60 lines
1.8 KiB
Markdown
60 lines
1.8 KiB
Markdown
---
|
|
base_model: unsloth/Qwen3-8B-unsloth-bnb-4bit
|
|
tags:
|
|
- text-generation
|
|
- conversational
|
|
- medical
|
|
- qa
|
|
- transformers
|
|
- unsloth
|
|
- qwen3
|
|
license: apache-2.0
|
|
language:
|
|
- en
|
|
- pt
|
|
---
|
|
|
|
# Qwen3-8B Medical (Fine-tuned)
|
|
|
|
- **Developed by:** lgsantini1
|
|
- **License:** apache-2.0
|
|
- **Finetuned from:** unsloth/Qwen3-8B-unsloth-bnb-4bit
|
|
|
|
## Overview
|
|
This is a Qwen3-8B model fine-tuned for medical-style question answering based on publicly available QA datasets.
|
|
|
|
## Training data
|
|
This model was fine-tuned using data sourced from:
|
|
|
|
- **PubMedQA** — A dataset of question answering pairs grounded in biomedical research abstracts.
|
|
Repo: https://github.com/pubmedqa/pubmedqa
|
|
(Used only the content available in the repository; no additional web crawling.)
|
|
|
|
> If you also used other datasets (e.g., MedQuAD), add them here with links and licenses.
|
|
|
|
## Intended use
|
|
- Educational / informational assistance for medical QA style prompts.
|
|
- Useful for summarization, explanation of concepts, and drafting answers that should be **verified**.
|
|
|
|
## Limitations & safety
|
|
- This model can **hallucinate** or provide incomplete/incorrect medical guidance.
|
|
- **Not a medical device**. Do not use for diagnosis, treatment decisions, or emergency situations.
|
|
- Always verify answers with reliable sources and qualified professionals.
|
|
|
|
## How to use
|
|
|
|
### Transformers (Python)
|
|
```python
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM
|
|
import torch
|
|
|
|
repo_id = "lgsantini1/qwen3-8b-medical"
|
|
|
|
tok = AutoTokenizer.from_pretrained(repo_id)
|
|
model = AutoModelForCausalLM.from_pretrained(repo_id, torch_dtype="auto", device_map="auto")
|
|
|
|
prompt = "Explain hypertension in simple terms."
|
|
inputs = tok(prompt, return_tensors="pt").to(model.device)
|
|
out = model.generate(**inputs, max_new_tokens=200)
|
|
print(tok.decode(out[0], skip_special_tokens=True))
|