Offline Reinforcement Learning on D4RL AntMaze fixed, play, diverse
85AntMaze UMaze (Fixed) ScoreICQ
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| ICQrandom_seeds=32021.06 | 85 | 80 | 51 | 65 | 65 | 44 | |
| BCCategory=Baseline2021.10 | 82 | 0 | 0 | 47 | 0 | 0 | |
| COMBOCategory=Model-Based2021.10 | 80.3 | — | — | 57.3 | — | — | |
| BCQrandom_seeds=32021.06 | 78.9 | 0 | 6.7 | 55 | 0 | 2.2 | |
| BCQCategory=Model-Free2021.10 | 75.3 | — | — | — | — | — | |
| CQLrandom_seeds=32021.06 | 74 | 61.2 | 15.8 | 84 | 53.7 | 14.9 | |
| ROMI-BCQCategory=Proposed2021.10 | 68.7 | 35.3 | 20.2 | 61.2 | 27.3 | 41.2 | |
| BCrandom_seeds=32021.06 | 65 | 0 | 0 | 55 | 0 | 0 | |
| AWRrandom_seeds=32021.06 | 56 | 0 | 0 | 70.3 | 0 | 0 | |
| BRAC-prandom_seeds=32021.06 | 50 | 0 | 0 | 40 | 0 | 0 | |
| BAILCategory=Model-Free2021.10 | — | 1.7 | 2.2 | 54 | — | 1 | |
| CQLCategory=Model-Free2021.10 | — | — | — | — | 61.5 | — |