Softplus module
Applies the element-wise function:
\mbox{Softplus}(x) = \frac{1}{β} * \log(1 + \exp(β * x))
nn_softplus(beta = 1, threshold = 20)
beta |
the β value for the Softplus formulation. Default: 1 |
threshold |
values above this revert to a linear function. Default: 20 |
SoftPlus is a smooth approximation to the ReLU function and can be used to constrain the output of a machine to always be positive. For numerical stability the implementation reverts to the linear function when input \times β > threshold.
Input: (N, *) where * means, any number of additional
dimensions
Output: (N, *), same shape as the input
if (torch_is_installed()) {
m <- nn_softplus()
input <- torch_randn(2)
output <- m(input)
}Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.