Offline Behavior Distillation on D4RL Hopper (medium-expert)
107.3Normalized ReturnOffRL (Whole)
Evaluation Results
| Method | Links | |
|---|---|---|
| OffRL (Whole)Dataset=Whole offline dataset, Algorithm=Cal-QL2025.12 | 107.3 | |
| BC (Whole)Dataset=Whole offline dataset2025.12 | 51.7 | |
| SDWNsyn=256, tau=0.12025.12 | 42.6 | |
| Av-PBCNsyn=2562025.12 | 38.7 | |
| PBCNsyn=2562025.12 | 33.4 | |
| DBCNsyn=2562025.12 | 31.1 | |
| Rand (Dreal)Selection=Random, Source=Dreal, Nsyn=2562025.12 | 27.1 | |
| Rand (Doff)Selection=Random, Source=Doff, Nsyn=2562025.12 | 11.6 |