MNIST Sample Application

This application explores the classic MNIST handwritten-digit dataset in the context of AI model training and explainability. It pairs trame with XAITK saliency methods so you can inspect which parts of an input image drove a model's prediction, with optional GPU acceleration for training.