Tensor Basics
The Tensor class is the core data structure in LeanPass. It wraps a NumPy array and tracks operations for automatic differentiation.
How Autodiff Works
LeanPass uses reverse-mode automatic differentiation (backpropagation) to compute gradients. Here's what that actually means.
Operations
LeanPass supports a focused set of tensor operations — enough to build and train neural networks, without the kitchen sink.
Activations
Non-linear activation functions are what give neural networks their expressive power. LeanPass includes the essential ones.