tensorflow :: ops :: ApplyProximalAdagrad

#include <training_ops.h>

Perbarui '* var' dan '* akumulasi' sesuai dengan FOBOS dengan kecepatan pemelajaran Adagrad.

Ringkasan

akumulasi + = grad * grad prox_v = var - lr * grad * (1 / sqrt (akum)) var = tanda (prox_v) / (1 + lr * l2) * max {| prox_v | -lr * l1,0}

Argumen:

  • scope: Objek Scope
  • var: Harus dari Variabel ().
  • akumulasi: Harus dari Variabel ().
  • lr: Faktor penskalaan. Harus skalar.
  • l1: regularisasi L1. Harus skalar.
  • l2: regularisasi L2. Harus skalar.
  • grad: Gradien.

Atribut opsional (lihat Attrs ):

  • use_locking: Jika True, pembaruan tensor var dan akumulator akan dilindungi oleh kunci; jika tidak, perilakunya tidak ditentukan, tetapi mungkin menunjukkan lebih sedikit perselisihan.

Pengembalian:

  • Output : Sama seperti "var".

Pembuat dan Penghancur

ApplyProximalAdagrad (const :: tensorflow::Scope & scope, :: tensorflow::Input var, :: tensorflow::Input accum, :: tensorflow::Input lr, :: tensorflow::Input l1, :: tensorflow::Input l2, :: tensorflow::Input grad)
ApplyProximalAdagrad (const :: tensorflow::Scope & scope, :: tensorflow::Input var, :: tensorflow::Input accum, :: tensorflow::Input lr, :: tensorflow::Input l1, :: tensorflow::Input l2, :: tensorflow::Input grad, const ApplyProximalAdagrad::Attrs & attrs)

Atribut publik

operation
out

Fungsi publik

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

Fungsi statis publik

UseLocking (bool x)

Structs

tensorflow :: ops :: ApplyProximalAdagrad :: Attrs

Penyetel atribut opsional untuk ApplyProximalAdagrad .

Atribut publik

operasi

Operation operation

di luar

::tensorflow::Output out

Fungsi publik

ApplyProximalAdagrad

 ApplyProximalAdagrad(
  const ::tensorflow::Scope & scope,
  ::tensorflow::Input var,
  ::tensorflow::Input accum,
  ::tensorflow::Input lr,
  ::tensorflow::Input l1,
  ::tensorflow::Input l2,
  ::tensorflow::Input grad
)

ApplyProximalAdagrad

 ApplyProximalAdagrad(
  const ::tensorflow::Scope & scope,
  ::tensorflow::Input var,
  ::tensorflow::Input accum,
  ::tensorflow::Input lr,
  ::tensorflow::Input l1,
  ::tensorflow::Input l2,
  ::tensorflow::Input grad,
  const ApplyProximalAdagrad::Attrs & attrs
)

simpul

::tensorflow::Node * node() const

operator :: tensorflow :: Input

 operator::tensorflow::Input() const

operator :: tensorflow :: Keluaran

 operator::tensorflow::Output() const

Fungsi statis publik

UseLocking

Attrs UseLocking(
  bool x
)