Struct neuronika::nn::Linear

pub struct Linear {
    pub weight: Learnable<Ix2>,
    pub bias: Learnable<Ix1>,
}

Applies a linear transformation to the incoming data.

ʏ = xAᵀ + b

Fields

weight: Learnable<Ix2>

bias: Learnable<Ix1>

Implementations

impl Linear

pub fn new(in_features: usize, out_features: usize) -> Self

Creates a linear layer.

Arguments

• in_features – size of each input sample.

• out_features – size of each output sample.

The learnable weight of the layer is of shape (out_features, in_features). The learnable bias of the layer is of shape out_features.

The values for both the weight and bias are initialized from U(-k, k) where k = (1. / in_features as f32).sqrt().

pub fn forward<I, T, U>(
    &self,
    input: I
) -> VarDiff<impl Data<Dim = Ix2> + Forward, impl Gradient<Dim = Ix2> + Overwrite + Backward> where
    I: MatMatMulT<Learnable<Ix2>>,
    I::Output: Into<VarDiff<T, U>>,
    T: Data<Dim = Ix2>,
    U: Gradient<Dim = Ix2> + Overwrite, 

Applies the linear transformation y = xA^T + b to the incoming data.

Arguments

data - a variable of shape (N, in_features), the output’s shape will be (N, out_features).

Trait Implementations

impl Register for Linear

fn register_params(&self, params: &mut Vec<RawParam>)

Registers the weight and the bias of this Linear instance.

fn register_status(&mut self, _: Rc<Cell<bool>>)

Register self’s status to the model’s status state status.