Reinforcement Learning on Atari 57 (ALE) 200M frames sticky actions
1,047Median Human-Normalized ScoreMuZero Reanalyse
Evaluation Results
| Method | Links | |
|---|---|---|
| MuZero ReanalyseFrames=200M, Sticky actions=true2021.04 | 1,047 | |
| MuesliFrames=200M, Sticky actions=true2021.04 | 1,041 | |
| LASERFrames=200M2021.04 | 431 | |
| STACFrames=200M2021.04 | 364 | |
| Meta-gradient {γ, λ}Frames=200M2021.04 | 287 | |
| RainbowFrames=200M2021.04 | 231 | |
| IMPALAFrames=200M2021.04 | 192 | |
| DreamerV2Frames=200M2021.04 | 164 | |
| DQNFrames=200M2021.04 | 79 |