Multi-Hop QA on Musique (EM)
18.5Exact Match (EM)ReSeek
Evaluation Results
| Method | Links | |
|---|---|---|
| ReSeekBackbone=Qwen2.5-7b-Instruct2025.10 | 18.5 | |
| ZeroSearchBackbone=Qwen2.5-7b-Instruct2025.10 | 18.4 | |
| Search-R1Backbone=Qwen2.5-7b-Instruct2025.10 | 14.6 | |
| Search-R1Backbone=Qwen2.5-3b-Instruct2025.10 | 10.3 | |
| ReSeekBackbone=Qwen2.5-3b-Instruct2025.10 | 10.3 | |
| ZeroSearchBackbone=Qwen2.5-3b-Instruct2025.10 | 9.8 | |
| R1Backbone=Qwen2.5-7b-Instruct2025.10 | 7.2 | |
| R1Backbone=Qwen2.5-3b-Instruct2025.10 | 6 | |
| RAGBackbone=Qwen2.5-7b-Instruct2025.10 | 5.8 | |
| Search-o1Backbone=Qwen2.5-7b-Instruct2025.10 | 5.8 | |
| Search-o1Backbone=Qwen2.5-3b-Instruct2025.10 | 5.4 | |
| RAGBackbone=Qwen2.5-3b-Instruct2025.10 | 4.7 | |
| Direct InferenceBackbone=Qwen2.5-7b-Instruct2025.10 | 3.1 | |
| CoTBackbone=Qwen2.5-7b-Instruct2025.10 | 2.2 | |
| Direct InferenceBackbone=Qwen2.5-3b-Instruct2025.10 | 2 | |
| CoTBackbone=Qwen2.5-3b-Instruct2025.10 | 0.2 |