Offline Reinforcement Learning on D4RL Adroit hammer-cloned
2,280Normalized ScoreEPQ
Evaluation Results
| Method | Links | |
|---|---|---|
| EPQPolicy Type=Gaussian, Seeds=82026.02 | 2,280 | |
| NEUBAYModel paradigm=Ours, pi_D=0.7, Number of seeds=6, Evaluation episodes=20, Evaluation step=final step2025.12 | 1,440 | |
| FACPolicy Type=Flow, Seeds=82026.02 | 1,110 | |
| FQLPolicy Type=Flow, Seeds=82026.02 | 1,100 | |
| ReBRACPolicy Type=Gaussian, Seeds=82026.02 | 670 | |
| ReBRACModel paradigm=Model-free, pi_D=0.7, Number of seeds=6, Evaluation episodes=20, Evaluation step=final step2025.12 | 500 | |
| VIPOModel paradigm=Conservative model-based, pi_D=0.7, Number of seeds=6, Evaluation episodes=20, Evaluation step=final step2025.12 | 210 | |
| CQLPolicy Type=Gaussian, Seeds=82026.02 | 210 | |
| IDQLPolicy Type=Diffusion, Seeds=82026.02 | 200 | |
| SRPOPolicy Type=Diffusion, Seeds=82026.02 | 200 | |
| IQLModel paradigm=Model-free, pi_D=0.7, Number of seeds=6, Evaluation episodes=20, Evaluation step=final step2025.12 | 150 | |
| MOBILEModel paradigm=Conservative model-based, pi_D=0.7, Number of seeds=6, Evaluation episodes=20, Evaluation step=final step2025.12 | 150 | |
| MCQPolicy Type=Gaussian, Seeds=82026.02 | 140 | |
| IQLPolicy Type=Gaussian, Seeds=82026.02 | 110 | |
| CACPolicy Type=Diffusion, Seeds=82026.02 | 100 | |
| TD3+BCPolicy Type=Gaussian, Seeds=82026.02 | 80 | |
| ARMORModel paradigm=Conservative model-based, pi_D=0.7, Number of seeds=6, Evaluation episodes=20, Evaluation step=final step2025.12 | 70 | |
| MoMoModel paradigm=Conservative model-based, pi_D=0.7, Number of seeds=6, Evaluation episodes=20, Evaluation step=final step2025.12 | 70 | |
| BCModel paradigm=Model-free, pi_D=0.7, Number of seeds=6, Evaluation episodes=20, Evaluation step=final step2025.12 | 60 | |
| MOPOModel paradigm=Conservative model-based, pi_D=0.7, Number of seeds=6, Evaluation episodes=20, Evaluation step=final step2025.12 | 50 | |
| EDACModel paradigm=Model-free, pi_D=0.7, Number of seeds=6, Evaluation episodes=20, Evaluation step=final step2025.12 | 30 | |
| IQLDomain=Adroit, Hyperparameter Tuning=individually tuned2025.11 | 11.6 | |
| MXQLDomain=Adroit, Hyperparameter Tuning=individually tuned2025.11 | 11.1 | |
| SAC-RNDPolicy Type=Gaussian, Seeds=82026.02 | 10 | |
| BCDomain=Adroit, Hyperparameter Tuning=individually tuned2025.11 | 8.9 | |
| QQLDomain=Adroit, Hyperparameter Tuning=consistent2025.11 | 8 | |
| XQLDomain=Adroit, Hyperparameter Tuning=individually tuned2025.11 | 1.1 | |
| TD3+BCDomain=Adroit, Hyperparameter Tuning=individually tuned2025.11 | 0.9 | |
| CQLDomain=Adroit, Hyperparameter Tuning=individually tuned2025.11 | 0.3 |