Multi-hop Question Answering on 2WikiMultiHopQA (EM, F1)
59.1EMSARDI
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SARDITraining Paradigm=Training-free, Backbone=DREAM-7B, Retrieval Strategy=SARDI, Diffusion Threshold (tau_c)=0.952026.06 | 59.1 | — | |
| AR W/ RET@1Training Paradigm=Training-free, Backbone=Qwen2.5-7B, Retrieval Strategy=RET@12026.06 | 58.8 | — | |
| SARDITraining Paradigm=Training-free, Backbone=DREAM-7B, Retrieval Strategy=SARDI, Diffusion Threshold (tau_c)=0.92026.06 | 57.8 | — | |
| AR W/ RET@10Training Paradigm=Training-free, Backbone=Qwen2.5-7B, Retrieval Strategy=RET@102026.06 | 53.6 | — | |
| GFM-RAG + CROSSAUGFramework=GFM-RAG, Setting=CROSSAUG2026.05 | 53.2 | 60.44 | |
| GFM-RAGFramework=GFM-RAG, Setting=Base2026.05 | 52.9 | 60.43 | |
| Search-R1Training Paradigm=RL-trained, Backbone=Qwen2.5-7B2026.06 | 52.4 | — | |
| HippoRAG2 + CROSSAUGFramework=HippoRAG2, Setting=CROSSAUG2026.05 | 51.9 | 58.15 | |
| HippoRAG2Framework=HippoRAG2, Setting=Base2026.05 | 51.2 | 57.53 | |
| LinearRAGFramework=LinearRAG, Setting=-2026.05 | 47.9 | 65.16 | |
| Critic-Embedtop k=5, Reasoner=Search-R12026.05 | 47.63 | 53.67 | |
| Critic-Embedtop k=3, Reasoner=Search-R12026.05 | 47 | 53.09 | |
| Agentic-Rtop k=5, Reasoner=Search-R12026.05 | 46.82 | 53.13 | |
| AR W/ RET@ADAPTIVETraining Paradigm=Training-free, Backbone=Qwen2.5-7B, Retrieval Strategy=ADAPTIVE2026.06 | 46.6 | — | |
| Agentic-Rtop k=3, Reasoner=Search-R12026.05 | 46.43 | 52.55 | |
| Stella-400Mtop k=5, Reasoner=Search-R12026.05 | 45.75 | 51.51 | |
| Critic-Embedtop k=1, Reasoner=Search-R12026.05 | 44.97 | 50.83 | |
| Stella-400Mtop k=3, Reasoner=Search-R12026.05 | 44.79 | 50.59 | |
| AR W/ RET@STATICTraining Paradigm=Training-free, Backbone=Qwen2.5-7B, Retrieval Strategy=STATIC2026.06 | 44.5 | — | |
| Agentic-Rtop k=1, Reasoner=Search-R12026.05 | 43.92 | 49.66 | |
| DLM W/ RET@STATICTraining Paradigm=Training-free, Backbone=DREAM-7B, Retrieval Strategy=STATIC, Diffusion Threshold (tau_c)=0.92026.06 | 43.7 | — | |
| Stella-400Mtop k=1, Reasoner=Search-R12026.05 | 43.67 | 49.46 | |
| ReActTraining Paradigm=Training-free, Backbone=Qwen2.5-7B2026.06 | 42.7 | — | |
| AdaptiveRAGTraining Paradigm=Training-free, Backbone=Qwen2.5-7B2026.06 | 34.1 | — | |
| LightRAG + CROSSAUGFramework=LightRAG, Setting=CROSSAUG2026.05 | 29.3 | 37.34 | |
| LightRAGFramework=LightRAG, Setting=Base2026.05 | 28.4 | 36.93 |