Mastering Atari with Discrete World Models

TL;DR

DreamerV2 is a novel reinforcement learning agent that learns from a discrete latent space world model, achieving human-level performance on Atari games and demonstrating versatility in continuous action tasks.

Contribution

It introduces DreamerV2, the first agent to achieve human-level Atari performance using a separately trained discrete world model from pixel inputs.

Findings

Achieves human-level performance on 55 Atari tasks

Surpasses top single-GPU agents IQN and Rainbow in sample efficiency

Successfully applies to continuous control tasks with pixel inputs

Abstract

Intelligent agents need to generalize from past experience to achieve goals in complex environments. World models facilitate such generalization and allow learning behaviors from imagined outcomes to increase sample-efficiency. While learning world models from image inputs has recently become feasible for some tasks, modeling Atari games accurately enough to derive successful behaviors has remained an open challenge for many years. We introduce DreamerV2, a reinforcement learning agent that learns behaviors purely from predictions in the compact latent space of a powerful world model. The world model uses discrete representations and is trained separately from the policy. DreamerV2 constitutes the first agent that achieves human-level performance on the Atari benchmark of 55 tasks by learning behaviors inside a separately trained world model. With the same computational budget and wall-clock time, Dreamer V2 reaches 200M frames and surpasses the final performance of the top single-GPU agents IQN and Rainbow. DreamerV2 is also applicable to tasks with continuous actions, where it learns an accurate world model of a complex humanoid robot and solves stand-up and walking from only pixel inputs.