|
|
|
|
|
|
by thomasahle
2355 days ago
|
|
|
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 matrix multiplication. I wonder how well it would compete with this GPT-2 engine. |
|
|