This layer represents a LSTM operation. More...

#include <LstmLayer.hpp>

Inheritance diagram for LstmLayer:

Collaboration diagram for LstmLayer:

Public Member Functions
virtual std::unique_ptr< IWorkload >	CreateWorkload (const IWorkloadFactory &factory) const override
	Makes a workload for the LSTM type.

LstmLayer *	Clone (Graph &graph) const override
	Creates a dynamically-allocated copy of this layer.

void	ValidateTensorShapesFromInputs () override
	Check if the input tensor shape(s) will lead to a valid configuration of LstmLayer.

std::vector< TensorShape >	InferOutputShapes (const std::vector< TensorShape > &inputShapes) const override
	By default returns inputShapes if the number of inputs are equal to number of outputs, otherwise infers the output shapes from given input shapes and layer properties.

void	ExecuteStrategy (IStrategy &strategy) const override
	Apply a visitor to this layer.

Public Member Functions inherited from LayerWithParameters< LstmDescriptor >
const LstmDescriptor &	GetParameters () const override
	If the layer has a descriptor return it.

void	SerializeLayerParameters (ParameterStringifyFunction &fn) const override
	Helper to serialize the layer parameters to string (currently used in DotSerializer and company).

Public Member Functions inherited from Layer
	Layer (unsigned int numInputSlots, unsigned int numOutputSlots, LayerType type, const char *name)

	Layer (unsigned int numInputSlots, unsigned int numOutputSlots, LayerType type, DataLayout layout, const char *name)

const std::string &	GetNameStr () const

const OutputHandler &	GetOutputHandler (unsigned int i=0) const

OutputHandler &	GetOutputHandler (unsigned int i=0)

ShapeInferenceMethod	GetShapeInferenceMethod () const

bool	GetAllowExpandedDims () const

const std::vector< InputSlot > &	GetInputSlots () const

const std::vector< OutputSlot > &	GetOutputSlots () const

std::vector< InputSlot >::iterator	BeginInputSlots ()

std::vector< InputSlot >::iterator	EndInputSlots ()

std::vector< OutputSlot >::iterator	BeginOutputSlots ()

std::vector< OutputSlot >::iterator	EndOutputSlots ()

bool	IsOutputUnconnected ()

void	ResetPriority () const

LayerPriority	GetPriority () const

LayerType	GetType () const override
	Returns the armnn::LayerType of this layer.

DataType	GetDataType () const

const BackendId &	GetBackendId () const

void	SetBackendId (const BackendId &id) override
	Set the backend of the IConnectableLayer.

virtual void	CreateTensorHandles (const TensorHandleFactoryRegistry &registry, const IWorkloadFactory &factory, const bool IsMemoryManaged=true)

void	VerifyLayerConnections (unsigned int expectedConnections, const CheckLocation &location) const

virtual void	ReleaseConstantData ()

template<typename Op >
void	OperateOnConstantTensors (Op op)

const char *	GetName () const override
	Returns the name of the layer.

unsigned int	GetNumInputSlots () const override
	Returns the number of connectable input slots.

unsigned int	GetNumOutputSlots () const override
	Returns the number of connectable output slots.

const InputSlot &	GetInputSlot (unsigned int index) const override
	Get a const input slot handle by slot index.

InputSlot &	GetInputSlot (unsigned int index) override
	Get the input slot handle by slot index.

const OutputSlot &	GetOutputSlot (unsigned int index=0) const override
	Get the const output slot handle by slot index.

OutputSlot &	GetOutputSlot (unsigned int index=0) override
	Get the output slot handle by slot index.

void	SetGuid (LayerGuid guid)

LayerGuid	GetGuid () const final
	Returns the unique id of the layer.

void	AddRelatedLayerName (const std::string layerName)

const std::list< std::string > &	GetRelatedLayerNames ()

virtual void	Reparent (Graph &dest, std::list< Layer * >::const_iterator iterator)=0

void	BackendSelectionHint (Optional< BackendId > backend) final
	Provide a hint for the optimizer as to which backend to prefer for this layer.

Optional< BackendId >	GetBackendHint () const

void	SetShapeInferenceMethod (ShapeInferenceMethod shapeInferenceMethod)

void	SetAllowExpandedDims (bool allowExpandedDims)

template<typename T >
std::shared_ptr< T >	GetAdditionalInformation () const

void	SetAdditionalInfoForObject (const AdditionalInfoObjectPtr &additionalInfo)

Public Attributes
LstmBasicParameters	m_BasicParameters

LstmOptCifgParameters	m_CifgParameters

LstmOptProjectionParameters	m_ProjectionParameters

LstmOptPeepholeParameters	m_PeepholeParameters

LstmOptLayerNormParameters	m_LayerNormParameters

Protected Member Functions
	LstmLayer (const LstmDescriptor &param, const char *name)
	Constructor to create a LstmLayer.

