Start by loading the model and associated processor from a checkpoint on the Hugging Face Hub.