List your company

Multi-Hop Question Answering on 2wiki 2018 Wikipedia dump (dev)

43.6Accuracy (%)MR-Search
16.5623.5830.637.62Mar 11, 2026

Evaluation results

MethodLinks
MR-SearchBackbone=Qwen2.5-7b2026.0343.6
StepResearchBackbone=Qwen2.5-7b2026.0341.8
MR-SearchBackbone=Qwen2.5-3b2026.0340.1
Search-R1Backbone=Qwen2.5-7b2026.0338.7
ReSearchBackbone=Qwen2.5-7b2026.0338.6
PPRMBackbone=Qwen2.5-7b2026.0335.5
PPRMBackbone=Qwen2.5-3b2026.0334
StepResearchBackbone=Qwen2.5-3b2026.0333.8
Search-R1Backbone=Qwen2.5-3b2026.0331
ReSearchBackbone=Qwen2.5-3b2026.0327.2
Direct InferenceBackbone=Qwen2.5-7b2026.0325
Direct InferenceBackbone=Qwen2.5-3b2026.0324.4
Search-o1Backbone=Qwen2.5-3b2026.0321.8
Search-o1Backbone=Qwen2.5-7b2026.0317.6