Multi-turn Dialogue Quality on Dementia Personas Simulation
4.44Pers. ScoreDemMA
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| DemMAEvaluator=Gemini-2.5-pro2026.01 | 4.44 | 4.56 | 4.5 | |
| DemMAEvaluator=Qwen3-32B2026.01 | 4.35 | 4.75 | 4.55 | |
| DemMAEvaluator=GPT-5.2-pro2026.01 | 4.11 | 3.78 | 3.95 | |
| DemMAEvaluator=Expert Evaluation2026.01 | 4 | 3.5 | 3.75 | |
| Clinical-Profile PromptEvaluator=Expert Evaluation2026.01 | 2.5 | 2 | 2.25 | |
| SFT-UtteranceEvaluator=Expert Evaluation2026.01 | 2.5 | 3 | 2.75 | |
| SFT-UtteranceEvaluator=Qwen3-32B2026.01 | 2.4 | 3.65 | 3.03 | |
| Clinical-Profile PromptEvaluator=GPT-5.2-pro2026.01 | 2.33 | 1.67 | 2 | |
| SFT-UtteranceEvaluator=GPT-5.2-pro2026.01 | 2.33 | 2.89 | 2.61 | |
| SFT-UtteranceEvaluator=Gemini-2.5-pro2026.01 | 2.31 | 3.22 | 2.76 | |
| Clinical-Profile PromptEvaluator=Qwen3-32B2026.01 | 2.3 | 3.15 | 2.73 | |
| Clinical-Profile PromptEvaluator=Gemini-2.5-pro2026.01 | 2.22 | 1.56 | 1.89 | |
| VanillaEvaluator=GPT-5.2-pro2026.01 | 2.16 | 1.5 | 1.83 | |
| VanillaEvaluator=Qwen3-32B2026.01 | 2.05 | 2.95 | 2.5 | |
| VanillaEvaluator=Expert Evaluation2026.01 | 2 | 1.5 | 1.75 | |
| VanillaEvaluator=Gemini-2.5-pro2026.01 | 1.8 | 1.6 | 1.7 |