	~LstmLayer ()=default
	Default destructor.

Layer::ImmutableConstantTensors	GetConstantTensorsByRef () const override
	Retrieve the handles to the constant values stored by the layer.

Protected Member Functions inherited from LayerWithParameters< LstmDescriptor >
	LayerWithParameters (unsigned int numInputSlots, unsigned int numOutputSlots, LayerType type, const LstmDescriptor &param, const char *name)

	~LayerWithParameters ()=default

WorkloadInfo	PrepInfoAndDesc (QueueDescriptor &descriptor) const
	Helper function to reduce duplication in *LayerCreateWorkload.

Layer::ImmutableConstantTensors	GetConnectedConstantAsInputTensors () const

Protected Member Functions inherited from Layer
virtual	~Layer ()=default

template<typename QueueDescriptor >
void	CollectQueueDescriptorInputs (QueueDescriptor &descriptor, WorkloadInfo &info) const

template<typename QueueDescriptor >
void	CollectQueueDescriptorOutputs (QueueDescriptor &descriptor, WorkloadInfo &info) const

void	ValidateAndCopyShape (const TensorShape &outputShape, const TensorShape &inferredShape, const ShapeInferenceMethod shapeInferenceMethod, const std::string &layerName, const unsigned int outputSlotIndex=0)

void	VerifyShapeInferenceType (const TensorShape &outputShape, ShapeInferenceMethod shapeInferenceMethod)

template<typename QueueDescriptor >
WorkloadInfo	PrepInfoAndDesc (QueueDescriptor &descriptor) const
	Helper function to reduce duplication in *LayerCreateWorkload.

template<typename LayerType , typename ... Params>
LayerType *	CloneBase (Graph &graph, Params &&... params) const

virtual ConstantTensors	GetConstantTensorsByRef () override final

void	SetAdditionalInfo (QueueDescriptor &descriptor) const

Protected Member Functions inherited from IConnectableLayer
	~IConnectableLayer ()
	Objects are not deletable via the handle.

Additional Inherited Members
Public Types inherited from LayerWithParameters< LstmDescriptor >
using	DescriptorType

Public Types inherited from IConnectableLayer
using	ConstantTensors = std::vector<std::reference_wrapper<std::shared_ptr<ConstTensorHandle>>>

using	ImmutableConstantTensors = std::vector<std::reference_wrapper<const std::shared_ptr<ConstTensorHandle>>>

Protected Attributes inherited from LayerWithParameters< LstmDescriptor >
LstmDescriptor	m_Param
	The parameters for the layer (not including tensor-valued weights etc.).

Protected Attributes inherited from Layer
AdditionalInfoObjectPtr	m_AdditionalInfoObject

std::vector< OutputHandler >	m_OutputHandlers

ShapeInferenceMethod	m_ShapeInferenceMethod

Detailed Description

This layer represents a LSTM operation.

Definition at line 16 of file LstmLayer.hpp.

Constructor & Destructor Documentation

◆ LstmLayer()

LstmLayer	(	const LstmDescriptor &	param,
		const char *	name )

protected

Constructor to create a LstmLayer.

Parameters

[in]	param	LstmDescriptor to configure the lstm operation.
[in]	name	Optional name for the layer.

Definition at line 17 of file LstmLayer.cpp.

        : LayerWithParameters(3, 4, LayerType::Lstm, param, name)
{
}

◆ ~LstmLayer()

~LstmLayer ( )

protecteddefault

Default destructor.

Member Function Documentation

◆ Clone()

LstmLayer * Clone ( Graph & graph ) const

overridevirtual

Creates a dynamically-allocated copy of this layer.

Parameters

[in] graph The graph into which this layer is being cloned.

Implements Layer.

Definition at line 80 of file LstmLayer.cpp.

{
    auto layer = CloneBase<LstmLayer>(graph, m_Param, GetName());
 
    layer->m_BasicParameters.m_InputToForgetWeights = m_BasicParameters.m_InputToForgetWeights ?
            m_BasicParameters.m_InputToForgetWeights
                : nullptr;
    layer->m_BasicParameters.m_InputToCellWeights = m_BasicParameters.m_InputToCellWeights ?
            m_BasicParameters.m_InputToCellWeights : nullptr;
    layer->m_BasicParameters.m_InputToOutputWeights = m_BasicParameters.m_InputToOutputWeights ?
            m_BasicParameters.m_InputToOutputWeights : nullptr;
    layer->m_BasicParameters.m_RecurrentToForgetWeights = m_BasicParameters.m_RecurrentToForgetWeights ?
            m_BasicParameters.m_RecurrentToForgetWeights : nullptr;
    layer->m_BasicParameters.m_RecurrentToCellWeights = m_BasicParameters.m_RecurrentToCellWeights ?
            m_BasicParameters.m_RecurrentToCellWeights : nullptr;
    layer->m_BasicParameters.m_RecurrentToOutputWeights = m_BasicParameters.m_RecurrentToOutputWeights ?
            m_BasicParameters.m_RecurrentToOutputWeights : nullptr;
    layer->m_BasicParameters.m_ForgetGateBias = m_BasicParameters.m_ForgetGateBias ?
            m_BasicParameters.m_ForgetGateBias : nullptr;
    layer->m_BasicParameters.m_CellBias = m_BasicParameters.m_CellBias ?
            m_BasicParameters.m_CellBias : nullptr;
    layer->m_BasicParameters.m_OutputGateBias = m_BasicParameters.m_OutputGateBias ?
            m_BasicParameters.m_OutputGateBias : nullptr;
 
    if (!m_Param.m_CifgEnabled)
    {
        layer->m_CifgParameters.m_InputToInputWeights = m_CifgParameters.m_InputToInputWeights ?
                m_CifgParameters.m_InputToInputWeights : nullptr;
        layer->m_CifgParameters.m_RecurrentToInputWeights = m_CifgParameters.m_RecurrentToInputWeights ?
                m_CifgParameters.m_RecurrentToInputWeights : nullptr;
        layer->m_CifgParameters.m_InputGateBias = m_CifgParameters.m_InputGateBias ?
                m_CifgParameters.m_InputGateBias : nullptr;
    }
 
    if (m_Param.m_ProjectionEnabled)
    {
        layer->m_ProjectionParameters.m_ProjectionWeights = m_ProjectionParameters.m_ProjectionWeights ?
               m_ProjectionParameters.m_ProjectionWeights : nullptr;
        layer->m_ProjectionParameters.m_ProjectionBias = m_ProjectionParameters.m_ProjectionBias ?
               m_ProjectionParameters.m_ProjectionBias : nullptr;
    }
 
    if (m_Param.m_PeepholeEnabled)
    {
        if (!m_Param.m_CifgEnabled)
        {
            layer->m_PeepholeParameters.m_CellToInputWeights = m_PeepholeParameters.m_CellToInputWeights ?
                m_PeepholeParameters.m_CellToInputWeights : nullptr;
        }
        layer->m_PeepholeParameters.m_CellToForgetWeights = m_PeepholeParameters.m_CellToForgetWeights ?
               m_PeepholeParameters.m_CellToForgetWeights : nullptr;
        layer->m_PeepholeParameters.m_CellToOutputWeights = m_PeepholeParameters.m_CellToOutputWeights ?
               m_PeepholeParameters.m_CellToOutputWeights : nullptr;
    }
 
    if (m_Param.m_LayerNormEnabled)
    {
        layer->m_LayerNormParameters.m_InputLayerNormWeights = m_LayerNormParameters.m_InputLayerNormWeights ?
               m_LayerNormParameters.m_InputLayerNormWeights : nullptr;
        layer->m_LayerNormParameters.m_ForgetLayerNormWeights = m_LayerNormParameters.m_ForgetLayerNormWeights ?
               m_LayerNormParameters.m_ForgetLayerNormWeights : nullptr;
        layer->m_LayerNormParameters.m_CellLayerNormWeights = m_LayerNormParameters.m_CellLayerNormWeights ?
               m_LayerNormParameters.m_CellLayerNormWeights : nullptr;
        layer->m_LayerNormParameters.m_OutputLayerNormWeights = m_LayerNormParameters.m_OutputLayerNormWeights ?
               m_LayerNormParameters.m_OutputLayerNormWeights : nullptr;
    }
 
    return std::move(layer);
}

◆ CreateWorkload()

std::unique_ptr< IWorkload > CreateWorkload ( const IWorkloadFactory & factory ) const

overridevirtual

Makes a workload for the LSTM type.

Parameters

[in]	graph	The graph where this layer can be found.
[in]	factory	The workload factory which will create the workload.

Returns: A pointer to the created workload, or nullptr if not created.

Implements Layer.

Definition at line 22 of file LstmLayer.cpp.

