List your company

Open-domain Question Answering on PopQA (EM, RAG Improvement)

45.5Accuracy (EM)PyRAG
3.79614.62325.4536.277May 13, 2026

Evaluation results

MethodLinks
PyRAGBackbone=Qwen2.5-72B-Instruct2026.0545.512.3
IRCoTBackbone=Qwen2.5-72B-Instruct2026.0544.6—
ITER-RETGENBackbone=Qwen2.5-72B-Instruct2026.0543.4—
Self-AskBackbone=Qwen2.5-72B-Instruct2026.0541.4—
PyRAGBackbone=Qwen2.5-7B-Instruct2026.0533.56.8
Vanilla RAGBackbone=Qwen2.5-72B-Instruct2026.0533.2—
IRCoTBackbone=Qwen2.5-7B-Instruct2026.0532.6—
ITER-RETGENBackbone=Qwen2.5-7B-Instruct2026.0531.4—
Self-AskBackbone=Qwen2.5-7B-Instruct2026.0529.4—
Vanilla RAGBackbone=Qwen2.5-7B-Instruct2026.0526.7—
CoTBackbone=Qwen2.5-72B-Instruct2026.0524.4—
Direct InferenceBackbone=Qwen2.5-72B-Instruct2026.0519.7—
Direct InferenceBackbone=Qwen2.5-7B-Instruct2026.0514—
CoTBackbone=Qwen2.5-7B-Instruct2026.055.4—