List your company
ResearchFormerly WizWand

Reinforcement Learning on Atari 2600 Montezuma's Revenge

18,003,200ScoreGo-Explore
-720,1284,140,7369,001,60013,862,464Jul 31, 2018Mar 24, 2019Nov 15, 2019Jul 8, 2020Mar 1, 2021Oct 23, 2021Jun 16, 2022
Updated 8mo ago

Evaluation results

MethodLinks
Go-ExploreAlgorithm Category=Reinforcement Learning, Domain knowledge utilization=true, Score Aggregation=Best2019.0118,003,200—
Human World RecordAlgorithm Category=Human2019.011,219,200—
Go-ExploreAlgorithm Category=Reinforcement Learning, Domain knowledge utilization=true, Score Aggregation=Average2019.01666,474—
LfSDAlgorithm Category=Imitation Learning, Score Aggregation=Best2019.0174,500—
Go-ExploreAlgorithm Category=Reinforcement Learning, Domain knowledge utilization=false, Score Aggregation=Average2019.0143,763—
TDC+CMCAlgorithm Category=Imitation Learning2019.0141,098—
Human ExpertAlgorithm Category=Human2019.0134,900—
Ape-X DQfDAlgorithm Category=Imitation Learning2019.0129,384—
PPO+CoExAlgorithm Category=Reinforcement Learning2019.0111,540—
RNDAlgorithm Category=Reinforcement Learning2019.0111,347—
BYOL-ExplorePure-exploration regime=true, Episodes=10, Seeds=32022.065,146.73—
Average HumanPure-exploration regime=true2022.064,753.3—
Average HumanAlgorithm Category=Human2019.014,753—
DQfDAlgorithm Category=Imitation Learning2019.014,638—
DQN-CTSAlgorithm Category=Reinforcement Learning2019.013,705—
DeepCSAlgorithm Category=Reinforcement Learning2019.013,500—
UBEAlgorithm Category=Reinforcement Learning2019.013,000—
DQN MMC CTSTraining frames=100 million, Sticky actions (s)=0.25, Frame skip=52018.072,941.9—
Feature-EBAlgorithm Category=Reinforcement Learning2019.012,745—
ReactorAlgorithm Category=Reinforcement Learning2019.012,643.5—
IMPALAAlgorithm Category=Reinforcement Learning2019.012,643.5—
DQN-PixelCNNAlgorithm Category=Reinforcement Learning2019.012,514—
Ape-XAlgorithm Category=Reinforcement Learning2019.012,500—
RNDPure-exploration regime=true, Episodes=10, Seeds=32022.062,305.65—
DQN MMC PIXELCNNTraining frames=100 million, Sticky actions (s)=0.25, Frame skip=52018.071,671.7—
DQN MMC + SRTraining frames=100 million, Sticky actions (s)=0.25, Frame skip=52018.071,395.41,121.8
A3C-CTSAlgorithm Category=Reinforcement Learning2019.011,127—
RNDTraining frames=100 million, Sticky actions (s)=0.25, Frame skip=52018.07524.8314
SARSAAlgorithm Category=Reinforcement Learning2019.01259—
BASS-hashAlgorithm Category=Reinforcement Learning2019.01238—
RainbowAlgorithm Category=Reinforcement Learning2019.01154—
GorilaAlgorithm Category=Reinforcement Learning2019.0184—
A3CAlgorithm Category=Reinforcement Learning2019.0167—
DDQNAlgorithm Category=Reinforcement Learning2019.0142—
Duel. DQNAlgorithm Category=Reinforcement Learning2019.0122—
Prior. DQNAlgorithm Category=Reinforcement Learning2019.0113—
LinearAlgorithm Category=Reinforcement Learning2019.0110.7—
ICMPure-exploration regime=true, Episodes=10, Seeds=32022.062.68—
DQNTraining frames=100 million, Sticky actions (s)=0.25, Frame skip=52018.0700
DQN MMCTraining frames=100 million, Sticky actions (s)=0.25, Frame skip=52018.0700
DQNAlgorithm Category=Reinforcement Learning2019.010—
MP-EBAlgorithm Category=Reinforcement Learning2019.010—
Pop-ArtAlgorithm Category=Reinforcement Learning2019.010—
ESAlgorithm Category=Reinforcement Learning2019.010—
C51Algorithm Category=Reinforcement Learning2019.010—