Multi-hop Question Answering on MuSiQue (Contain-Acc., GPT-Acc.)
31.1Contain-Acc.Ours
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Ours2026.06 | 31.1 | 36.2 | |
| HippoRAG22026.06 | 31 | 35 | |
| GFM-RAG2026.06 | 29.9 | 34.6 | |
| HippoRAG2026.06 | 29.3 | 24.1 | |
| LightRAG2026.06 | 27.4 | 28.6 | |
| RetrievalTop-k=52026.06 | 26.1 | 29.6 | |
| KGP2026.06 | 25.6 | 30.1 | |
| RetrievalTop-k=32026.06 | 25.1 | 27.5 | |
| E2GraphRAG2026.06 | 23.8 | 26.2 | |
| RAPTOR2026.06 | 23.3 | 27.4 | |
| RetrievalTop-k=12026.06 | 17.8 | 21.1 | |
| G-retriever2026.06 | 14.4 | 15.5 | |
| GPT-4o-miniModel=GPT-4o-mini2026.06 | 13.6 | 15.8 | |
| GPT-3.5-turboModel=GPT-3.5-turbo2026.06 | 10.3 | 21.9 | |
| llama-8BModel=llama-8B2026.06 | 7.4 | 8.1 | |
| llama-13BModel=llama-13B2026.06 | 3.3 | 4.4 |