List your company

Multi-Hop Question Answering on Bamboogle (Accuracy)

47.6AccuracyMultiSearch
-1.90410.94823.836.652May 13, 2026

Evaluation results

MethodLinks
MultiSearchBackbone=Qwen2.5-7B, Model Variant=Base2026.0547.6
CriticSearchBackbone=Qwen2.5-7B2026.0547.2
StepSearchBackbone=Qwen2.5-7B, Model Variant=Base2026.0546.7
MultiSearchBackbone=Qwen2.5-7B, Model Variant=Instruct2026.0546.3
AutoRefineBackbone=Qwen2.5-7B, Model Variant=Base2026.0543.5
AutoRefineBackbone=Qwen2.5-7B, Model Variant=Instruct2026.0541.6
Search-R1Backbone=Qwen2.5-7B, Model Variant=Instruct2026.0540
MultiSearchBackbone=Qwen2.5-3B, Model Variant=Base2026.0539
MultiSearchBackbone=Qwen2.5-3B, Model Variant=Instruct2026.0537.1
CriticSearchBackbone=Qwen2.5-3B2026.0536.8
ReasonRAGBackbone=Qwen2.5-7B2026.0536
Dr.ZeroBackbone=Qwen2.5-7B2026.0536
Search-R1Backbone=Qwen2.5-7B, Model Variant=Base2026.0536
StepSearchBackbone=Qwen2.5-3B, Model Variant=Instruct2026.0534.4
AutoRefineBackbone=Qwen2.5-3B, Model Variant=Base2026.0534.4
AutoRefineBackbone=Qwen2.5-3B, Model Variant=Instruct2026.0533.6
StepSearchBackbone=Qwen2.5-3B, Model Variant=Base2026.0532.8
Search-o1Backbone=Qwen2.5-3B2026.0532
StepSearchBackbone=Qwen2.5-7B, Model Variant=Instruct2026.0531.2
R1Backbone=Qwen2.5-7B, Model Variant=Base2026.0529.6
Search-o1Backbone=Qwen2.5-7B2026.0529.6
R1Backbone=Qwen2.5-7B, Model Variant=Instruct2026.0529.3
AdaSearchBackbone=Qwen2.5-3B2026.0528
ReSearchBackbone=Qwen2.5-3B, Model Variant=Instruct2026.0526.6
IRCoTBackbone=Qwen2.5-3B2026.0524
Search-R1Backbone=Qwen2.5-3B, Model Variant=Instruct2026.0523.2
CoTBackbone=Qwen2.5-7B2026.0523.2
R1Backbone=Qwen2.5-3B, Model Variant=Base2026.0522.4
IRCoTBackbone=Qwen2.5-7B2026.0522.4
RAGBackbone=Qwen2.5-7B2026.0520.8
Dr.ZeroBackbone=Qwen2.5-3B2026.0520
R1Backbone=Qwen2.5-3B, Model Variant=Instruct2026.0519.2
ZeroSearchBackbone=Qwen2.5-3B, Model Variant=Instruct2026.0518.1
ZeroSearchBackbone=Qwen2.5-3B, Model Variant=Base2026.0515.3
ZeroSearchBackbone=Qwen2.5-7B, Model Variant=Base2026.0513.9
ReSearchBackbone=Qwen2.5-3B, Model Variant=Base2026.0512.8
Search-R1Backbone=Qwen2.5-3B, Model Variant=Base2026.0512.8
Direct InferenceBackbone=Qwen2.5-7B2026.0512
SFTBackbone=Qwen2.5-3B2026.0511.2
SFTBackbone=Qwen2.5-7B2026.0511.2
ZeroSearchBackbone=Qwen2.5-7B, Model Variant=Instruct2026.0511.1
RAGBackbone=Qwen2.5-3B2026.058
Direct GenerationBackbone=Qwen2.5-3B2026.052.4
CoTBackbone=Qwen2.5-3B2026.050