Reinforcement Learning on Atari 57 30 no-ops
591.9Mean HNSPrior. Duel Clip
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Prior. Duel ClipArchitecture=Dueling, Experience Replay=Prioritized, Gradient Clipping=true2015.11 | 591.9 | 172.1 | |
| Prior. SingleArchitecture=Single, Experience Replay=Prioritized, Gradient Clipping=true2015.11 | 434.6 | 123.7 | |
| Duel ClipArchitecture=Dueling, Experience Replay=Uniform, Gradient Clipping=true2015.11 | 373.1 | 151.5 | |
| Single ClipArchitecture=Single, Experience Replay=Uniform, Gradient Clipping=true2015.11 | 341.2 | 132.6 | |
| SingleArchitecture=Single, Experience Replay=Uniform, Gradient Clipping=false2015.11 | 307.3 | 117.8 | |
| Nature DQNArchitecture=Nature DQN2015.11 | 227.9 | 79.1 |