Offline Reinforcement Learning on Walker2d (Score Reduction Analysis)
3,718Clean ScoreGlobal Budget Allocation
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Global Budget AllocationVictim=CQL, Config (rho, epsilon)=(0.01, 0.5)2025.12 | 3,718 | — | — | — | 681 | 81.7 | |
| Local GreedyVictim=CQL, Config (rho, epsilon)=(0.01, 0.5)2025.12 | 3,718 | — | — | 929 | — | — | |
| Random SubsetVictim=CQL, Config (rho, epsilon)=(0.01, 0.5)2025.12 | 3,718 | — | 2,129 | — | — | — | |
| Random NoiseVictim=CQL, Config (rho, epsilon)=(0.01, 0.5)2025.12 | 3,718 | 3,152 | — | — | — | — | |
| Global Budget AllocationVictim=CQL, Config (rho, epsilon)=(0.015, 0.33)2025.12 | 3,718 | — | — | — | 924 | 75.1 | |
| Local GreedyVictim=CQL, Config (rho, epsilon)=(0.015, 0.33)2025.12 | 3,718 | — | — | 859 | — | — | |
| Global Budget AllocationVictim=CQL, Config (rho, epsilon)=(0.02, 0.25)2025.12 | 3,718 | — | — | — | 795 | 78.6 | |
| Global Budget AllocationVictim=CQL, Config (rho, epsilon)=(0.025, 0.2)2025.12 | 3,718 | — | — | — | 1,124 | 69.8 | |
| Global Budget AllocationVictim=CQL, Config (rho, epsilon)=(0.033, 0.15)2025.12 | 3,718 | — | — | — | 1,491 | 59.9 | |
| Global Budget AllocationVictim=CQL, Config (rho, epsilon)=(0.05, 0.1)2025.12 | 3,718 | — | — | — | 2,188 | 41.1 | |
| Global Budget AllocationVictim=CQL, Config (rho, epsilon)=(0.1, 0.05)2025.12 | 3,718 | — | — | — | 2,693 | 27.6 | |
| Global Budget AllocationVictim=BCQ, Config (rho, epsilon)=(0.01, 0.5)2025.12 | 2,984 | — | — | — | 522 | 82.5 | |
| Local GreedyVictim=BCQ, Config (rho, epsilon)=(0.01, 0.5)2025.12 | 2,984 | — | — | 652 | — | — | |
| Global Budget AllocationVictim=BCQ, Config (rho, epsilon)=(0.05, 0.1)2025.12 | 2,984 | — | — | — | 1,956 | 34.5 | |
| Global Budget AllocationVictim=BEAR, Config (rho, epsilon)=(0.01, 0.5)2025.12 | 2,763 | — | — | — | 411 | 85.1 | |
| Local GreedyVictim=BEAR, Config (rho, epsilon)=(0.01, 0.5)2025.12 | 2,763 | — | — | 565 | — | — | |
| Global Budget AllocationVictim=BEAR, Config (rho, epsilon)=(0.1, 0.05)2025.12 | 2,763 | — | — | — | 2,216 | 19.8 | |
| Global Budget AllocationVictim=IQL, Config (rho, epsilon)=(0.01, 0.5)2025.12 | 1,779 | — | — | — | 1,197 | 32.7 | |
| Local GreedyVictim=IQL, Config (rho, epsilon)=(0.01, 0.5)2025.12 | 1,779 | — | — | 1,291 | — | — | |
| Global Budget AllocationVictim=IQL, Config (rho, epsilon)=(0.05, 0.1)2025.12 | 1,779 | — | — | — | 1,418 | 20.3 | |
| Global Budget AllocationVictim=IQL, Config (rho, epsilon)=(0.1, 0.05)2025.12 | 1,779 | — | — | — | 1,527 | 14.2 |