This code classifies Fashion MNIST images using PyTorch.

It defines and trains five models: one-layer, two-layer, three-layer neural network, a CNN, and a two-layer sigmoid network.

The dataset is preprocessed, normalized, and split into training, validation, and test sets. Each model is trained, evaluated, and compared using metrics like accuracy, loss, confusion matrices, and visualizations.