Multi-hop QA on HotpotQA (test)
45.2Exact Match (EM)ParetoPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ParetoPOTool Use=Yes, Base Model=Qwen2.5-7B2026.06 | 45.2 | 0.9 | |
| ParetoPOTool Use=Yes, Base Model=Qwen2.5-3B2026.06 | 42.2 | 1.2 | |
| OTC-GRPOTool Use=Yes, Base Model=Qwen2.5-7B2026.06 | 36.6 | 1 | |
| OTC-GRPOTool Use=Yes, Base Model=Qwen2.5-3B2026.06 | 36.5 | 1.4 | |
| Search-R1Tool Use=Yes, Base Model=Qwen2.5-7B2026.06 | 34.1 | 2.1 | |
| MO-GRPOTool Use=Yes, Base Model=Qwen2.5-7B2026.06 | 33.1 | 1.6 | |
| MO-GRPOTool Use=Yes, Base Model=Qwen2.5-3B2026.06 | 32 | 1.9 | |
| Search-R1Tool Use=Yes, Base Model=Qwen2.5-3B2026.06 | 31.2 | 1.8 | |
| RAGTool Use=Yes, Base Model=Qwen2.5-7B2026.06 | 29.9 | 1 | |
| RAGTool Use=Yes, Base Model=Qwen2.5-3B2026.06 | 25.5 | 1 | |
| R1-BaseTool Use=No, Base Model=Qwen2.5-7B2026.06 | 24.2 | — | |
| SFTTool Use=No, Base Model=Qwen2.5-7B2026.06 | 21.7 | — | |
| R1-BaseTool Use=No, Base Model=Qwen2.5-3B2026.06 | 20.1 | — | |
| SFTTool Use=No, Base Model=Qwen2.5-3B2026.06 | 18.6 | — |