List your company

Multi-Hop Question Answering on 2Wiki (Accuracy)

43.6Accuracy (2Wiki)ReasonRAG
0.4411.64522.8534.055May 13, 2026

Evaluation results

MethodLinks
ReasonRAGBackbone=Qwen2.5-7B2026.0543.6
CriticSearchBackbone=Qwen2.5-7B2026.0542.8
MultiSearchBackbone=Qwen2.5-7B, Model Variant=Base2026.0541.6
MultiSearchBackbone=Qwen2.5-3B, Model Variant=Base2026.0541.3
MultiSearchBackbone=Qwen2.5-3B, Model Variant=Instruct2026.0541.2
CriticSearchBackbone=Qwen2.5-3B2026.0540.9
AutoRefineBackbone=Qwen2.5-3B, Model Variant=Base2026.0539.3
AdaSearchBackbone=Qwen2.5-3B2026.0538.5
StepSearchBackbone=Qwen2.5-7B, Model Variant=Base2026.0538.5
AutoRefineBackbone=Qwen2.5-3B, Model Variant=Instruct2026.0538
AutoRefineBackbone=Qwen2.5-7B, Model Variant=Base2026.0536.9
StepSearchBackbone=Qwen2.5-7B, Model Variant=Instruct2026.0536.6
ZeroSearchBackbone=Qwen2.5-3B, Model Variant=Base2026.0535.6
MultiSearchBackbone=Qwen2.5-7B, Model Variant=Instruct2026.0535.1
Dr.ZeroBackbone=Qwen2.5-7B2026.0534.7
Search-R1Backbone=Qwen2.5-7B, Model Variant=Instruct2026.0534.6
ZeroSearchBackbone=Qwen2.5-7B, Model Variant=Base2026.0534.6
StepSearchBackbone=Qwen2.5-3B, Model Variant=Base2026.0533.9
AutoRefineBackbone=Qwen2.5-7B, Model Variant=Instruct2026.0532.5
StepSearchBackbone=Qwen2.5-3B, Model Variant=Instruct2026.0532
Search-R1Backbone=Qwen2.5-3B, Model Variant=Instruct2026.0531
ZeroSearchBackbone=Qwen2.5-7B, Model Variant=Instruct2026.0530
Search-R1Backbone=Qwen2.5-7B, Model Variant=Base2026.0529.7
R1Backbone=Qwen2.5-7B, Model Variant=Instruct2026.0529.2
Dr.ZeroBackbone=Qwen2.5-3B2026.0529.1
R1Backbone=Qwen2.5-3B, Model Variant=Instruct2026.0527.5
Search-R1Backbone=Qwen2.5-3B, Model Variant=Base2026.0527.4
R1Backbone=Qwen2.5-7B, Model Variant=Base2026.0527.3
ReSearchBackbone=Qwen2.5-3B, Model Variant=Instruct2026.0527.2
ReSearchBackbone=Qwen2.5-3B, Model Variant=Base2026.0527.2
R1Backbone=Qwen2.5-3B, Model Variant=Base2026.0526.8
SFTBackbone=Qwen2.5-7B2026.0525.9
Direct InferenceBackbone=Qwen2.5-7B2026.0525
SFTBackbone=Qwen2.5-3B2026.0524.8
Direct GenerationBackbone=Qwen2.5-3B2026.0524.4
RAGBackbone=Qwen2.5-7B2026.0523.5
RAGBackbone=Qwen2.5-3B2026.0522.6
Search-o1Backbone=Qwen2.5-3B2026.0521.8
ZeroSearchBackbone=Qwen2.5-3B, Model Variant=Instruct2026.0521.4
Search-o1Backbone=Qwen2.5-7B2026.0517.6
IRCoTBackbone=Qwen2.5-3B2026.0517.1
IRCoTBackbone=Qwen2.5-7B2026.0514.9
CoTBackbone=Qwen2.5-7B2026.0511.1
CoTBackbone=Qwen2.5-3B2026.052.1