Continuous Control on OpenAI Gym Mujoco 100K steps v2 (train)
1,000InvertedPendulum-v2 ScoreSSPG
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| SSPGUTD ratio=10, Random seeds=5, Test rollouts=1002022.10 | 1,000 | 3,314 | 4,428 | 8,897 | 5,163 | 4,992 | |
| REDQUTD ratio=20, Random seeds=5, Test rollouts=1002022.10 | 1,000 | 2,994 | 1,989 | 5,613 | 3,132 | 1,402 | |
| IAPORandom seeds=5, Test rollouts=1002022.10 | 1,000 | 425 | 476 | 4,122 | 5 | 441 | |
| SAC-20UTD ratio=20, Random seeds=5, Test rollouts=1002022.10 | 1,000 | 2,718 | 2,043 | 5,831 | 496 | 495 | |
| MBPOUTD ratio=20, Random seeds=5, Test rollouts=1002022.10 | 963 | 3,271 | 3,393 | 9,533 | 1,596 | 559 |