Sparsemax
Applies the SparseMax activation.
nnf_contrib_sparsemax(input, dim = -1)
input
: the input tensordim
: The dimension over which to apply the sparsemax function. (-1)The SparseMax activation is described in 'From Softmax to Sparsemax: A Sparse Model of Attention and Multi-Label Classification'
The implementation is based on aced125/sparsemax
Useful links