Reinforcement Learning on Cheetah-Gravity Start OpenAI Gym Mujoco (initialization)
160.88Average RewardVBLRL
Evaluation results
| Method | Links | |
|---|---|---|
| VBLRLsamples per iteration=1x2022.10 | 160.88 | |
| T-HiP-MDPsamples per iteration=1x2022.10 | 126.95 | |
| LPG-FTWsamples per iteration=2x2022.10 | -81.59 | |
| Single-task MBRLmode=train from scratch2022.10 | -83.96 | |
| EWCsamples per iteration=2x2022.10 | -3,426.79 |