Reinforcement Learning on ALE (10 Game Score Subset)
12,301.1Asterix ScoreQDagger
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| QDaggerNumber of runs=32022.06 | 12,301.1 | 6,428.3 | 34.1 | 277.6 | 1,765.6 | 5,110.6 | 16,198.8 | 19,496.7 | 17,056 | 6,891.9 | |
| KickstartingNumber of runs=32022.06 | 7,872 | 5,330.9 | 31.3 | 237.6 | 1,167.3 | 4,196.8 | 10,028.4 | 16,160.5 | 7,849.9 | 2,459.6 | |
| Offline PretrainingNumber of runs=32022.06 | 4,132.2 | 1,854.5 | 25.5 | 126.3 | 1,167.9 | 3,687.2 | 16,421.6 | 14,137.7 | 12,099.4 | 958.7 |