Reinforcement Learning on Atari 2600 (20 Game Subset)
4.4Freeway ScoreSAC
Evaluation results
| Method | Links | ||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SACrandom seeds=5, metric aggregation=mean2019.10 | 4.4 | 690.9 | 0.8 | 4,386.7 | 280.5 | 160.8 | 432.1 | 350 | 68.3 | 211.6 | 272 | 29.3 | 216.9 | 305.3 | 59.4 | 7.9 | 3,668.7 | 0.7 | 250.7 | -20.98 | |
| RainbowSource=Kaiser et al., 20192019.10 | 0.1 | 364.3 | 0.53 | 3,363.5 | 235.6 | 135.1 | 365.6 | 300.3 | 61.7 | 206.3 | 285.7 | 38.7 | 290.6 | 524.1 | 140.1 | 20.8 | 12,558.3 | 3.3 | 1,346.3 | -19.5 | |
| Randompolicy=purely random2019.10 | 0 | 235.2 | 0 | 2,895 | 166.1 | 148 | 372.1 | 233.7 | 29.2 | 61.1 | 248.8 | 42 | 184.8 | 0 | 74 | 11.8 | 7,339.5 | 0.9 | 488.4 | -20.4 |