Neural Network to play the game of Go
Original project by Surag Nair
Alpha Zero General
Just run python train.py, it would complete a single cycle of training. You can then continue training with the existing model.
You can simply change the arg to Game() to a desired size, beware though, training on larger board takes much longer and also NN arch might need to get improved as well as some additional changes to game logic and scoring.
You can play vs the model you've trained yourself or download pre-trained models from the releases. Adjust path to NN model in play-nn.py or play-old-vs-new.py and play vs the network yourself or watch it playing itself or other models