Multi-hop Question Answering on MuSiQue standard (val)
22.6Exact Match (EM)StepSearch-instruct
Evaluation Results
| Method | Links | |
|---|---|---|
| StepSearch-instructType=Agentic RAG, Backbone=Qwen2.5-7B-Instruct2026.03 | 22.6 | |
| ZeroSearch-instructType=Agentic RAG, Backbone=Qwen2.5-7B-Instruct2026.03 | 18.4 | |
| StepSearch-instructType=Agentic RAG, Backbone=Qwen2.5-3B-Instruct2026.03 | 17.4 | |
| Apex-SearcherType=Agentic RAG, Backbone=Qwen2.5-7B-Instruct2026.03 | 16.4 | |
| Search-R1-InstructType=Agentic RAG, Backbone=Qwen2.5-7B-Instruct2026.03 | 14.6 | |
| Apex-SearcherType=Agentic RAG, Backbone=Qwen2.5-3B-Instruct2026.03 | 13.6 | |
| ReasonRAGType=Agentic RAG, Backbone=Qwen2.5-7B-Instruct2026.03 | 12.8 | |
| ZeroSearch-instructType=Agentic RAG, Backbone=Qwen2.5-3B-Instruct2026.03 | 9.8 | |
| IRCoTType=Iterative RAG, Backbone=Qwen2.5-7B-Instruct2026.03 | 7.2 | |
| IRCoTType=Iterative RAG, Backbone=Qwen2.5-3B-Instruct2026.03 | 6.7 | |
| RAGType=Standard RAG, Backbone=Qwen2.5-7B-Instruct2026.03 | 5.8 | |
| Search-o1Type=Agentic RAG, Backbone=Qwen2.5-7B-Instruct2026.03 | 5.8 | |
| Search-o1Type=Agentic RAG, Backbone=Qwen2.5-3B-Instruct2026.03 | 5.4 | |
| Search-R1-InstructType=Agentic RAG, Backbone=Qwen2.5-3B-Instruct2026.03 | 4.9 | |
| RAGType=Standard RAG, Backbone=Qwen2.5-3B-Instruct2026.03 | 4.7 | |
| Direct InferenceType=Non-Retrieval, Backbone=Qwen2.5-7B-Instruct2026.03 | 3.1 | |
| CoTType=Non-Retrieval, Backbone=Qwen2.5-7B-Instruct2026.03 | 2.2 | |
| Direct InferenceType=Non-Retrieval, Backbone=Qwen2.5-3B-Instruct2026.03 | 2 | |
| CoTType=Non-Retrieval, Backbone=Qwen2.5-3B-Instruct2026.03 | 0.2 |