Reinforcement Learning on Atari 2600 55 games (test)
1,426Mean Human-Normalized ScoreFQF
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| FQFrandom no-op starts=up to 30, seeds=32019.11 | 1,426 | 272 | 44 | 54 | |
| RAINBOWrandom no-op starts=up to 30, seeds=32019.11 | 1,213 | 227 | 42 | 52 | |
| IQNrandom no-op starts=up to 30, seeds=32019.11 | 1,112 | 218 | 39 | 54 | |
| QR-DQNrandom no-op starts=up to 30, seeds=32019.11 | 902 | 193 | 41 | 54 | |
| C51random no-op starts=up to 30, seeds=32019.11 | 701 | 178 | 40 | 50 | |
| PRIOR.random no-op starts=up to 30, seeds=32019.11 | 580 | 124 | 39 | 48 | |
| DQNrandom no-op starts=up to 30, seeds=32019.11 | 221 | 79 | 24 | 0 |