Multi-task Reinforcement Learning on Meta-World MT50 Conditioned v1
61Average Success RateSoft Modularization (Deep)
Evaluation Results
| Method | Links | |
|---|---|---|
| Soft Modularization (Deep)Network Depth=Deep2020.03 | 61 | |
| Soft Modularization (Shallow)Network Depth=Shallow2020.03 | 60.4 | |
| Mix-ExpertImplementation=Internal implementation baseline2020.03 | 37.5 | |
| MT-MH-SACImplementation=Internal implementation baseline2020.03 | 34.2 | |
| Hard RoutingImplementation=Internal implementation baseline2020.03 | 29.1 | |
| MT-SACImplementation=Internal implementation baseline2020.03 | 28.3 |