Trait Forward 

pub trait Forward {
    // Required method
    fn forward(&self, input: &Tensor) -> Tensor;
}

A feed-forward function of a tensor.

This is roughtly equivalent to PyTorch’s module class except it is not treated as the base interface of all modules because not all modules implement the batch Tensor -> Tensor transformation.

Required Methods

fn forward(&self, input: &Tensor) -> Tensor

Apply a batch feed-forward transformation to a tensor.

Applies a vector-to-vector map on the last dimension of the input tensor, replicated over all other dimensions.

• Input shape: [BATCH_SHAPE.., INPUT_DIM].
• Output shape: [BATCH_SHAPE.., OUTPUT_DIM].

Implementations on Foreign Types

impl<M: Forward> Forward for [M]

fn forward(&self, input: &Tensor) -> Tensor

impl<M: Forward, const N: usize> Forward for [M; N]

fn forward(&self, input: &Tensor) -> Tensor

impl<T: Forward + ?Sized> Forward for &T

fn forward(&self, input: &Tensor) -> Tensor

impl<T: Forward + ?Sized> Forward for Box<T>

fn forward(&self, input: &Tensor) -> Tensor

Implementors

impl Forward for Activation

impl Forward for Linear

impl Forward for Mlp

impl<A, B> Forward for Chain<A, B>

where A: Forward, B: Forward,

impl<T, F> Forward for BatchMap<T, F>

where T: Forward, F: Fn(Tensor) -> Tensor,