Chess engine powered with reinforcement learning.
Many techniques used in this engine are inspired by the revolutionary AlphaZero program as well as the extensive Chess Programming Wiki.
- Parallel Monte Carlo Tree Search
- Bitboard move generation
- Incremental input layer updates
- CMake 3.14+
- LibTorch 1.8.1+
- (optional) GoogleTest for tests
GCC 8+ is required for debug and test builds.
$ ./compile.sh
$ build/bin/nczero
# cp build/bin/nczero /usr/bin
$ ./test.sh