XOR Classification
The "hello world" of neural networks — solving the XOR problem, which a single linear layer cannot do.
Binary Classification
Classifying data into one of two categories using a neural network with sigmoid output.
Multi-Class Classification
Classifying data into more than two categories using softmax output and cross-entropy loss.
Regression with Neural Networks
Using a multi-layer network for non-linear regression — fitting a curve, not just a line.
MNIST Digit Classification
Training a neural network to recognize handwritten digits from the MNIST dataset.
Gradient Checking
Verifying that your autodiff gradients are correct using finite difference approximation.
Custom Layers
Extending LeanPass with your own layer types by subclassing Module.