Online Reinforcement Learning on WalkerWalk DMControl (final)
919.61Normalized ReturnGoRL(FM)
Evaluation Results
| Method | Links | |
|---|---|---|
| GoRL(FM)Seeds=5, Decoder=Flow-Matching2025.12 | 919.61 | |
| GoRL(Diff)Seeds=5, Decoder=Diffusion2025.12 | 908.96 | |
| PPOSeeds=52025.12 | 825.65 | |
| DPPOSeeds=52025.12 | 345.59 | |
| FPOSeeds=52025.12 | 29 |