Multi-hop Question Answering on HotpotQA (RAG and CoT Evaluation)
75.63AccuracyRIRS
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| RIRSModel=Qwen-Plus2025.01 | 75.63 | 82.85 | 81.86 | 45.37 | |
| ChameleonModel=Qwen-Plus2025.01 | 73.19 | 84.1 | 78.7 | 44.56 | |
| RIRSModel=Mixed2025.01 | 69.33 | 80.27 | 73.17 | 53.7 | |
| RouterDCModel=Qwen-Plus2025.01 | 67.24 | 72.26 | 75.64 | 45.32 | |
| RIRSModel=LLaMA-3.1-8B2025.01 | 58.69 | 77.33 | 64.42 | 39.23 | |
| CoTModel=Qwen-Plus2025.01 | 54.58 | — | — | 54.58 | |
| ChameleonModel=LLaMA-3.1-8B2025.01 | 50.2 | 77.75 | 55.81 | 30.56 | |
| RouterDCModel=LLaMA-3.1-8B2025.01 | 47.74 | 74.51 | 50.92 | 38.51 | |
| CoTModel=LLaMA-3.1-8B2025.01 | 44.23 | — | — | 44.23 |