Question Answering on FanOutQA
44.1F1 ScoreHybridDeepSearcher
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| HybridDeepSearcherSearch Strategy=Hybrid Search (Ours), Backbone=Qwen3-8B2025.08 | 44.1 | 48.4 | 20 | 3.1 | 0.15 | |
| HybridDeepSearcherSearch Strategy=Hybrid Search (Ours), Backbone=Qwen2.5-7B-Instruct2025.08 | 37.4 | 43.4 | 17.4 | 3.4 | 0.13 | |
| RAG-R1Search Strategy=Parallel Search, Backbone=Qwen2.5-7B-Instruct2025.08 | 28.2 | 36.7 | 10 | 1.9 | 0.09 | |
| Search-o1Search Strategy=Sequential Search, Backbone=Qwen3-8B2025.08 | 26.7 | 32.9 | 8.7 | 5.2 | 0.06 | |
| DeepResearcherSearch Strategy=Parallel Search, Backbone=Qwen2.5-7B-Instruct2025.08 | 26.4 | 35.4 | 6.45 | 3.6 | 0.05 | |
| Standard RAGSearch Strategy=Non-iterative, Backbone=Qwen3-8B2025.08 | 20.6 | 32.1 | 5.6 | — | — | |
| R1-SearcherSearch Strategy=Sequential Search, Backbone=Qwen2.5-7B-Instruct2025.08 | 18.8 | 30.2 | 2.5 | 3.1 | 0.02 | |
| Naïve GenSearch Strategy=Non-iterative, Backbone=Qwen3-8B2025.08 | 10.9 | 27.5 | 3.2 | — | — | |
| Search-R1Search Strategy=Sequential Search, Backbone=Qwen2.5-7B-Instruct2025.08 | 10.1 | 23.1 | 1.2 | 4.3 | 0.01 |