Agentic Search on 2WikiMultiHopQA
69.9String-F1Vanilla-Qwen3-32B
Evaluation Results
| Method | Links | |
|---|---|---|
| Vanilla-Qwen3-32BASP=false2026.04 | 69.9 | |
| OPD-Qwen3-1.7BASP=true2026.04 | 62.9 | |
| SFT-Llama-3.2-3BASP=true2026.04 | 58.8 | |
| SFT-Qwen3-1.7BASP=true2026.04 | 58.5 | |
| Vanilla-Qwen3-8BASP=false2026.04 | 58.1 | |
| Mixed-Qwen3-1.7BASP=true2026.04 | 57.8 | |
| Vanilla-Qwen3-4BASP=false2026.04 | 54.7 | |
| SFT-Llama-3.2-1BASP=true2026.04 | 48.2 | |
| SFT-Qwen3-0.6BASP=true2026.04 | 47.4 | |
| Distilled-Llama3.2-3BASP=false2026.04 | 46.3 | |
| Distilled-Qwen3-1.7BASP=false2026.04 | 44.1 | |
| Vanilla-Qwen3-1.7BASP=false2026.04 | 39.8 | |
| Distilled-Llama3.2-1BASP=false2026.04 | 30.9 | |
| Vanilla-Qwen3-0.6BASP=false2026.04 | 26 |