Multi-hop QA on CUAD GE (test)
38.58Token F1Qwen3-32B-Gemini-SFT
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Qwen3-32B-Gemini-SFTBackbone=Qwen3-32B, SFT=Gemini data2026.05 | 38.58 | 2.34 | 89.04 | |
| Qwen3-32B-BaseBackbone=Qwen3-32B, SFT=None2026.05 | 21.66 | 2.74 | 53.98 |
| Method | Links | |||
|---|---|---|---|---|
| Qwen3-32B-Gemini-SFTBackbone=Qwen3-32B, SFT=Gemini data2026.05 | 38.58 | 2.34 | 89.04 | |
| Qwen3-32B-BaseBackbone=Qwen3-32B, SFT=None2026.05 | 21.66 | 2.74 | 53.98 |