himanshutewatia771/gemma-4-medical-lora

⚕️ Gemma-4 Medical LoRA Adapter

This model is a QLoRA fine-tuned adapter for Gemma-4 (4.5B), trained on the VQA-RAD dataset for medical image visual question answering.

🧠 Model Details

• Base Model: Gemma-4 Vision
• Training Dataset: VQA-RAD
• Fine-Tuning Method: QLoRA
• Primary Use Case: Medical anomaly detection and conversational radiological workflows (Research purposes only).

🚀 How to Use

The model weights are zipped in my_medical_model.zip. Extract them and load using PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("google/gemma-4-4b-it")
model = PeftModel.from_pretrained(base_model, "./path_to_extracted_zip")