Multi-task Reinforcement Learning on Atari-57 unclipped
10,700Median Human Normalised ScorePopArt-IMPALA
Evaluation Results
| Method | Links | |
|---|---|---|
| PopArt-IMPALAEvaluation Start Type=Random2018.09 | 10,700 | |
| PopArt-IMPALAEvaluation Start Type=Human2018.09 | 9,370 | |
| IMPALAEvaluation Start Type=Human2018.09 | 100 | |
| IMPALAEvaluation Start Type=Random2018.09 | 30 |