Exploratory Reasoning Problems on WikiWeb-ERP 1.0 (test)
26.43AccuracyPlanRAG
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| PlanRAGexternal_retrieval=enabled2026.07 | 26.43 | 30.15 | 23.36 | 30.68 | |
| LEGO-GraphRAGArchitecture=Graph-based2026.07 | 25.48 | 28.92 | 21.27 | 28.3 | |
| DualRAGArchitecture=Iteration-based2026.07 | 24.85 | 26.65 | 20.47 | 28.35 | |
| KiRAGArchitecture=Iteration-based2026.07 | 24.72 | 27.83 | 19.65 | 27.92 | |
| ChainRAGArchitecture=Graph-based2026.07 | 24.34 | 25.56 | 20.23 | 26.14 | |
| HopRAGArchitecture=Graph-based2026.07 | 22.43 | 24.97 | 20.43 | 27.14 | |
| GenGroundArchitecture=Iteration-based2026.07 | 21.97 | 23.21 | 18.44 | 27.54 | |
| PlanRAGexternal_retrieval=absent2026.07 | 21.28 | 25.95 | 19.82 | 26.74 | |
| RetGenArchitecture=Iteration-based2026.07 | 19.34 | 23.45 | 15.78 | 24.76 | |
| NaiveRAG2026.07 | 17.54 | 20.33 | 14.65 | 22.19 | |
| DirectLLM2026.07 | 12.31 | 16.55 | 10.48 | 16.88 |