37 lines
1.4 KiB
Markdown
37 lines
1.4 KiB
Markdown
|
---
|
|||
|
|
license: llama3
|
|||
|
|
datasets:
|
|||
|
|
- Henrychur/MMedC
|
|||
|
|
- Henrychur/MedS-Ins
|
|||
|
|
language:
|
|||
|
|
- en
|
|||
|
|
base_model: Henrychur/MMedS-Llama-3-8B
|
|||
|
|
tags:
|
|||
|
|
- medical
|
|||
|
|
library_name: transformers
|
|||
|
|
---
|
|||
|
|
|
|||
|
|
|
|||
|
|
# MMedS-Llama3
|
|||
|
|
[💻Github Repo](https://github.com/MAGIC-AI4Med/MedS-Ins) [🖨️arXiv Paper](https://arxiv.org/abs/2408.12547)
|
|||
|
|
|
|||
|
|
The official codes for "Towards Evaluating and Building Versatile Large Language Models for Medicine"
|
|||
|
|
|
|||
|
|
|
|||
|
|
## Introduction
|
|||
|
|
This repository hosts MMedS-Llama-3-8B. Its foundation model, [MMed-Llama-3-8B](https://huggingface.co/Henrychur/MMed-Llama-3-8B),
|
|||
|
|
is a multilingual medical language model which has undergone additional continuous pretraining on MMedC. Furthermore, the model has
|
|||
|
|
been fine-tuned under supervision using MedS-Ins, a comprehensive dataset designed specifically for supervised fine-tuning (SFT),
|
|||
|
|
featuring 13.5 million samples across 122 tasks. For more details, please refer to our paper.
|
|||
|
|
|
|||
|
|
|
|||
|
|
## Usage
|
|||
|
|
The model can be loaded as follows:
|
|||
|
|
```py
|
|||
|
|
import torch
|
|||
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM
|
|||
|
|
tokenizer = AutoTokenizer.from_pretrained("Henrychur/MMed-Llama-3-8B-EnIns")
|
|||
|
|
model = AutoModelForCausalLM.from_pretrained("Henrychur/MMed-Llama-3-8B-EnIns", torch_dtype=torch.float16)
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
- Inference format is the same as Llama 3, you can check the inference code [here](https://github.com/MAGIC-AI4Med/MedS-Ins/blob/main/Inference/model.py).
|