Offline Behavior Distillation on D4RL Walker2D (medium-expert)
109Normalized ReturnOffRL (Whole)
Evaluation Results
| Method | Links | |
|---|---|---|
| OffRL (Whole)Dataset=Whole offline dataset, Algorithm=Cal-QL2025.12 | 109 | |
| BC (Whole)Dataset=Whole offline dataset2025.12 | 89.6 | |
| SDWNsyn=256, tau=0.12025.12 | 44.6 | |
| Av-PBCNsyn=2562025.12 | 42.1 | |
| PBCNsyn=2562025.12 | 34 | |
| Rand (Dreal)Selection=Random, Source=Dreal, Nsyn=2562025.12 | 17.8 | |
| DBCNsyn=2562025.12 | 11.7 | |
| Rand (Doff)Selection=Random, Source=Doff, Nsyn=2562025.12 | 6.7 |