A julia package to train AIs that can play two-player boardgames. It is based on the Alpha-Zero paradigm and is inspired by open implementations like KataGo.
The following games are currently part of Jtac:
- M,n,k-game (including tic-tac-toe)
- Meta tic-tac-toe
- Three men's morris
Extending Jtac by adding support for new games is easy and independent from the rest of Jtac's functionality.
Some tutorial will soon follow here
We are also writing a frontend in Elm to help with inspecting models. It lives at https://github.com/kreibaum/Jtac-frontend