nn_log_sigmoid: LogSigmoid module

In torch: Tensors and Neural Networks with 'GPU' Acceleration

LogSigmoid module

Description

Applies the element-wise function:

\mbox{LogSigmoid}(x) = \log\left(\frac{ 1 }{ 1 + \exp(-x)}\right)

Usage

nn_log_sigmoid()

Shape

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

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

Examples

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

torch documentation built on May 29, 2024, 9:54 a.m.