Medical Question Answering on MedQA 5-options Full (test)
83.6AccuracyMDAgents
Evaluation results
| Method | Links | |
|---|---|---|
| MDAgentsCategory=Adaptive, Model=GPT-4o mini2024.04 | 83.6 | |
| ReconcileCategory=Multi-agent (Single-model), Model=GPT-4o mini2024.04 | 80.2 | |
| Weighted VotingCategory=Multi-agent (Multi-model), Model=GPT-4o mini2024.04 | 79.1 | |
| + CoT-SCCategory=Single-agent, Model=GPT-4o mini2024.04 | 77.2 | |
| + CoTCategory=Single-agent, Model=GPT-4o mini2024.04 | 76.6 | |
| Majority VotingCategory=Multi-agent (Multi-model), Model=GPT-4o mini2024.04 | 76.3 | |
| Borda CountCategory=Multi-agent (Multi-model), Model=GPT-4o mini2024.04 | 76.1 | |
| 3-shotCategory=Single-agent, Model=GPT-4o mini2024.04 | 72.3 | |
| Zero-shotCategory=Single-agent, Model=GPT-4o mini2024.04 | 71.5 | |
| LLaMAModel Size=65B2025.12 | 47 | |
| FBPRCorpus=Dolma2025.12 | 46.7 | |
| FBPRCorpus=RedPajama2025.12 | 44.5 | |
| OLMo InstructModel Size=7B, Model Variant=Instruct2025.12 | 44.1 | |
| Random Chance2025.12 | 20 |