Deep Neural Network Library (DNNL)  1.3.0
Performance library for Deep Learning
Public Member Functions | List of all members
dnnl::lstm_backward::desc Struct Reference

Descriptor for an LSTM backward propagation primitive. More...

#include <dnnl.hpp>

Collaboration diagram for dnnl::lstm_backward::desc:
Collaboration graph
[legend]

Public Member Functions

 desc (prop_kind prop_kind, rnn_direction direction, const memory::desc &src_layer_desc, const memory::desc &src_iter_desc, const memory::desc &src_iter_c_desc, const memory::desc &weights_layer_desc, const memory::desc &weights_iter_desc, const memory::desc &weights_peephole_desc, const memory::desc &bias_desc, const memory::desc &dst_layer_desc, const memory::desc &dst_iter_desc, const memory::desc &dst_iter_c_desc, const memory::desc &diff_src_layer_desc, const memory::desc &diff_src_iter_desc, const memory::desc &diff_src_iter_c_desc, const memory::desc &diff_weights_layer_desc, const memory::desc &diff_weights_iter_desc, const memory::desc &diff_weights_peephole_desc, const memory::desc &diff_bias_desc, const memory::desc &diff_dst_layer_desc, const memory::desc &diff_dst_iter_desc, const memory::desc &diff_dst_iter_c_desc, rnn_flags flags=rnn_flags::undef)
 Constructs an LSTM (with or without peephole) descriptor for backward propagation using prop_kind, direction, and memory descriptors. More...
 
 desc (prop_kind prop_kind, rnn_direction direction, const memory::desc &src_layer_desc, const memory::desc &src_iter_desc, const memory::desc &src_iter_c_desc, const memory::desc &weights_layer_desc, const memory::desc &weights_iter_desc, const memory::desc &bias_desc, const memory::desc &dst_layer_desc, const memory::desc &dst_iter_desc, const memory::desc &dst_iter_c_desc, const memory::desc &diff_src_layer_desc, const memory::desc &diff_src_iter_desc, const memory::desc &diff_src_iter_c_desc, const memory::desc &diff_weights_layer_desc, const memory::desc &diff_weights_iter_desc, const memory::desc &diff_bias_desc, const memory::desc &diff_dst_layer_desc, const memory::desc &diff_dst_iter_desc, const memory::desc &diff_dst_iter_c_desc, rnn_flags flags=rnn_flags::undef)
 Constructs an LSTM descriptor for backward propagation using prop_kind, direction, and memory descriptors. More...
 

Detailed Description

Descriptor for an LSTM backward propagation primitive.

Examples:
rnn_training_f32.cpp.

Constructor & Destructor Documentation

◆ desc() [1/2]

dnnl::lstm_backward::desc::desc ( prop_kind  prop_kind,
rnn_direction  direction,
const memory::desc src_layer_desc,
const memory::desc src_iter_desc,
const memory::desc src_iter_c_desc,
const memory::desc weights_layer_desc,
const memory::desc weights_iter_desc,
const memory::desc weights_peephole_desc,
const memory::desc bias_desc,
const memory::desc dst_layer_desc,
const memory::desc dst_iter_desc,
const memory::desc dst_iter_c_desc,
const memory::desc diff_src_layer_desc,
const memory::desc diff_src_iter_desc,
const memory::desc diff_src_iter_c_desc,
const memory::desc diff_weights_layer_desc,
const memory::desc diff_weights_iter_desc,
const memory::desc diff_weights_peephole_desc,
const memory::desc diff_bias_desc,
const memory::desc diff_dst_layer_desc,
const memory::desc diff_dst_iter_desc,
const memory::desc diff_dst_iter_c_desc,
rnn_flags  flags = rnn_flags::undef 
)
inline

Constructs an LSTM (with or without peephole) descriptor for backward propagation using prop_kind, direction, and memory descriptors.

The src_iter_desc together with diff_iter_desc, src_iter_c_desc together with src_iter_c_desc, weights_peephole_desc together with diff_weights_peephole_desc, bias_desc together with diff_bias_desc, dst_iter_desc together with diff_dst_iter_desc, and dst_iter_c_desc together with diff_dst_iter_c_desc, may point to a zero memory descriptor. This would then indicate that the LSTM backward propagation primitive should not use them and should default to zero values instead.

Note
All memory descriptors can be initialized with dnnl::memory::format_tag::any value of format_tag.

