Google I/O'yu ayarladığınız için teşekkür ederiz. İsteğe bağlı olarak tüm oturumları görüntüleyin İsteğe bağlı olarak izleyin

tensorflow :: ops :: MatrixDiag

#include <array_ops.h>

Verilen toplu çapraz değerlere sahip bir toplu çapraz tensör döndürür.

Özet

Bir Verilen diagonal , bu işlem bir ile tensörünü döndüren diagonal sıfır gelerek ve her şeyi. Köşegen şu şekilde hesaplanır:

Varsayalım diagonal sahip k boyutları [I, J, K, ..., N] , daha sonra çıkış sıralaması bir tensör olan k+1 boyutları, [I, J, K, ..., N, N-] 'burada:

output[i, j, k, ..., m, n] = 1{m=n} * diagonal[i, j, k, ..., n] .

Örneğin:

# 'diagonal' is [[1, 2, 3, 4], [5, 6, 7, 8]]

and diagonal.shape = (2, 4)

tf.matrix_diag(diagonal) ==> [[[1, 0, 0, 0]
                                     [0, 2, 0, 0]
                                     [0, 0, 3, 0]
                                     [0, 0, 0, 4]],
                                    [[5, 0, 0, 0]
                                     [0, 6, 0, 0]
                                     [0, 0, 7, 0]
                                     [0, 0, 0, 8]]]

which has shape (2, 4, 4)

Argümanlar:

  • kapsam: Bir Scope nesnesi
  • diyagonal: Derece k , burada k >= 1 .

İadeler:

  • Output : output.shape = diagonal.shape + [diagonal.shape[-1]] ile output.shape = diagonal.shape + [diagonal.shape[-1]] k+1 .

Yapıcılar ve Yıkıcılar

MatrixDiag (const :: tensorflow::Scope & scope, :: tensorflow::Input diagonal)

Genel özellikler

operation
output

Kamusal işlevler

node () const
::tensorflow::Node *
operator::tensorflow::Input () const
operator::tensorflow::Output () const

Genel özellikler

operasyon

Operation operation

çıktı

::tensorflow::Output output

Kamusal işlevler

MatrixDiag

 MatrixDiag(
  const ::tensorflow::Scope & scope,
  ::tensorflow::Input diagonal
)

düğüm

::tensorflow::Node * node() const 

operator :: tensorflow :: Girdi

 operator::tensorflow::Input() const 

operator :: tensorflow :: Çıktı

 operator::tensorflow::Output() const