chept-2

Second iteration of the ChePT Neural Chess Engine - Refined and Improved

Enter the Virtual Environment

$ python -m venv venv
$ source venv/bin/activate
$ pip install -r requirements.txt

Install and Preprocess the Datasets

$ python data/get_datasets.py --download --preprocess

Pretrain

$ python run.py pretrain

Then accept the default presets when prompted. You may need to use the --batch_size flag depending on your hardware constraints, as the model is extremely video memory intensive when using CUDA.