Atari Games

Edit

198 papers with code • 63 benchmarks • 6 datasets

The Atari 2600 Games task (and dataset) involves training an agent to achieve high game scores.

( Image credit: Playing Atari with Deep Reinforcement Learning )

Benchmarks

Add a Result

Dataset	Best Model	Paper Title	Compare
Atari 2600 Freeway	TRPO-hash	#Exploration: A Study of Count-Based Exploration for Deep Reinforcement Learning	See all
Atari 2600 Q*Bert	Agent57	Agent57: Outperforming the Atari Human Benchmark	See all
Atari 2600 Seaquest	GDI-H3	GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning	See all
Atari 2600 Breakout	MuZero	Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model	See all
Atari 2600 Space Invaders	GDI-H3	GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning	See all
Atari 2600 Frostbite	MuZero	Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model	See all
Atari 2600 Venture	Agent57	Agent57: Outperforming the Atari Human Benchmark	See all
Atari 2600 Pong	DT	Decision Transformer: Reinforcement Learning via Sequence Modeling	See all
Atari 2600 Gravitar	Agent57	Agent57: Outperforming the Atari Human Benchmark	See all
Atari 2600 Alien	MuZero	Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model	See all
Atari 2600 Amidar	Agent57	Agent57: Outperforming the Atari Human Benchmark	See all
Atari 2600 Asterix	GDI-H3	GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning	See all
Atari 2600 Crazy Climber	Agent57	Agent57: Outperforming the Atari Human Benchmark	See all
Atari 2600 Private Eye	Go-Explore	First return, then explore	See all
Atari 2600 Battle Zone	Agent57	Agent57: Outperforming the Atari Human Benchmark	See all
Atari 2600 Enduro	GDI-I3	GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning	See all
Atari 2600 Beam Rider	MuZero	Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model	See all
Atari 2600 Kangaroo	Agent57	Agent57: Outperforming the Atari Human Benchmark	See all
Atari 2600 Montezuma's Revenge	Go-Explore	First return, then explore	See all
Atari 2600 Ms. Pacman	MuZero	Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model	See all
Atari 2600 Assault	MuZero	Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model	See all
Atari 2600 Bank Heist	MuZero (Res2 Adam)	Online and Offline Reinforcement Learning by Planning with a Learned Model	See all
Atari 2600 Boxing	MuZero	Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model	See all
Atari 2600 Chopper Command	GDI-H3	GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning	See all
Atari 2600 Demon Attack	GDI-H3	GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning	See all
Atari 2600 James Bond	GDI-H3	GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning	See all
Atari 2600 Krull	GDI-H3	GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning	See all
Atari 2600 Asteroids	GDI-H3	GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning	See all
Atari 2600 Bowling	MuZero	Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model	See all
Atari 2600 Centipede	Go-Explore	First return, then explore	See all
Atari 2600 Fishing Derby	MuZero	Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model	See all
Atari 2600 Gopher	GDI-I3	GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning	See all
Atari 2600 HERO	Agent57	Agent57: Outperforming the Atari Human Benchmark	See all
Atari 2600 Road Runner	GDI-H3	GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning	See all
Atari 2600 Time Pilot	MuZero	Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model	See all
Atari 2600 Tutankham	Agent57	Agent57: Outperforming the Atari Human Benchmark	See all
Atari 2600 Atlantis	GDI-H3	GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning	See all
Atari 2600 Double Dunk	UCT	The Arcade Learning Environment: An Evaluation Platform for General Agents	See all
Atari 2600 Ice Hockey	R2D2	Recurrent Experience Replay in Distributed Reinforcement Learning	See all
Atari 2600 Kung-Fu Master	R2D2	Recurrent Experience Replay in Distributed Reinforcement Learning	See all
Atari 2600 Name This Game	MuZero	Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model	See all
Atari 2600 Tennis	GDI-I3	GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning	See all
Atari 2600 Up and Down	GDI-I3	GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning	See all
Atari 2600 Video Pinball	R2D2	Recurrent Experience Replay in Distributed Reinforcement Learning	See all
Atari 2600 Robotank	MuZero	Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model	See all
Atari 2600 Star Gunner	Agent57	Agent57: Outperforming the Atari Human Benchmark	See all
Atari 2600 Wizard of Wor	MuZero	Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model	See all
Atari 2600 Berzerk	Go-Explore	First return, then explore	See all
Atari 2600 Zaxxon	MuZero	Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model	See all
Atari 2600 River Raid	MuZero	Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model	See all
Atari 2600 Skiing	Best Learner	The Arcade Learning Environment: An Evaluation Platform for General Agents	See all
Atari 2600 Pitfall!	Go-Explore	Go-Explore: a New Approach for Hard-Exploration Problems	See all
Atari 2600 Defender	CGP	Evolving simple programs for playing Atari games	See all
Atari 2600 Phoenix	GDI-H3	GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning	See all
Atari 2600 Solaris	Agent57	Agent57: Outperforming the Atari Human Benchmark	See all
Atari-57	MuZero	Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model	See all
Atari 2600 Yars Revenge	Agent57	Agent57: Outperforming the Atari Human Benchmark	See all
Atari 2600 Surround	NoisyNet-Dueling	Noisy Networks for Exploration	See all
Atari games	GDI-H3(200M frames)	GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning	See all
Atari 2600 Elevator Action	Persistent AL	Increasing the Action Gap: New Operators for Reinforcement Learning	See all
Atari 2600 Pooyan	UCT	The Arcade Learning Environment: An Evaluation Platform for General Agents	See all
Atari 2600 Carnival	UCT	The Arcade Learning Environment: An Evaluation Platform for General Agents	See all
Atari 2600 Journey Escape	UCT	The Arcade Learning Environment: An Evaluation Platform for General Agents	See all

