Scalable Distributed Training for Deep RL via Importance-Weighted Actor-Learner Architecture and V-trace
Deep Reinforcement Learning (DeepRL) has achieved remarkable success in a range of tasks, from continuous control problems in robotics to playing games