Feedforward Neural Network with PyTorch

In this assignment you will build and train a feedforward neural network (multi-layer perceptron) using PyTorch.

You will cover:

• Model definition: Define a multi-layer perceptron as a torch.nn.Module subclass, specifying the number of hidden layers, hidden units, and activation functions.
• Forward pass: Implement the forward method to pass input through each layer.
• Training loop: Write a complete training loop — compute the loss, call loss.backward(), and update parameters with an optimizer (SGD or Adam).
• Regularization: Experiment with dropout and weight decay to reduce overfitting.
• Evaluation: Track training and validation loss across epochs and report test accuracy.

Problems will be released closer to the due date.