Generalized Planning on Blocksworld
55ScaleHeuristically regularized Q-value policy (Ω^Heu.)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Heuristically regularized Q-value policy (Ω^Heu.)Architecture=R-GNN2026.03 | 55 | 48.2 | |
| Explicitly regularized Q-value policy (Ω^Exp.)Architecture=R-GNN2026.03 | 54 | 47.8 | |
| State-value policy (V)Architecture=R-GNN2026.03 | 52 | 44.5 | |
| Explicitly regularized Q-value policy (Ω^Exp.)Architecture=OAE2026.03 | 52 | 44.9 | |
| Heuristically regularized Q-value policy (Ω^Heu.)Architecture=OAE2026.03 | 52 | 44.6 | |
| Heuristically regularized Q-value policy (Ω^Heu.)Architecture=OE2026.03 | 50 | 42.4 | |
| Explicitly regularized Q-value policy (Ω^Exp.)Architecture=OE2026.03 | 48 | 40.5 | |
| State-value policy (V)Architecture=OAE2026.03 | 37 | 27.5 | |
| State-value policy (V)Architecture=OE2026.03 | 14 | 9.2 | |
| Vanilla Q-value policy (Q)Architecture=R-GNN2026.03 | 5 | 1.6 | |
| Vanilla Q-value policy (Q)Architecture=OE2026.03 | 5 | 1.2 | |
| Vanilla Q-value policy (Q)Architecture=OAE2026.03 | 4 | 1.3 |