Multi-hop Question Answering on 2WikiMultiHopQA v1.0 (test)
2.29Task Latency (s)No Retrieval
Evaluation Results
| Method | Links | |
|---|---|---|
| No RetrievalBackbone=Llama3.1-8B2025.03 | 2.29 | |
| VanillaRAGBackbone=Llama3.1-8B, k=42025.03 | 2.6 | |
| L-RAGBackbone=Llama3.1-8B, k=42025.03 | 2.76 | |
| VanillaRAGBackbone=Llama3.1-8B, k=82025.03 | 2.79 | |
| L-RAGBackbone=Llama3.1-8B, k=82025.03 | 2.87 | |
| HyDEBackbone=Llama3.1-8B2025.03 | 5.6 | |
| SelfAskBackbone=Llama3.1-8B2025.03 | 14.8 | |
| IterRetGenBackbone=Llama3.1-8B2025.03 | 20.07 | |
| IRCoTBackbone=Llama3.1-8B2025.03 | 25.47 |