Exact Match Accuracy on HotpotQA
43.5Exact Match AccuracyReSearch
Evaluation Results
| Method | Links | |
|---|---|---|
| ReSearchBackbone=Qwen2.5-7B-Instruct2026.05 | 43.5 | |
| PyRAG-RLBackbone=LLaMA-3.1-8B-Instruct2026.05 | 43.2 | |
| PyRAG-RLBackbone=Qwen2.5-7B-Instruct2026.05 | 40.5 | |
| StepSearchBackbone=Qwen2.5-7B-Instruct2026.05 | 38.6 | |
| PyRAG-RLBackbone=Qwen3-4B-Instruct2026.05 | 38.4 | |
| RAG-RLBackbone=LLaMA-3.1-8B-Instruct2026.05 | 37.5 | |
| Search-R1Backbone=Qwen2.5-7B-Instruct2026.05 | 37 | |
| RAG-RLBackbone=Qwen2.5-7B-Instruct2026.05 | 35.2 | |
| ZEROSEARCHBackbone=Qwen2.5-7B-Instruct2026.05 | 34.6 | |
| RAG-SFTBackbone=LLaMA-3.1-8B-Instruct2026.05 | 34.1 | |
| RAG-RLBackbone=Qwen3-4B-Instruct2026.05 | 33.2 | |
| RAG-SFTBackbone=Qwen2.5-7B-Instruct2026.05 | 32.4 | |
| RAG-SFTBackbone=Qwen3-4B-Instruct2026.05 | 30.5 | |
| Vanilla RAGBackbone=LLaMA-3.1-8B-Instruct2026.05 | 30.3 | |
| Vanilla RAGBackbone=Qwen2.5-7B-Instruct2026.05 | 28.9 | |
| Vanilla RAGBackbone=Qwen3-4B-Instruct2026.05 | 27.1 |