torch (version 0.8.1)

nn_softshrink: Softshrink module

Description

Applies the soft shrinkage function elementwise:

Usage

nn_softshrink(lambd = 0.5)

Arguments

lambd

the \(\lambda\) (must be no less than zero) value for the Softshrink formulation. Default: 0.5

Shape

• Input: \((N, *)\) where * means, any number of additional dimensions

• Output: \((N, *)\), same shape as the input

Details

$$ \mbox{SoftShrinkage}(x) = \left\{ \begin{array}{ll} x - \lambda, & \mbox{ if } x > \lambda \\ x + \lambda, & \mbox{ if } x < -\lambda \\ 0, & \mbox{ otherwise } \end{array} \right. $$

Examples

if (torch_is_installed()) {
m <- nn_softshrink()
input <- torch_randn(2)
output <- m(input)
}