Modules
Elixir-native micrograd: a tiny scalar reverse-mode automatic differentiation engine plus the small neural-network library from the original project.
Small deterministic two-dimensional datasets for MicrogradEx demos.
A small supervised two-dimensional classification dataset.
The immutable result of a reverse-mode automatic differentiation pass.
Extracts scalar computation graphs from MicrogradEx.Value expressions.
Loss functions for small scalar MicrogradEx models.
Result of evaluating a supervised scalar loss.
Public facade for the tiny neural-network library.
A layer is a list of neurons with the same input width.
A multi-layer perceptron composed of Layer structs.
A scalar neuron: weighted sum, bias, and optional ReLU.
Converts MicrogradEx datasets and training runs into plain plotting rows.
Small immutable training loops for MicrogradEx models.
Result of a MicrogradEx training run.
A scalar value that remembers the expression graph that produced it.
A single local derivative from one operation output back to one parent value.
The immutable record stored in a value's computation graph.