Programmer's guide - QueensGambit/CrazyAra GitHub Wiki

Programmer's guide

First steps

C++

The engine is written in C++ with the following source directory.

• Follow the compilation instructions on how to setup a working environment.

Python

Parts of the project such as training neural networks and converting pgn-files into training data is still done in python. There is also a MCTS search engine in python which was used for releases <= 0.5.1.

• Make sure you have installed the dependencies by following the installation guide.

• Copy and rename the DeepCrazyhouse/configs/main_config_sample.py to DeepCrazyhouse/configs/main_config.py and adjust the paths to the correct location on your file system. The first entries to change are model_architecture_dir and model_weights_dir.

• The most important source files of this project are the files which are included in a release.

• You can find more details about the code structure in the following README.md.

• If you want to create a PR which includes jupyter notebook files, make sure that you don't commit the output of the cells by using the script ipynb_output_filter.py of the following repository.

  • https://github.com/toobaz/ipynb_output_filter

The repository includes instructions on how to set-up an automatic workflow.