List your company

General Question Answering on PopQA out-of-domain (val test)

50.1Exact Match (EM)Search-R2
3.61215.68127.7539.819Feb 3, 2026

Evaluation results

MethodLinks
Search-R2Backbone=Qwen2.5-32B2026.0250.1
Search-R1Backbone=Qwen2.5-32B2026.0247
Search-R2Backbone=Qwen3-8B2026.0246.6
Search-R1Backbone=Qwen3-8B2026.0241.8
Search-R2Backbone=Qwen2.5-7B2026.0241
RAGBackbone=Qwen2.5-7B2026.0239.2
Search-R1Backbone=Qwen2.5-7B2026.0238.8
Rejection SamplingBackbone=Qwen2.5-7B2026.0238
IRCoTBackbone=Qwen2.5-7B2026.0230.1
R1-baseBackbone=Qwen2.5-7B2026.0220.2
R1-instructBackbone=Qwen2.5-7B2026.0219.9
Direct InferenceBackbone=Qwen2.5-7B2026.0214
Search-o1Backbone=Qwen2.5-7B2026.0213.1
SFTBackbone=Qwen2.5-7B2026.0212.1
CoTBackbone=Qwen2.5-7B2026.025.4