stockfish our model achieves an Elo rating of $1198 \pm 22$. Based on an analysis of its moves Chessformer has a playing style very different from both humans and Stockfish, despite being trained on human chess games.

Abstract [sv]

Vi introducerar \textit{Chessformer}: en schackspelande transformermodell, tränad till att spela genom att förutse nästa drag i mänskliga schackspel. Syftet är att visa att transformers kan lära sig schack då de tränas på detta vis och att träna en modell med en mer mänsklig spelstil än andra schackbotar.

Chessformer är tränad på ett dataset av $16$ miljoner schackmatcher, och lär sig genom detta att spela regelrätta och strategiska drag. I slumpvalda positioner utanför träningsdata spelar den bästa versionen av Chessformer regelrätta drag $99.2\% \pm 0.1\%$ av gångerna. Under spel mot Stockfish uppnår vår modell en Elo-rating på $1198 \pm 22$. Baserat på en analys av dess drag har Chessformer en spelstil som skiljer sig avsevärt från både människor och Stockfish, trots att den tränas på matcher mellan människor.

Place, publisher, year, edition, pages
2025. , p. 513-521
Series
TRITA-EECS-EX ; 2025:151
National Category
Electrical Engineering, Electronic Engineering, Information Engineering
Identifiers
URN: urn:nbn:se:kth:diva-376172OAI: oai:DiVA.org:kth-376172DiVA, id: diva2:2034542
Supervisors
Examiners
Projects
Kandidatexamensarbete i Elektroteknik 2025, EECS, KTHAvailable from: 2026-02-02 Created: 2026-02-02

Open Access in DiVA

fulltext(80627 kB)21 downloads
File information
File name FULLTEXT01.pdfFile size 80627 kBChecksum SHA-512
35ce0a386dafe4649eb99cbe0efdfed651a3c9044e3339612422234d17a7e8ec21d4fd4aa201500c3c7a8f57194994b78b3e0cfbd5319ecd49f18a5d8a7ff775
Type fulltextMimetype application/pdf

By organisation
School of Electrical Engineering and Computer Science (EECS)
Electrical Engineering, Electronic Engineering, Information Engineering

Search outside of DiVA

GoogleGoogle Scholar
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 3913 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf