Multi-hop Question Answering on PopQA fixed split (test)
50.78Exact Match (EM)Graph-R1
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Graph-R1Backbone=Qwen2.5-7B-Instruct, Interaction Type=Training, Knowledge Representation=Graph-based2026.05 | 50.78 | 54.01 | |
| Graph-AdaptR1Backbone=Qwen2.5-7B-Instruct, Interaction Type=Training, Knowledge Representation=Graph-based2026.05 | 49.22 | 55.03 | |
| Search-R1Backbone=Qwen2.5-7B-Instruct, Interaction Type=Training, Knowledge Representation=Chunk-based2026.05 | 43.75 | 47.03 | |
| Search-AdaptR1Backbone=Qwen2.5-7B-Instruct, Interaction Type=Training, Knowledge Representation=Chunk-based2026.05 | 42.19 | 46.8 | |
| R1-SearcherBackbone=Qwen2.5-7B-Instruct, Interaction Type=Training, Knowledge Representation=Chunk-based2026.05 | 41.41 | 47.12 | |
| R1Backbone=Qwen2.5-7B-Instruct, Interaction Type=Training, Knowledge Representation=None2026.05 | 26.56 | 30.35 | |
| SFTBackbone=Qwen2.5-7B-Instruct, Interaction Type=Training, Knowledge Representation=Chunk-based2026.05 | 20.31 | 27.93 | |
| StandardRAGBackbone=GPT-4o-mini, Interaction Type=Prompt Engineering, Knowledge Representation=Chunk-based2026.05 | 18.75 | 30.58 | |
| HyperGraphRAGBackbone=GPT-4o-mini, Interaction Type=Prompt Engineering, Knowledge Representation=Graph-based2026.05 | 13.28 | 29.48 | |
| NaiveGenerationBackbone=GPT-4o-mini, Interaction Type=Prompt Engineering, Knowledge Representation=None2026.05 | 10.36 | 25.95 | |
| HippoRAG2Backbone=GPT-4o-mini, Interaction Type=Prompt Engineering, Knowledge Representation=Graph-based2026.05 | 9.38 | 21.1 | |
| GraphRAGBackbone=GPT-4o-mini, Interaction Type=Prompt Engineering, Knowledge Representation=Graph-based2026.05 | 8.59 | 20.92 | |
| LightRAGBackbone=GPT-4o-mini, Interaction Type=Prompt Engineering, Knowledge Representation=Graph-based2026.05 | 5.47 | 24.47 | |
| StandardRAGBackbone=Qwen2.5-7B-Instruct, Interaction Type=Prompt Engineering, Knowledge Representation=Chunk-based2026.05 | 3.12 | 13.1 | |
| PathRAGBackbone=GPT-4o-mini, Interaction Type=Prompt Engineering, Knowledge Representation=Graph-based2026.05 | 2.34 | 15.65 | |
| NaiveGenerationBackbone=Qwen2.5-7B-Instruct, Interaction Type=Prompt Engineering, Knowledge Representation=None2026.05 | 0.78 | 12.82 |