nn_threshold: Threshold module

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

Threshold module

Description

Thresholds each element of the input Tensor.

Usage

nn_threshold(threshold, value, inplace = FALSE)

Arguments

threshold	The value to threshold at
value	The value to replace with
inplace	can optionally do the operation in-place. Default: FALSE

Details

Threshold is defined as:

y = \left\{ \begin{array}{ll} x, &\mbox{ if } x > \mbox{threshold} \\ \mbox{value}, &\mbox{ otherwise } \end{array} \right.

Shape

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

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

Examples

if (torch_is_installed()) {
m <- nn_threshold(0.1, 20)
input <- torch_randn(2)
output <- m(input)
}

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