---
license: cc-by-4.0
base_model: google/medgemma-1.5-4b-it
tags:
- medical
- silverguard
- qlora
- medgemma
---

# 🏥 SilverGuard-Adapter (V1.0 Impact Edition)

This is the **QLoRA Fine-tuned Adapter** for the **SilverGuard Project** (Google MedGemma Impact Challenge).
It is designed to work with `google/medgemma-1.5-4b-it` to detect medication errors in Taiwan's localized context.

### 🔧 Model Details
- **Base Model:** [google/medgemma-1.5-4b-it](https://huggingface.co/google/medgemma-1.5-4b-it)
- **Training Method:** QLoRA (4-bit quantization)
- **Task:** Vision-Language Medication Safety Checking
- **Developer:** Wang, Yuan-dao

### 🚀 How to use
```python
from peft import PeftModel, PeftConfig
from transformers import AutoModelForCausalLM

config = PeftConfig.from_pretrained("markwang941108/SilverGuard-Adapter-V1")
base_model = AutoModelForCausalLM.from_pretrained("google/medgemma-1.5-4b-it")
model = PeftModel.from_pretrained(base_model, "markwang941108/SilverGuard-Adapter-V1")