Fork me on GitHub
Arraymancer Technical reference Tutorial Spellbook (How-To's) Under the hood

Module nnp_softmax

Search:
Group by:

Procs

proc softmax[T](input: Tensor[T]): Tensor[T] {.
noInit
.}
For each sample in a tensor:
do an exponential normalization of each of its class features xi exp(xi) / ∑i exp(xi)
Input:
  • A tensor of shape [batch_size, number_of_classes]
Output:
  • A tensor of shape [batch_size, number_of_classes]
  Source Edit