Offline Reinforcement Learning on D4RL Adroit hammer-cloned v0
6.7Normalized ScoreReBRAC
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| ReBRACevaluations=final 10, random seeds=52024.10 | 6.7 | — | — | |
| PPLevaluations=final 10, random seeds=52024.10 | 4.74 | — | — | |
| Easy BCQAlgorithmic Template=One-step2021.06 | 2.8 | — | — | |
| Fu et al.Algorithmic Template=Iterative2021.06 | 2.1 | — | — | |
| CQLevaluations=final 10, random seeds=52024.10 | 2.1 | — | — | |
| Exp. WeightAlgorithmic Template=One-step2021.06 | 1.2 | — | — | |
| IQLevaluations=final 10, random seeds=52024.10 | 1.1 | — | — | |
| BCevaluations=final 10, random seeds=52024.10 | 0.8 | — | — | |
| OTR+IQLevaluations=final 10, random seeds=52024.10 | 0.8 | — | — | |
| BCAlgorithmic Template=One-step2021.06 | 0.5 | — | — | |
| TD3+BCevaluations=final 10, random seeds=52024.10 | 0.5 | — | — | |
| Rev. KL Reg.Algorithmic Template=One-step2021.06 | 0.2 | — | — | |
| BC2021.10 | — | 0.7 | — | |
| CQLImplementation Source=Paper2021.10 | — | 5.7 | — | |
| CQLImplementation Source=Reproduced2021.10 | — | 1.4 | — | |
| EDACImplementation Source=Ours2021.10 | — | 0.3 | — | |
| Flex-f-DICE2026.02 | — | — | 1.3 | |
| Flex-f-Q2026.02 | — | — | 0.9 | |
| IQL2026.02 | — | — | 2.2 | |
| Opti-DICE2026.02 | — | — | 0.2 | |
| REM2021.10 | — | -0.3 | — | |
| SAC2021.10 | — | 0.1 | — | |
| SAC-NImplementation Source=Ours2021.10 | — | 0.2 | — |