Multi-Hop Question Answering on Bamboogle (Accuracy)
47.6AccuracyMultiSearch
Evaluation Results
| Method | Links | |
|---|---|---|
| MultiSearchBackbone=Qwen2.5-7B, Model Variant=Base2026.05 | 47.6 | |
| CriticSearchBackbone=Qwen2.5-7B2026.05 | 47.2 | |
| StepSearchBackbone=Qwen2.5-7B, Model Variant=Base2026.05 | 46.7 | |
| MultiSearchBackbone=Qwen2.5-7B, Model Variant=Instruct2026.05 | 46.3 | |
| AutoRefineBackbone=Qwen2.5-7B, Model Variant=Base2026.05 | 43.5 | |
| AutoRefineBackbone=Qwen2.5-7B, Model Variant=Instruct2026.05 | 41.6 | |
| Search-R1Backbone=Qwen2.5-7B, Model Variant=Instruct2026.05 | 40 | |
| MultiSearchBackbone=Qwen2.5-3B, Model Variant=Base2026.05 | 39 | |
| MultiSearchBackbone=Qwen2.5-3B, Model Variant=Instruct2026.05 | 37.1 | |
| CriticSearchBackbone=Qwen2.5-3B2026.05 | 36.8 | |
| ReasonRAGBackbone=Qwen2.5-7B2026.05 | 36 | |
| Dr.ZeroBackbone=Qwen2.5-7B2026.05 | 36 | |
| Search-R1Backbone=Qwen2.5-7B, Model Variant=Base2026.05 | 36 | |
| StepSearchBackbone=Qwen2.5-3B, Model Variant=Instruct2026.05 | 34.4 | |
| AutoRefineBackbone=Qwen2.5-3B, Model Variant=Base2026.05 | 34.4 | |
| AutoRefineBackbone=Qwen2.5-3B, Model Variant=Instruct2026.05 | 33.6 | |
| StepSearchBackbone=Qwen2.5-3B, Model Variant=Base2026.05 | 32.8 | |
| Search-o1Backbone=Qwen2.5-3B2026.05 | 32 | |
| StepSearchBackbone=Qwen2.5-7B, Model Variant=Instruct2026.05 | 31.2 | |
| R1Backbone=Qwen2.5-7B, Model Variant=Base2026.05 | 29.6 | |
| Search-o1Backbone=Qwen2.5-7B2026.05 | 29.6 | |
| R1Backbone=Qwen2.5-7B, Model Variant=Instruct2026.05 | 29.3 | |
| AdaSearchBackbone=Qwen2.5-3B2026.05 | 28 | |
| ReSearchBackbone=Qwen2.5-3B, Model Variant=Instruct2026.05 | 26.6 | |
| IRCoTBackbone=Qwen2.5-3B2026.05 | 24 | |
| Search-R1Backbone=Qwen2.5-3B, Model Variant=Instruct2026.05 | 23.2 | |
| CoTBackbone=Qwen2.5-7B2026.05 | 23.2 | |
| R1Backbone=Qwen2.5-3B, Model Variant=Base2026.05 | 22.4 | |
| IRCoTBackbone=Qwen2.5-7B2026.05 | 22.4 | |
| RAGBackbone=Qwen2.5-7B2026.05 | 20.8 | |
| Dr.ZeroBackbone=Qwen2.5-3B2026.05 | 20 | |
| R1Backbone=Qwen2.5-3B, Model Variant=Instruct2026.05 | 19.2 | |
| ZeroSearchBackbone=Qwen2.5-3B, Model Variant=Instruct2026.05 | 18.1 | |
| ZeroSearchBackbone=Qwen2.5-3B, Model Variant=Base2026.05 | 15.3 | |
| ZeroSearchBackbone=Qwen2.5-7B, Model Variant=Base2026.05 | 13.9 | |
| ReSearchBackbone=Qwen2.5-3B, Model Variant=Base2026.05 | 12.8 | |
| Search-R1Backbone=Qwen2.5-3B, Model Variant=Base2026.05 | 12.8 | |
| Direct InferenceBackbone=Qwen2.5-7B2026.05 | 12 | |
| SFTBackbone=Qwen2.5-3B2026.05 | 11.2 | |
| SFTBackbone=Qwen2.5-7B2026.05 | 11.2 | |
| ZeroSearchBackbone=Qwen2.5-7B, Model Variant=Instruct2026.05 | 11.1 | |
| RAGBackbone=Qwen2.5-3B2026.05 | 8 | |
| Direct GenerationBackbone=Qwen2.5-3B2026.05 | 2.4 | |
| CoTBackbone=Qwen2.5-3B2026.05 | 0 |