Continuous Control on HalfCheetah v1 (train)
9,721Max Average ReturnMB-MPO + Min Attn
Evaluation results
| Method | Links | |
|---|---|---|
| MB-MPO + Min AttnRegularization=Minimum Attention2025.05 | 9,721 | |
| TD3Time Steps=1 million2018.02 | 9,636.95 | |
| Our DDPGTime Steps=1 million, Implementation=Modified architecture2018.02 | 8,577.29 | |
| MB-MPORegularization=None2025.05 | 6,692 | |
| DDPGTime Steps=1 million2018.02 | 3,305.6 | |
| SACTime Steps=1 million2018.02 | 2,347.19 | |
| PPOTime Steps=1 million2018.02 | 1,795.43 | |
| ACKTRTime Steps=1 million2018.02 | 1,450.46 | |
| TRPOTime Steps=1 million2018.02 | -15.57 |