Pediatric Medical Question Answering on PediaBench MC
82.89AccuracySafeMed-R1
Evaluation results
| Method | Links | |
|---|---|---|
| SafeMed-R1Optimization=CoT reasoning, safety-ethics alignment, improved GRPO2026.05 | 82.89 | |
| Qwen3-32B+SFT2026.05 | 81.77 | |
| qwen3-30B-A3B-Instruct-25072026.05 | 81.38 | |
| Qwen3-32B2026.05 | 81.35 | |
| Baichuan-M2-32B2026.05 | 80.62 | |
| gemini-2.5-flash2026.05 | 78.8 | |
| gpt-5-mini2026.05 | 77.29 | |
| DeepSeek-R1-Distill-Qwen-32B2026.05 | 74.52 | |
| Qwen2.5-32B-Instruct2026.05 | 73.35 | |
| google_medgemma-27b-text-it2026.05 | 68.46 | |
| HuatuoGPT-o1-70B2026.05 | 67.2 | |
| Lingshu-32B2026.05 | 49.33 |