Multi-hop Question Answering on 2WikiMultiHopQA (RAG/CoT Evaluation)
65.78AccuracyRIRS
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| RIRSModel=Qwen-Plus2025.01 | 65.78 | 78.98 | 74.68 | 31.88 | |
| RIRSModel=Mixed2025.01 | 62.06 | 76.93 | 69.65 | 36.74 | |
| ChameleonModel=Qwen-Plus2025.01 | 60.12 | 68.92 | 72.86 | 31.87 | |
| RouterDCModel=Qwen-Plus2025.01 | 52.12 | 62.2 | 62.47 | 35.63 | |
| RIRSModel=LLaMA-3.1-8B2025.01 | 45.65 | 72.85 | 51.62 | 29.6 | |
| ChameleonModel=LLaMA-3.1-8B2025.01 | 40.27 | 67.81 | 45.89 | 28.41 | |
| CoTModel=Qwen-Plus2025.01 | 37.31 | — | — | 37.31 | |
| RouterDCModel=LLaMA-3.1-8B2025.01 | 30.59 | 68.33 | 34.25 | 22.69 | |
| CoTModel=LLaMA-3.1-8B2025.01 | 25.99 | — | — | 25.99 |