List your company

Multi-hop Question Answering on 2Wiki (EM, LJ)

48.89EMM2R
8.53819.01429.4939.966Apr 10, 2026

Evaluation results

MethodLinks
M2RBackbone=Qwen2.5-7B-Instruct2026.0448.8957.01
ReSearchBackbone=Qwen2.5-7B-Instruct2026.0447.5954.22
SUREBackbone=Qwen2.5-7B-Instruct2026.0445.6553.93
COFTBackbone=Qwen2.5-7B-Instruct2026.0441.8648.7
M2RBackbone=Qwen2.5-3B-Instruct2026.0440.0748.34
ReSearchBackbone=Qwen2.5-3B-Instruct2026.0438.947.41
SUREBackbone=Qwen2.5-3B-Instruct2026.0436.242.38
COFTBackbone=Qwen2.5-3B-Instruct2026.0432.8239.76
Iter-RetGenBackbone=Qwen2.5-7B-Instruct2026.0427.9231.86
Naive RAGBackbone=Qwen2.5-7B-Instruct2026.0425.7829.52
Naive GenerationBackbone=Qwen2.5-7B-Instruct2026.0425.7627.87
IRCoTBackbone=Qwen2.5-7B-Instruct2026.0421.5730.65
Iter-RetGenBackbone=Qwen2.5-3B-Instruct2026.0414.7528.86
Naive RAGBackbone=Qwen2.5-3B-Instruct2026.0413.8223.08
IRCoTBackbone=Qwen2.5-3B-Instruct2026.0412.523.54
Naive GenerationBackbone=Qwen2.5-3B-Instruct2026.0410.0919.79