Dataset · epfl-llm
Clinical guidelines for medical LLMs
epfl-llm/guidelines
47k clinical guidelines from 17 medical sources for training and evaluating medical LLMs.
What's inside
Clinical practice guideline texts in English. The corpus features about 47k documents from 17 high-quality online medical sources; a subset of about 37k is publicly available, plus 1,627 guidelines from PubMed and PubMed Central added in December 2023. This is part of the original Meditron LLM training corpus.
What you can do with it
- Fine-tuning (SFT) a medical LLM on clinical texts
- Evaluating models on clinical guideline knowledge
- Fine-tuning LoRA for medical protocol responses
- Tasks
- Text generation
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 2.5K
- Likes
- 167
Models trained on it
- microsoft/MediPhi-InstructText generation 3K
- microsoft/MediPhiText generation 2.6K
- TheBloke/meditron-7B-GGUF 2K
- TheBloke/meditron-70B-GGUFText generation 1K
- mradermacher/MediPhi-i1-GGUF 872
- mradermacher/Gemma-3-Earthen-Completion-v0.1-4B-GGUF 708
- microsoft/MediPhi-GuidelinesText generation 671
- mradermacher/MediPhi-Instruct-i1-GGUF 667
How to load it
from datasets import load_dataset
ds = load_dataset("epfl-llm/guidelines")