Simple Question Answering on NQ* (test)
40.09RAG F1HippoRAG + DeepRefine-8B
Evaluation Results
| Method | Links | |
|---|---|---|
| HippoRAG + DeepRefine-8BConstructor=AR1, Refinement Model=DeepRefine-8B2026.05 | 40.09 | |
| HippoRAGConstructor=AR12026.05 | 39.84 | |
| HippoRAG + DeepRefine-4BConstructor=AR1, Refinement Model=DeepRefine-4B2026.05 | 39.75 | |
| HippoRAG2 + DeepRefine-8BConstructor=AR1, Refinement Model=DeepRefine-8B2026.05 | 38.6 | |
| HippoRAG2 + DeepRefine-4BConstructor=AR1, Refinement Model=DeepRefine-4B2026.05 | 37.71 | |
| HippoRAG2Constructor=AR12026.05 | 37.25 | |
| HippoRAG2 + DeepRefine-8BConstructor=Naive, Refinement Model=DeepRefine-8B2026.05 | 36.03 | |
| HippoRAG2 + DeepRefine-4BConstructor=Naive, Refinement Model=DeepRefine-4B2026.05 | 35.93 | |
| HippoRAG2Constructor=Naive2026.05 | 35.91 | |
| HippoRAG + DeepRefine-4BConstructor=Naive, Refinement Model=DeepRefine-4B2026.05 | 35.85 | |
| HippoRAG + DeepRefine-8BConstructor=Naive, Refinement Model=DeepRefine-8B2026.05 | 35.69 | |
| HippoRAGConstructor=Naive2026.05 | 35.54 | |
| ToG + DeepRefine-8BConstructor=AR1, Refinement Model=DeepRefine-8B2026.05 | 29.91 | |
| ToGConstructor=AR12026.05 | 29.88 | |
| ToG + DeepRefine-4BConstructor=AR1, Refinement Model=DeepRefine-4B2026.05 | 29.47 | |
| Subgraph + DeepRefine-8BConstructor=AR1, Refinement Model=DeepRefine-8B2026.05 | 29.41 | |
| SubgraphConstructor=AR12026.05 | 29.27 | |
| Subgraph + DeepRefine-4BConstructor=AR1, Refinement Model=DeepRefine-4B2026.05 | 28.1 | |
| Subgraph + DeepRefine-4BConstructor=Naive, Refinement Model=DeepRefine-4B2026.05 | 27.5 | |
| Subgraph + DeepRefine-8BConstructor=Naive, Refinement Model=DeepRefine-8B2026.05 | 26.65 | |
| SubgraphConstructor=Naive2026.05 | 26.43 | |
| ToG + DeepRefine-8BConstructor=Naive, Refinement Model=DeepRefine-8B2026.05 | 25.72 | |
| ToG + DeepRefine-4BConstructor=Naive, Refinement Model=DeepRefine-4B2026.05 | 25.6 | |
| ToGConstructor=Naive2026.05 | 25.55 | |
| Subgraph + DeepRefine-8BConstructor=Graphify (LLM-Wiki), Refinement Model=DeepRefine-8B2026.05 | 21.26 | |
| Subgraph + DeepRefine-4BConstructor=Graphify (LLM-Wiki), Refinement Model=DeepRefine-4B2026.05 | 20.97 | |
| SubgraphConstructor=Graphify (LLM-Wiki)2026.05 | 20.42 | |
| ToG + DeepRefine-8BConstructor=Graphify (LLM-Wiki), Refinement Model=DeepRefine-8B2026.05 | 18.22 | |
| ToG + DeepRefine-4BConstructor=Graphify (LLM-Wiki), Refinement Model=DeepRefine-4B2026.05 | 17.9 | |
| ToGConstructor=Graphify (LLM-Wiki)2026.05 | 16.69 | |
| HippoRAG2 + DeepRefine-4BConstructor=Graphify (LLM-Wiki), Refinement Model=DeepRefine-4B2026.05 | 13.9 | |
| HippoRAG2 + DeepRefine-8BConstructor=Graphify (LLM-Wiki), Refinement Model=DeepRefine-8B2026.05 | 13.77 | |
| HippoRAG2Constructor=Graphify (LLM-Wiki)2026.05 | 13.62 | |
| HippoRAG + DeepRefine-8BConstructor=Graphify (LLM-Wiki), Refinement Model=DeepRefine-8B2026.05 | 13.46 | |
| HippoRAGConstructor=Graphify (LLM-Wiki)2026.05 | 13.23 | |
| HippoRAG + DeepRefine-4BConstructor=Graphify (LLM-Wiki), Refinement Model=DeepRefine-4B2026.05 | 12.9 |