Generalized Planning on IPC Floortile 2023 (test)
37.8CoverageR-GNN (Ω-Heu)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| R-GNN (Ω-Heu)Architecture=R-GNN, Policy Formulation=Heuristic regularized (Ω-Heu), Time Limit=1 minute, Memory Limit=8 GB2026.03 | 37.8 | 109.3 | |
| R-GNN (Ω-Exp)Architecture=R-GNN, Policy Formulation=Explicit regularized (Ω-Exp), Time Limit=1 minute, Memory Limit=8 GB2026.03 | 36.7 | 93.5 | |
| Horčík2026.05 | 32 | — | |
| GSPWA⋆Search Strategy=WA⋆ (w = 2) heuristic, Search Budget=10,000 expansions or 1 hour2026.05 | 28 | 57 | |
| GSPπSearch Strategy=greedy policy, Search Budget=10,000 expansions or 1 hour2026.05 | 20 | 54 | |
| LAMA-firstTime Limit=1 minute, Memory Limit=8 GB2026.03 | 16.9 | 51 | |
| LAMA2026.05 | 13 | 47 | |
| WL-f2026.05 | 3 | — | |
| Distincter2026.05 | 2 | — | |
| R-GNN (V)Architecture=R-GNN, Policy Formulation=State-value (V), Time Limit=1 minute, Memory Limit=8 GB2026.03 | 0 | — | |
| Lifted HER2026.05 | 0 | — | |
| α0Search Strategy=AlphaZero-style learning2026.05 | 0 | — |