The DeepActorCriticAgent class's initialization is straightforward. We will quickly have a look into it and then see how we actually define and initialize the actor and critic networks.
The agent's initialization function is shown here:
class DeepActorCriticAgent(mp.Process): def __init__(self, id, env_name, agent_params): """ An Advantage Actor-Critic Agent that uses a Deep Neural Network to represent it's Policy and the Value function :param id: An integer ID to identify the agent in case there are multiple agent instances :param env_name: Name/ID of the environment :param agent_params: Parameters to be used by the agent """ super(DeepActorCriticAgent, self).__init__() self.id = id self.actor_name ...
