Question Answering on Neuroscience QA Overall (Full set)
89.5Average AccuracyQwen 14B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen 14BTraining Phase=SFT + RL2026.05 | 89.5 | |
| Qwen 14BTraining Phase=SFT2026.05 | 87.3 | |
| Gemini 3.1 ProModel=Gemini 3.1 Pro2026.05 | 84.8 | |
| Qwen 14BTraining Phase=Base2026.05 | 78.1 |