Open realBravo opened 1 week ago
I don't think the games you have played as a human will be enough to train a NN engine and this will also be hard. Some easier solutions would be:
Both of these solutions will need you to get to the exact position to play the move you would have played, and not just a similar position.
I have cbuild install how do i use it. I've gone through the readme and set everything up. What now.
for option one do i just download maia and turn lichess-opening-explorer to true
I have cbuild install how do i use it. I've gone through the readme and set everything up. What now.
After downloading your games as a pgn file, create an environment in cbuild (pipenv run python cbuild.py -e user --variant standard
) the put the pgn file in the envs/user/source
folder. Then run python cbuild.py -e user -f
and then envs/user/source
folder. Then run python cbuild.py -e user -b
and you will have a bin book created in the envs/user/book
folder.
for option one do i just download maia and turn lichess-opening-explorer to true
Download Maia (and Leela) (see here), and then in your config.yml
under lichess_opening_explorer
change enabled
to true
, source
to player
, player_name
to YOUR LICHESS USERNAME
and lower min_games
to 1 or 2.
Is there a way to make a custom engine based on my games with both black and white. I want to always be playing chess if you catch my drift. I've looked into it and it seem I have to make my own custom engine that is trained off my games but im not sure how to get them off lichess. How do i do that?