Efficient Model Deployment: Running Large AI Models on Edge Devices