The project uses the simulator CybORG. Please follow the instructions to install the simulator.
The Cyborg Simulator and pytorch
The objective is each group gets a red agent that maximizes the reward using reinforcement learning.
In the current version, the red agent produces a random action sampled from the action space.
The Agents/RedAgent.py file implements the class that will contain the policy.
Currently, this function implements a random policy. Each group should modify the get action function to obtain an action from their red agent. Similarly, each group needs to modify the train function to implement the training function that will return the red agent policy.
Additionally, each group needs to modify the red train.py to call their implemented train function. Finally, the groups can use the red evaluation.py to evaluate their agent’s policy.