Multi-hop Question Answering on MuSiQue (ACCE, ACCL)
28.4ACCETC+FM*
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TC+FM*Model=Qwen-72B2026.02 | 28.4 | 36.6 | |
| TC+FM*Model=Qwen-32B2026.02 | 26 | 30.4 | |
| Search-o1Model=Qwen-72B2026.02 | 24.6 | 30.8 | |
| TC+FM*Model=Qwen-14B2026.02 | 24 | 28.2 | |
| Search-o1Model=Qwen-32B2026.02 | 22.8 | 27.8 | |
| TC+FM*Model=Llama3.1-8B2026.02 | 21.8 | 25.2 | |
| Search-o1Model=Qwen-7B2026.02 | 20.8 | 23.8 | |
| Search-o1Model=Qwen-14B2026.02 | 20.6 | 25.6 | |
| TC+FM*Model=Qwen-7B2026.02 | 20 | 23.8 | |
| Standard RAGModel=Qwen-72B2026.02 | 19 | 23.8 | |
| Standard RAGModel=Qwen-32B2026.02 | 16.2 | 19.6 | |
| Search-o1Model=Llama3.1-8B2026.02 | 15.4 | 18.2 | |
| TC+FM*Model=Qwen-3B2026.02 | 14.2 | 15.2 | |
| Standard RAGModel=Qwen-14B2026.02 | 14 | 18.6 | |
| Standard RAGModel=Qwen-7B2026.02 | 13.2 | 16.8 | |
| Naive GenModel=Qwen-72B2026.02 | 12.8 | 18.4 | |
| Naive GenModel=Llama3.1-8B2026.02 | 12.8 | 18.4 | |
| Standard RAGModel=Llama3.1-8B2026.02 | 12.2 | 15.2 | |
| Naive GenModel=Qwen-32B2026.02 | 10.8 | 15.2 | |
| Search-o1Model=Qwen-3B2026.02 | 10.4 | 12.6 | |
| Standard RAGModel=Qwen-3B2026.02 | 10.2 | 13 | |
| Naive GenModel=Qwen-14B2026.02 | 8.8 | 12.4 | |
| Naive GenModel=Qwen-7B2026.02 | 7.6 | 10.8 | |
| Naive GenModel=Qwen-3B2026.02 | 3.6 | 4.6 |