Any results?
Have you tested the model in some games? Does it always make legal moves? Whats its estimated elo?
Have you tested the model in some games? Does it always make legal moves? Whats its estimated elo?
I had some interesting games with it that I collected in the demo app ( https://huggingface.co/spaces/philipp-zettl/ChessPT/discussions/1#6700138e4e07efa9575af7b7 ).
In this app, I implemented a loop that tests predictions against their validity and enforces the app to respond with only valid moves. The model itself produces increasingly more wrong moves the longer you progress in the game. I encountered plenty cases when it gets stuck on predicting wrong moves :/
I didn't evaluate it's ELO. From my experience so far it's not even comparable with someone that understands the basic rules of chess ':D