---
title: README
emoji: 👁
colorFrom: yellow
colorTo: gray
sdk: static
pinned: false
---

We are the LLM team at EPFL (Swiss Federal Institute of Technology), led by Prof. Antoine Bosselut from [NLP lab](https://nlp.epfl.ch), Prof. Martin Jaggi from [MLO lab](https://www.epfl.ch/labs/mlo/) and Prof. Mary-Anne Hartley from [LiGHT lab](https://www.yale-light.org). 

<p align="center">
  <img src="medllama.jpeg" width="35%">
</p>


Our latest project is **MEDITRON**, currently the best open-source medical Large Language Model in the world. 

We've publicly released the weights for [Meditron-70B](https://huggingface.co/epfl-llm/meditron-70b) and [Meditron-7B](https://huggingface.co/epfl-llm/meditron-7b) on Huggingface. 

- 🦾 **GitHub**: [epfLLM/meditron](https://github.com/epfLLM/meditron) and [epfLLM/Megatron-LLM](https://github.com/epfLLM/Megatron-LLM)

- 📖 **Paper**: [MEDITRON-70B: Scaling Medical Pre-Training For Large Language Models](https://arxiv.org/abs/2311.16079) (pre-print)

- 🗞️ **Press Release**: [EPFL's new Large Language Model for Medical Knowledge](https://actu.epfl.ch/news/epfl-s-new-large-language-model-for-medical-knowle/)

- 🧬 **NEW medical dataset**: [Clinical Guidelines Corpus](https://huggingface.co/datasets/epfl-llm/guidelines)