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

#include <dnnl.hpp>

Collaboration diagram for dnnl::lstm_forward::desc:

[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, rnn_flags flags=rnn_flags::undef)
	Constructs a descriptor for an LSTM (with or without peephole) forward propagation primitive. 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, rnn_flags flags=rnn_flags::undef)
	Constructs a descriptor for an LSTM forward propagation primitive. More...

Detailed Description

Descriptor for an LSTM forward propagation primitive.

Examples:: cpu_rnn_inference_f32.cpp, cpu_rnn_inference_int8.cpp, and rnn_training_f32.cpp.

Constructor & Destructor Documentation

◆ desc() [1/2]

dnnl::lstm_forward::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,
		rnn_flags	flags = `rnn_flags::undef`
	)

inline

Constructs a descriptor for an LSTM (with or without peephole) forward propagation primitive.

The src_iter_desc, src_iter_c_desc, weights_peephole_desc, bias_desc, dst_iter_desc, and dst_iter_c_desc may point to a zero memory descriptor. This would then indicate that the LSTM forward propagation primitive should not use them and should default to zero values instead.

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

Inputs:

src_layer (dnnl::primitive_desc_base::src_desc (0))
src_iter (dnnl::primitive_desc_base::src_desc (1)), if used
src_iter_c (dnnl::primitive_desc_base::src_desc (2)), if used
weights_layer (dnnl::primitive_desc_base::weights_desc (0))
weights_iter (dnnl::primitive_desc_base::weights_desc (1))
weights_peephole (dnnl::primitive_desc_base::weights_desc (2)), if used
bias (dnnl::primitive_desc_base::weights_desc (2)), if used and LSTM is without peephole
bias (dnnl::primitive_desc_base::weights_desc (3)), if used and LSTM is with peephole

Outputs:

dst_layer (dnnl::primitive_desc_base::dst_desc (0))
dst_iter (dnnl::primitive_desc_base::dst_desc (1)), if used
dst_iter_c (dnnl::primitive_desc_base::dst_desc (2)), if used
workspace (dnnl::primitive_desc_base::workspace_desc (0)), if prop_kind equals dnnl::prop_kind::forward_training; must be queried for using dnnl::primitive_desc_base::query_md() after a corresponding primitive descriptor is created

Parameters

prop_kind	Propagation kind. Possible values are dnnl::prop_kind::forward_training, and dnnl::prop_kind::forward_inference.
direction	RNN direction. See dnnl::rnn_direction for more info.
src_layer_desc	Memory descriptor for the input vector.
src_iter_desc	Memory descriptor for the input recurrent hidden state vector.
src_iter_c_desc	Memory descriptor for the input recurrent cell state vector.
weights_layer_desc	Memory descriptor for the weights applied to the layer input.
weights_iter_desc	Memory descriptor for the weights applied to the recurrent input.
weights_peephole_desc	Memory descriptor for the weights applied to the cell states (according to the Peephole LSTM formula).
bias_desc	Bias memory descriptor.
dst_layer_desc	Memory descriptor for the output vector.
dst_iter_desc	Memory descriptor for the output recurrent hidden state vector.
dst_iter_c_desc	Memory descriptor for the output recurrent cell state vector.
flags	Unused.

◆ desc() [2/2]

dnnl::lstm_forward::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,
		rnn_flags	flags = `rnn_flags::undef`
	)

inline

Constructs a descriptor for an LSTM forward propagation primitive.

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

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