Medical history-taking on MIMIC
37.8PrecisionLlama-3.1-8B (IGFT)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Llama-3.1-8B (IGFT)Fine-tuning=IGFT, Base Model=Llama-3.1-8B-Instruct2026.01 | 37.8 | 34 | 33.6 | |
| Llama-3.1-8B-InstructFine-tuning=None2026.01 | 35 | 32.7 | 30.8 | |
| DeepSeek-R1-7B (IGFT)Fine-tuning=IGFT, Base Model=DeepSeek-R1-Distill-Qwen-7B2026.01 | 34.8 | 28 | 28.9 | |
| GPT-4o-mini2026.01 | 32.5 | 34.4 | 30.4 | |
| HuatuoGPT-o1-7BDomain-specific=true2026.01 | 32.2 | 30.2 | 28.3 | |
| DeepSeek-R1-Distill-Qwen-7BFine-tuning=None2026.01 | 31.4 | 26.5 | 25.6 | |
| UltraMedical-8BDomain-specific=true2026.01 | 30.6 | 33 | 28.6 |