---
license: llama3
base_model: unsloth/llama-3-8b-bnb-4bit-Instruct
tags:
- llama-3
- instruct
- medical
- pubmed
- fine-tuned
- unsloth
- lora
- text-generation
datasets:
- qiaojin/PubMedQA
language:
- en
pipeline_tag: text-generation
---

# 🩺 LLaMA-3-8B Medical Assistant

A compassionate medical AI assistant fine-tuned on PubMedQA dataset using Unsloth and LoRA.

## 🎯 Model Description

This model is designed to provide empathetic, caring medical guidance while encouraging professional medical consultation.

- **Base Model**: Llama-3-8B-Instruct
- **Fine-tuning**: LoRA on PubMedQA dataset
- **Use Case**: Educational and research purposes

## 🚀 Usage

```python
from unsloth import FastLanguageModel

model, tokenizer = FastLanguageModel.from_pretrained(
    model_name="chimochimo/llama3-8b-pubmedqa-medical",
    max_seq_length=2048,
    dtype=None,
    load_in_4bit=True,
)

FastLanguageModel.for_inference(model)

prompt = '''### Instruction:
You are a compassionate, understanding, and caring doctor speaking with a patient.

### Input:
What should I do if I have persistent headaches?

### Response:
'''

inputs = tokenizer([prompt], return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=128, use_cache=True)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```

## ⚠️ Disclaimer

This model is for educational and research purposes only. Always consult qualified healthcare professionals for medical advice.

## 🏷️ Training Details

- **Framework**: Unsloth + LoRA
- **Dataset**: PubMedQA
- **Training Steps**: 60
- **Batch Size**: 2
- **Learning Rate**: 2e-4