Continuous Control on Reacher v5
-3.9Average Episodic RewardPDA
Evaluation results
| Method | Links | |
|---|---|---|
| PDAEnvironment steps=1M, Seeds=10, Tests per epoch=102026.03 | -3.9 | |
| MaxEntDPNFE=20 × 102026.04 | -4.4 | |
| SACNFE=12026.04 | -4.9 | |
| TRFP(ours)NFE=4 × 42026.04 | -4.9 | |
| TRFP(one-step)NFE=1 × 42026.04 | -5 | |
| SDACNFE=20 × 322026.04 | -6.5 | |
| TD3NFE=12026.04 | -8.9 | |
| PPOEnvironment steps=1M, Seeds=10, Tests per epoch=102026.03 | -11.1 |