Question Answering on Neuroscience QA 3-hop
90.7AccuracyQwen 14B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen 14BTraining Phase=SFT + RL2026.05 | 90.7 | |
| Qwen 14BTraining Phase=SFT2026.05 | 88.7 | |
| Gemini 3.1 ProModel=Gemini 3.1 Pro2026.05 | 86.9 | |
| Qwen 14BTraining Phase=Base2026.05 | 81.2 |