Function shrink 

pub fn shrink<const D: usize, B>(
    tensor: Tensor<B, D>,
    lambda: f64,
    bias: f64,
) -> Tensor<B, D>
where
    B: Backend,

Applies the Shrink function element-wise.

shrink(x) = x - bias if x > lambda, x + bias if x < -lambda, 0 otherwise

Arguments

• lambda: the lambda value for the Shrink formulation.
• bias: the bias value for the Shrink formulation.