Struct Model

pub struct Model<'a> { /* private fields */ }

A neural network model, containing the layers of the model, and the outputs.

Implementations

impl<'a> Model<'a>

pub fn new( layers: Vec<&'a mut dyn Layer>, optimizer: &'a dyn Optimizer, cost: &'a CostFunction, ) -> Model<'a>

Constructs a new model given the layers.

pub fn forward(&mut self, input: Array) -> Array

Computes the forward pass of a model. The input should have the dimensions batch size by input size.

pub fn backward(&mut self, target: Array) -> Float

Computes the backward pass of a model, and updates parameters.

pub fn update(&mut self)

Updates all parameters of the model.