Biomedical Question Answering on MedMCQA In-Domain (test)
90AccuracyHuman (expert)
Evaluation Results
| Method | Links | |
|---|---|---|
| Human (expert)Setting=Manual2023.08 | 90 | |
| BioMedGPT-10BSetting=Fine-tuning2023.08 | 51.4 | |
| PMC-LlamaSetting=Fine-tuning2023.08 | 50.5 | |
| Llama2-ChatSetting=Fine-tuning2023.08 | 48.3 | |
| LlamaSetting=Fine-tuning2023.08 | 48.2 | |
| ChatGPTSetting=Zero-shot2023.08 | 44.7 | |
| InstructGPTSetting=Zero-shot2023.08 | 44 | |
| Llama2-ChatSetting=Zero-shot2023.08 | 35.5 | |
| Llama2Setting=Zero-shot2023.08 | 30.6 | |
| LlamaSetting=Zero-shot2023.08 | 24.3 |