Accompanying repository for COLA: Consistent Learning with Opponent-Learning Awareness
We provide one Google Colab notebook, in which you can run COLA and other general-sum learning algorithms on a set of zero- and general-sum games. The notebook allows you plot the learning behavior, the consistency loss of COLA, and gradient fields.
Please raise issues if you encounter any problems!