Mastering Atari with Discrete World Models

Danijar Hafner, Timothy P Lillicrap, Mohammad Norouzi, Jimmy Ba

Keywords: Atari, world models, model-based reinforcement learning, reinforcement learning, planning, actor critic

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.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

One-sentence Summary: DreamerV2 is the first agent based on a world model to achieve human-level performance on the Atari benchmark.

Code: [ danijar/dreamerv2](https://github.com/danijar/dreamerv2) + [ 8 community implementations](https://paperswithcode.com/paper/?openreview=0oabwyZbOu)

Data: [Arcade Learning Environment](https://paperswithcode.com/dataset/arcade-learning-environment)