neural_network_chess VS python-chess

Basically, MCTS is used in combination with reinforcement learning. This ensures that we get better evaluations for non terminal positions in training games. You might want to have a look at my free book @ https://github.com/asdfjkl/neural_network_chess

So consider instead keeping a running score of the position, rather than recalculating it in full every time. One of the main tricks behind Stockfish NNUE, the current top chess engine, is a clever way of not needing to fully recalculate the full neural net every time, but to do a delta update for each move for a substantial part of the network. I forget the main paper about that, but you can read about it in https://github.com/asdfjkl/neural_network_chess.

My book https://github.com/asdfjkl/neural_network_chess covers some of these topics...

Neural Networks For Chess

Shameless self-plug: I wrote a book about neural networks in chess and you can get a free PDF here https://github.com/asdfjkl/neural_network_chess (no ads, no fremium)

If you do get around to the pi version- I highly recommend this library python-chess. It can handle valid moves in addition to querying an engine like stockfish.

There is a very good Python chess library for computing these sorts of things: https://python-chess.readthedocs.io/en/latest/ If you backed the board with this, you could do it pretty easily.

I've never done it, but it should be quite easy with python-chess to read in a detailed opening book, go through each game, and then find the last position in each game that occurs in the opening book.

pip install chess https://python-chess.readthedocs.io/en/latest/

database.lichess.org I use the python-chess package.

The tool requires this library to be installed.

python-chess

Do you need to reinvent the wheel? There's python-chess, for example: https://python-chess.readthedocs.io/en/latest/