MMLU Clinical Knowledge (Open Medical-LLM)
Science · 2024-04-19
The clinical-knowledge subset of MMLU, 265 questions, scored by the Open Medical-LLM Leaderboard under lm-evaluation-harness with a fixed prompt.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-4 | 86.4 |
| Llama 3 8B | 74.7 |
| GPT-3.5 Turbo 1106 | 74.7 |
| Solar 10.7B Instruct V1.0 | 72.1 |
| Llama 3 8B Instruct | 70.9 |
| Gemma 7B | 69.4 |
| Mistral 7B v0.1 Base | 68.7 |
| Qwen1.5 7B Chat | 68.3 |
| Qwen1.5 7B | 66.0 |
| Gemma 1.1 7B IT | 64.9 |
| Zephyr 7B Beta | 64.5 |
| Mistral 7B Instruct v0.1 | 61.1 |
| Gemma 7B IT | 55.1 |
| Phi-1.5 | 46.0 |
| Gemma 2B | 35.5 |
| Falcon 7B | 21.9 |