Multi-Turn Medical Dialogue on MedMCQA
63.31AccuracyMEDIQ
Evaluation Results
| Method | Links | |
|---|---|---|
| MEDIQModel=Gemini-2.5-Pro, Method Type=Prompt2026.02 | 63.31 | |
| TreePOModel=Qwen3-8B, Method Type=SFT+RL2026.02 | 54.74 | |
| ATPO (U1 + U2)Model=Qwen3-8B, Method Type=SFT+RL2026.02 | 53.66 | |
| ATPO (U1)Model=Qwen3-8B, Method Type=SFT+RL2026.02 | 53.22 | |
| MEDIQModel=GPT-4o, Method Type=Prompt2026.02 | 53.03 | |
| PPO (H-MDP)Model=Qwen3-8B, Method Type=SFT+RL2026.02 | 51.98 | |
| GRPOModel=Qwen3-8B, Method Type=SFT+RL2026.02 | 51.12 | |
| PPO (MDP)Model=Qwen3-8B, Method Type=SFT+RL2026.02 | 50 | |
| SFTModel=Qwen3-8B, Method Type=SFT2026.02 | 46.87 | |
| ATPO (U1)Model=Qwen3-4B, Method Type=SFT+RL2026.02 | 46.23 | |
| DirectModel=Qwen3-8B, Method Type=Prompt2026.02 | 46.16 | |
| ATPO (U1 + U2)Model=Qwen3-4B, Method Type=SFT+RL2026.02 | 45.93 | |
| TreePOModel=Qwen3-4B, Method Type=SFT+RL2026.02 | 45.22 | |
| GRPOModel=Qwen3-4B, Method Type=SFT+RL2026.02 | 43.84 | |
| PPO (H-MDP)Model=Qwen3-4B, Method Type=SFT+RL2026.02 | 43.32 | |
| PPO (MDP)Model=Qwen3-4B, Method Type=SFT+RL2026.02 | 42.5 | |
| DFTModel=Qwen3-8B, Method Type=SFT2026.02 | 42.2 | |
| MEDIQModel=Qwen3-8B, Method Type=Prompt2026.02 | 41.6 | |
| DirectModel=Qwen3-4B, Method Type=Prompt2026.02 | 41.53 | |
| ATPO (U1 + U2)Model=Qwen3-1.7B, Method Type=SFT+RL2026.02 | 39.93 | |
| SFTModel=Qwen3-4B, Method Type=SFT2026.02 | 39.18 | |
| ATPO (U1)Model=Qwen3-1.7B, Method Type=SFT+RL2026.02 | 38.66 | |
| TreePOModel=Qwen3-1.7B, Method Type=SFT+RL2026.02 | 38.47 | |
| GRPOModel=Qwen3-1.7B, Method Type=SFT+RL2026.02 | 36.57 | |
| PPO (MDP)Model=Qwen3-1.7B, Method Type=SFT+RL2026.02 | 35.37 | |
| PPO (H-MDP)Model=Qwen3-1.7B, Method Type=SFT+RL2026.02 | 34.89 | |
| DFTModel=Qwen3-4B, Method Type=SFT2026.02 | 33.28 | |
| DirectModel=Qwen3-1.7B, Method Type=Prompt2026.02 | 32.54 | |
| MEDIQModel=Qwen3-1.7B, Method Type=Prompt2026.02 | 32.35 | |
| MEDIQModel=Qwen3-4B, Method Type=Prompt2026.02 | 31.64 | |
| SFTModel=Qwen3-1.7B, Method Type=SFT2026.02 | 28.1 | |
| DFTModel=Qwen3-1.7B, Method Type=SFT2026.02 | 21.08 |