Chesstakov

A deep neural network based chess program.

Dependencies:

cat *.pgn | pgn-extract -s -D -#3000

to generate pgn files, each containing 3000 games.

Run parser.py to extract at most 10 random position from each game. For each pgn file, a new HDF5 file will be created.
Run concat_h5.py to generate merge all HDF5 files into one.
Run generate_training_dataset.py.
Run board_selection_tranining.py.

Name		Name	Last commit message	Last commit date
Latest commit History 19 Commits
chessboard_interface		chessboard_interface
README.md		README.md
__init__.py		__init__.py
autoencoder_trainer.py		autoencoder_trainer.py
aux.py		aux.py
board_selection_training.py		board_selection_training.py
board_selection_training_no_previous_autoencoder.py		board_selection_training_no_previous_autoencoder.py
chesstakov.py		chesstakov.py
chesstakov_negamax_vs_random.py		chesstakov_negamax_vs_random.py
chesstakov_negamax_vs_stockfish.py		chesstakov_negamax_vs_stockfish.py
chesstakov_vs_random.py		chesstakov_vs_random.py
concat_h5.py		concat_h5.py
evaluate_training.py		evaluate_training.py
generate_autoencoder_dataset.py		generate_autoencoder_dataset.py
generate_position_pair_dataset.py		generate_position_pair_dataset.py
generate_training_dataset.py		generate_training_dataset.py
model.py		model.py
parser.py		parser.py