This release contains two scenarios:
'discover_strike' is our simple scenario, which shows the agent learning modified 'push' actions.
'cook' is our complex scenario, which shows the agent learning how to cook something, learning to modify actions in its cook action sequence.
Enjoy!