thomasahle/fastchess
View on GitHub ↗ ★ 103 Python
Predicts the best chess move with 27.5% accuracy by a single matrix multiplication
5 mentions on Hacker News by 2 people: 0 stories, 5 comments · first 2019-06-01 · latest 2020-12-24
20082026
-
…I'm personally running bits for sunfish https://github.com/thomasahle/sunfish and fastchess https://github.com/thomasahle/fastchess There are way too many stockfish clones on there. I hope more hobby chess engine builders will join in.
comment by thomasahle · 2020-12-24 · in Building My Own Chess Engine -
Logistic regression can learn some quite amazing things. I trained a linear function to play chess: https://github.com/thomasahle/fastchess and it manages to predict the next moves of top engine games with 27% accuracy. A benefit of…
comment by thomasahle · 2020-06-25 · in Logistic regression from scratch -
I made similar chess engine using FastText: https://github.com/thomasahle/fastchess It is surprising to me that you can predict optimal/strong engine moves with 27% accuracy using a completely trivial linear model, that is by a single…
comment by thomasahle · 2020-01-08 · in Training GPT-2 to Play Chess -
…similar, learning a chess representation, and move predictor with a simple linear fastText model: https://github.com/thomasahle/fastchess It turns out that you can get surprisingly good accuracy (27.5%) at predicting moves from the chess…
comment by thomasahle · 2019-08-18 · in Chess2vec: Learning Vector Representations for Chess (2018) [pdf] -
There is an simple AI version with facebook fastText: https://github.com/thomasahle/fastchess
comment by ghosthamlet · 2019-06-01 · in Sunfish: A simple but strong chess engine written in Python