Single-Hop Question Answering on NQ (Accuracy)
49.1AccuracyMultiSearch
Evaluation Results
| Method | Links | |
|---|---|---|
| MultiSearchBackbone=Qwen2.5-7B, Model Variant=Base2026.05 | 49.1 | |
| MultiSearchBackbone=Qwen2.5-7B, Model Variant=Instruct2026.05 | 47.2 | |
| MultiSearchBackbone=Qwen2.5-3B, Model Variant=Base2026.05 | 47.1 | |
| MultiSearchBackbone=Qwen2.5-3B, Model Variant=Instruct2026.05 | 47 | |
| AutoRefineBackbone=Qwen2.5-7B, Model Variant=Base2026.05 | 47 | |
| AutoRefineBackbone=Qwen2.5-3B, Model Variant=Base2026.05 | 46.7 | |
| AutoRefineBackbone=Qwen2.5-3B, Model Variant=Instruct2026.05 | 43.6 | |
| ZeroSearchBackbone=Qwen2.5-3B, Model Variant=Base2026.05 | 43.4 | |
| ZeroSearchBackbone=Qwen2.5-7B, Model Variant=Base2026.05 | 43 | |
| Search-R1Backbone=Qwen2.5-7B, Model Variant=Instruct2026.05 | 42.9 | |
| ReSearchBackbone=Qwen2.5-3B, Model Variant=Base2026.05 | 42.7 | |
| Search-R1Backbone=Qwen2.5-3B, Model Variant=Base2026.05 | 42.1 | |
| AutoRefineBackbone=Qwen2.5-7B, Model Variant=Instruct2026.05 | 41.9 | |
| ZeroSearchBackbone=Qwen2.5-7B, Model Variant=Instruct2026.05 | 41.4 | |
| Dr.ZeroBackbone=Qwen2.5-7B2026.05 | 40.6 | |
| ZeroSearchBackbone=Qwen2.5-3B, Model Variant=Instruct2026.05 | 40.2 | |
| Search-R1Backbone=Qwen2.5-3B, Model Variant=Instruct2026.05 | 39.7 | |
| Search-R1Backbone=Qwen2.5-7B, Model Variant=Base2026.05 | 39.5 | |
| Dr.ZeroBackbone=Qwen2.5-3B2026.05 | 39.1 | |
| AdaSearchBackbone=Qwen2.5-3B2026.05 | 37.9 | |
| ReSearchBackbone=Qwen2.5-3B, Model Variant=Instruct2026.05 | 36.5 | |
| RAGBackbone=Qwen2.5-7B2026.05 | 34.9 | |
| RAGBackbone=Qwen2.5-3B2026.05 | 34.8 | |
| SFTBackbone=Qwen2.5-7B2026.05 | 31.8 | |
| R1Backbone=Qwen2.5-7B, Model Variant=Base2026.05 | 29.7 | |
| R1Backbone=Qwen2.5-7B, Model Variant=Instruct2026.05 | 27 | |
| SFTBackbone=Qwen2.5-3B2026.05 | 24.9 | |
| Search-o1Backbone=Qwen2.5-3B2026.05 | 23.8 | |
| R1Backbone=Qwen2.5-3B, Model Variant=Base2026.05 | 22.6 | |
| IRCoTBackbone=Qwen2.5-7B2026.05 | 22.4 | |
| R1Backbone=Qwen2.5-3B, Model Variant=Instruct2026.05 | 21 | |
| Search-o1Backbone=Qwen2.5-7B2026.05 | 15.1 | |
| Direct InferenceBackbone=Qwen2.5-7B2026.05 | 13.4 | |
| IRCoTBackbone=Qwen2.5-3B2026.05 | 11.1 | |
| Direct GenerationBackbone=Qwen2.5-3B2026.05 | 10.6 | |
| CoTBackbone=Qwen2.5-7B2026.05 | 4.8 | |
| CoTBackbone=Qwen2.5-3B2026.05 | 2.3 |