Trait neuronika::Gradient

pub trait Gradient {
    type Dim: Dimension;
    fn gradient(&self) -> Ref<'_, Array<f32, Self::Dim>>;
    fn gradient_mut(&self) -> RefMut<'_, Array<f32, Self::Dim>>;
}

Gradient representation.

This trait is implemented by all the internal backward components of VarDiff.

It provides the .gradient() method that is used to get a Ref to the data stored inside the node.

Associated Types

type Dim: Dimension

The gradient’s dimensionality.

Required methods

fn gradient(&self) -> Ref<'_, Array<f32, Self::Dim>>

Returns an immutable reference to the gradient inside self.

fn gradient_mut(&self) -> RefMut<'_, Array<f32, Self::Dim>>

Returns a mutable reference to the gradient inside self.

Implementations on Foreign Types

impl<T, D> Gradient for Rc<T> where
    T: Gradient<Dim = D>,
    D: Dimension, 

type Dim = D

fn gradient(&self) -> Ref<'_, Array<f32, Self::Dim>>

fn gradient_mut(&self) -> RefMut<'_, Array<f32, Self::Dim>>

Implementors