Inputs:

Outputs:

Parameters
prop_kindPropagation kind. Must be dnnl::prop_kind::backward.
directionRNN direction. See dnnl::rnn_direction for more info.
src_layer_descMemory descriptor for the input vector.
src_iter_descMemory descriptor for the input recurrent hidden state vector.
src_iter_c_descMemory descriptor for the input recurrent cell state vector.
weights_layer_descMemory descriptor for the weights applied to the layer input.
weights_iter_descMemory descriptor for the weights applied to the recurrent input.
weights_peephole_descMemory descriptor for the weights applied to the cell states (according to the Peephole LSTM formula).
bias_descBias memory descriptor.
dst_layer_descMemory descriptor for the output vector.
dst_iter_descMemory descriptor for the output recurrent hidden state vector.
dst_iter_c_descMemory descriptor for the output recurrent cell state vector.
diff_src_layer_descMemory descriptor for the diff of input vector.
diff_src_iter_descMemory descriptor for the diff of input recurrent hidden state vector.
diff_src_iter_c_descMemory descriptor for the diff of input recurrent cell state vector.
diff_weights_layer_descMemory descriptor for the diff of weights applied to the layer input.
diff_weights_iter_descMemory descriptor for the diff of weights applied to the recurrent input.
diff_weights_peephole_descMemory descriptor for the diff of weights applied to the cell states (according to the Peephole LSTM formula).
diff_bias_descDiff bias memory descriptor.
diff_dst_layer_descMemory descriptor for the diff of output vector.
diff_dst_iter_descMemory descriptor for the diff of output recurrent hidden state vector.
diff_dst_iter_c_descMemory descriptor for the diff of output recurrent cell state vector.
flagsUnused.

◆ desc() [2/2]

dnnl::lstm_backward::desc::desc ( prop_kind  prop_kind,
rnn_direction  direction,
const memory::desc src_layer_desc,
const memory::desc src_iter_desc,
const memory::desc src_iter_c_desc,
const memory::desc weights_layer_desc,
const memory::desc weights_iter_desc,
const memory::desc bias_desc,
const memory::desc dst_layer_desc,
const memory::desc dst_iter_desc,
const memory::desc dst_iter_c_desc,
const memory::desc diff_src_layer_desc,
const memory::desc diff_src_iter_desc,
const memory::desc diff_src_iter_c_desc,
const memory::desc diff_weights_layer_desc,
const memory::desc diff_weights_iter_desc,
const memory::desc diff_bias_desc,
const memory::desc diff_dst_layer_desc,
const memory::desc diff_dst_iter_desc,
const memory::desc diff_dst_iter_c_desc,
rnn_flags  flags = rnn_flags::undef 
)
inline

Constructs an LSTM descriptor for backward propagation using prop_kind, direction, and memory descriptors.

The src_iter_desc together with diff_iter_desc, src_iter_c_desc together with src_iter_c_desc, bias_desc together with diff_bias_desc, dst_iter_desc together with diff_dst_iter_desc, and dst_iter_c_desc together with diff_dst_iter_c_desc, may point to a zero memory descriptor. This would then indicate that the LSTM backward propagation primitive should not use them and should default to zero values instead.

Note
All memory descriptors can be initialized with dnnl::memory::format_tag::any value of format_tag.

Inputs:

Outputs:

Parameters
prop_kindPropagation kind. Must be dnnl::prop_kind::backward.
directionRNN direction. See dnnl::rnn_direction for more info.
src_layer_descMemory descriptor for the input vector.
src_iter_descMemory descriptor for the input recurrent hidden state vector.
src_iter_c_descMemory descriptor for the input recurrent cell state vector.
weights_layer_descMemory descriptor for the weights applied to the layer input.
weights_iter_descMemory descriptor for the weights applied to the recurrent input.
bias_descBias memory descriptor.
dst_layer_descMemory descriptor for the output vector.
dst_iter_descMemory descriptor for the output recurrent hidden state vector.
dst_iter_c_descMemory descriptor for the output recurrent cell state vector.
diff_src_layer_descMemory descriptor for the diff of input vector.
diff_src_iter_descMemory descriptor for the diff of input recurrent hidden state vector.
diff_src_iter_c_descMemory descriptor for the diff of input recurrent cell state vector.
diff_weights_layer_descMemory descriptor for the diff of weights applied to the layer input.
diff_weights_iter_descMemory descriptor for the diff of weights applied to the recurrent input.
diff_bias_descDiff bias memory descriptor.
diff_dst_layer_descMemory descriptor for the diff of output vector.
diff_dst_iter_descMemory descriptor for the diff of output recurrent hidden state vector.
diff_dst_iter_c_descMemory descriptor for the diff of output recurrent cell state vector.
flagsUnused.

The documentation for this struct was generated from the following file: