Advantage Actor-Critic (A2C) jointly trains a policy and a value function with a single composite loss.
