Graph-structured planning on Graph-structured planning pooled over four benchmarks (test)
83.8Success Rate (SR)GILP
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| GILP2026.06 | 83.8 | 75.8 | 6.5 | 7.9 | 1.51 | 15,021 | 6.8 | 0 | |
| GILP-VerifierStrategy=Verifier2026.06 | 82.5 | 71.7 | 4 | 7.5 | 1.516 | 21,261 | 6.3 | 0 | |
| Hybrid-Full-VerifierComponents=Full Hybrid, Strategy=Verifier2026.06 | 76.6 | 66.1 | 5.1 | 10.7 | 1.621 | 21,101 | 9.7 | 0 | |
| Hybrid-FullComponents=Full Hybrid2026.06 | 75 | 63.6 | 9.1 | 11.1 | 1.653 | 13,503 | 12.1 | 0 | |
| Hybrid-DeltaOnlyComponents=Delta Only2026.06 | 70.3 | 53.8 | 12.3 | 12.1 | 1.787 | 17,167 | 16.4 | 0 | |
| Agent-VerifierStrategy=Verifier2026.06 | 68.4 | 49.5 | 6.2 | 19.3 | 2.35 | 26,038 | 18.6 | 0.3 | |
| Agent-ReplanStrategy=Replan2026.06 | 66.8 | 47.1 | 16.9 | 20.5 | 2.451 | 16,389 | 21.8 | 10.5 | |
| Hybrid-ValidityOnlyComponents=Validity Only2026.06 | 66 | 47.1 | 8.7 | 19.2 | 2.451 | 18,302 | 21.5 | 5 | |
| Parametric-WM-MCTSModel=Parametric World Model, Planner=MCTS2026.06 | 62.5 | 55.9 | 8 | 6 | 1.1 | 262 | 11.7 | 0 | |
| Agent-CoTReasoning=Chain-of-Thought2026.06 | 59.1 | 39.4 | 20.8 | 22.5 | 2.447 | 21,301 | 28.6 | 20 | |
| Parametric-WM-MPCModel=Parametric World Model, Planner=MPC2026.06 | 56.5 | 50.2 | 8.9 | 6.9 | 1.096 | 291 | 14.5 | 0 | |
| Agent-DirectReasoning=Direct2026.06 | 49.6 | 26.5 | 25.7 | 25.2 | 2.48 | 9,961 | 35.8 | 27.6 |