ReLu6 module

Applies the element-wise function:

Usage

nn_relu6(inplace = FALSE)

Arguments

inplace

can optionally do the operation in-place. Default: FALSE

Details

$$ \mbox{ReLU6}(x) = \min(\max(0,x), 6) $$

Shape

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

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

Examples

if (torch_is_installed()) {
m <- nn_relu6()
input <- torch_randn(2)
output <- m(input)
}