Show all 63 benchmarks

Collapse benchmarks

Datasets

Subtasks

Montezuma's Revenge

Greatest papers with code

Deep Exploration via Bootstrapped DQN

tensorflow/models • • NeurIPS 2016

Efficient exploration in complex environments remains a major challenge for reinforcement learning.

Ranked #4 on Atari Games on Atari 2600 Breakout

Atari Games Efficient Exploration

70,696

Paper
Code

MICo: Learning improved representations via sampling-based state similarity for Markov decision processes

google-research/google-research • • 3 Jun 2021

We present a new behavioural distance over the state space of a Markov decision process, and demonstrate the use of this distance as an effective means of shaping the learnt representations of deep reinforcement learning agents.

Atari Games

18,536

Paper
Code

Learning Abstract Models for Strategic Exploration and Fast Reward Transfer

google-research/google-research • • 12 Jul 2020

Model-based reinforcement learning (RL) is appealing because (i) it enables planning and thus more strategic exploration, and (ii) by decoupling dynamics from rewards, it enables fast transfer to new reward functions.

Model-based Reinforcement Learning Montezuma's Revenge

18,536

Paper
Code

On Catastrophic Interference in Atari 2600 Games

google-research/google-research • • 28 Feb 2020

Our study provides a clear empirical link between catastrophic interference and sample efficiency in reinforcement learning.

Atari Games

18,536

Paper
Code

Model-Based Reinforcement Learning for Atari

tensorflow/tensor2tensor • • 1 Mar 2019

We describe Simulated Policy Learning (SimPLe), a complete model-based deep RL algorithm based on video prediction models and present a comparison of several model architectures, including a novel architecture that yields the best results in our setting.

Ranked #12 on Atari Games on Atari games

Atari Games Model-based Reinforcement Learning +1

11,426

Paper
Code

Implicit Quantile Networks for Distributional Reinforcement Learning

google/dopamine • • ICML 2018

In this work, we build on recent advances in distributional reinforcement learning to give a generally applicable, flexible, and state-of-the-art distributional variant of DQN.

Ranked #1 on Atari Games on Atari 2600 Freeway

Atari Games Distributional Reinforcement Learning

9,476

Paper
Code

Prioritized Experience Replay

google/dopamine • • 18 Nov 2015

Experience replay lets online reinforcement learning agents remember and reuse experiences from the past.

Ranked #3 on Atari Games on Atari 2600 Kangaroo

Atari Games

9,476

Paper
Code

RL Unplugged: A Suite of Benchmarks for Offline Reinforcement Learning

deepmind/deepmind-research • • 24 Jun 2020

We hope that our suite of benchmarks will increase the reproducibility of experiments and make it possible to study challenging tasks with a limited computational budget, thus making RL research both more systematic and more accessible across the community.

Atari Games DQN Replay Dataset +1

7,218

Paper
Code

Asynchronous Methods for Deep Reinforcement Learning

tensorpack/tensorpack • • 4 Feb 2016

We propose a conceptually simple and lightweight framework for deep reinforcement learning that uses asynchronous gradient descent for optimization of deep neural network controllers.

Ranked #8 on Atari Games on Atari 2600 Star Gunner

Atari Games

6,053

Paper
Code