Continuous Control on Visual DMC Suite
324Acrobot-Swingup ScoreDR.Q
Evaluation Results
| Method | Links | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| DR.QEnvironment steps=1M, Action repeat=22026.05 | 324 | 118 | 700 | 113 | 201 | 330 | 937 | 1 | 655 | 927 | 954 | 746 | 49.4 | 50 | 50.1 | |
| MR.QEnvironment steps=1M, Action repeat=22026.05 | 287 | 60 | 216 | 65 | 77 | 270 | 852 | 1 | 498 | 833 | 965 | 615 | 32.2 | 39.8 | 39.5 | |
| TDMPC2Environment steps=1M, Action repeat=22026.05 | 197 | 14 | 117 | 20 | 22 | 187 | 582 | 0 | 262 | 246 | 911 | 665 | 15.4 | 29.5 | 26.9 | |
| DrQ-v2Environment steps=1M, Action repeat=22026.05 | 168 | 10 | 43 | 14 | 22 | 224 | 917 | 1 | 459 | 750 | 705 | 546 | 24.1 | 19.1 | 32.1 | |
| DreamerV3Environment steps=1M, Action repeat=22026.05 | 121 | 9 | 61 | 14 | 11 | 205 | 888 | 1 | 328 | 316 | 338 | 669 | 16.8 | 13.4 | 24.7 | |
| PPOEnvironment steps=1M, Action repeat=22026.05 | 2 | 11 | 51 | 13 | 16 | 0 | 1 | 1 | 118 | 149 | 10 | 39 | 1.6 | 1.3 | 3.4 |