Atari Games On Atari Games
평가 지표
Mean Human Normalized Score
평가 결과
이 벤치마크에서 각 모델의 성능 결과
모델 이름 | Mean Human Normalized Score | Paper Title | Repository |
|---|---|---|---|
| SimPLe | 25.3% | Model-Based Reinforcement Learning for Atari | |
| NGU | 3169.90% | Never Give Up: Learning Directed Exploration Strategies | |
| Go-Explore | 4989.94% | First return, then explore | |
| IMPALA, deep | 957.34% | IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures | |
| Agent57 | 4763.69% | Agent57: Outperforming the Atari Human Benchmark | |
| LASER | 1741.36% | Off-Policy Actor-Critic with Shared Experience Replay | - |
| GDI-H3 | 9620.33% | Generalized Data Distribution Iteration | - |
| R2D2 | 3374.31% | Recurrent Experience Replay in Distributed Reinforcement Learning | - |
| Rainbow DQN | 873.97% | Rainbow: Combining Improvements in Deep Reinforcement Learning | |
| MuZero | 4996.20% | Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model | |
| GDI-I3 | 7810.1% | Generalized Data Distribution Iteration | - |
| DreamerV2 | 631.17% | Mastering Atari with Discrete World Models |
0 of 12 row(s) selected.