{
    LstmQueueDescriptor descriptor;
 
    // Basic parameters
    descriptor.m_InputToForgetWeights = m_BasicParameters.m_InputToForgetWeights.get();
    descriptor.m_InputToCellWeights = m_BasicParameters.m_InputToCellWeights.get();
    descriptor.m_InputToOutputWeights = m_BasicParameters.m_InputToOutputWeights.get();
    descriptor.m_RecurrentToForgetWeights = m_BasicParameters.m_RecurrentToForgetWeights.get();
    descriptor.m_RecurrentToCellWeights = m_BasicParameters.m_RecurrentToCellWeights.get();
    descriptor.m_RecurrentToOutputWeights = m_BasicParameters.m_RecurrentToOutputWeights.get();
    descriptor.m_ForgetGateBias = m_BasicParameters.m_ForgetGateBias.get();
    descriptor.m_CellBias = m_BasicParameters.m_CellBias.get();
    descriptor.m_OutputGateBias = m_BasicParameters.m_OutputGateBias.get();
 
    // Cifg parameters
    if (!m_Param.m_CifgEnabled)
    {
        descriptor.m_InputToInputWeights = m_CifgParameters.m_InputToInputWeights.get();
        descriptor.m_RecurrentToInputWeights = m_CifgParameters.m_RecurrentToInputWeights.get();
        descriptor.m_InputGateBias = m_CifgParameters.m_InputGateBias.get();
    }
 
    // Projection parameters
    if (m_Param.m_ProjectionEnabled)
    {
        descriptor.m_ProjectionWeights = m_ProjectionParameters.m_ProjectionWeights.get();
        descriptor.m_ProjectionBias    = m_ProjectionParameters.m_ProjectionBias.get();
    }
 
    // Peephole parameters
    if (m_Param.m_PeepholeEnabled)
    {
        if (!m_Param.m_CifgEnabled)
        {
            descriptor.m_CellToInputWeights = m_PeepholeParameters.m_CellToInputWeights.get();
        }
        descriptor.m_CellToForgetWeights = m_PeepholeParameters.m_CellToForgetWeights.get();
        descriptor.m_CellToOutputWeights = m_PeepholeParameters.m_CellToOutputWeights.get();
    }
 
    // Layer normalisation parameters
    if(m_Param.m_LayerNormEnabled)
    {
        if (!m_Param.m_CifgEnabled)
        {
            descriptor.m_InputLayerNormWeights = m_LayerNormParameters.m_InputLayerNormWeights.get();
        }
        descriptor.m_ForgetLayerNormWeights = m_LayerNormParameters.m_ForgetLayerNormWeights.get();
        descriptor.m_CellLayerNormWeights = m_LayerNormParameters.m_CellLayerNormWeights.get();
        descriptor.m_OutputLayerNormWeights = m_LayerNormParameters.m_OutputLayerNormWeights.get();
    }
 
    SetAdditionalInfo(descriptor);
 
    return factory.CreateWorkload(LayerType::Lstm, descriptor, PrepInfoAndDesc(descriptor));
}

◆ ExecuteStrategy()

void ExecuteStrategy ( IStrategy & strategy ) const

overridevirtual

Apply a visitor to this layer.

Reimplemented from LayerWithParameters< LstmDescriptor >.

Definition at line 402 of file LstmLayer.cpp.

{
    std::vector<ConstTensor> constTensors;
 
    LstmDescriptor descriptor = GetParameters();
 
    ManagedConstTensorHandle managedInputToForgetWeights(m_BasicParameters.m_InputToForgetWeights);
    ManagedConstTensorHandle managedInputToCellWeights(m_BasicParameters.m_InputToCellWeights);
    ManagedConstTensorHandle managedInputToOutputWeights(m_BasicParameters.m_InputToOutputWeights);
    ManagedConstTensorHandle managedRecurrentToForgetWeights(m_BasicParameters.m_RecurrentToForgetWeights);
    ManagedConstTensorHandle managedRecurrentToCellWeights(m_BasicParameters.m_RecurrentToCellWeights);
    ManagedConstTensorHandle managedRecurrentToOutputWeights(m_BasicParameters.m_RecurrentToOutputWeights);
    ManagedConstTensorHandle managedForgetGateBias(m_BasicParameters.m_ForgetGateBias);
    ManagedConstTensorHandle managedCellBias(m_BasicParameters.m_CellBias);
    ManagedConstTensorHandle managedOutputGateBias(m_BasicParameters.m_OutputGateBias);
 
    // Cifg parameters
    ManagedConstTensorHandle managedInputToInputWeights(m_CifgParameters.m_InputToInputWeights);
    ManagedConstTensorHandle managedRecurrentToInputWeights(m_CifgParameters.m_RecurrentToInputWeights);
    ManagedConstTensorHandle managedInputGateBias(m_CifgParameters.m_InputGateBias);
 
    // Projection parameters
    ManagedConstTensorHandle managedProjectionWeights(m_ProjectionParameters.m_ProjectionWeights);
    ManagedConstTensorHandle managedProjectionBias(m_ProjectionParameters.m_ProjectionBias);
 
    // Peephole parameters
    ManagedConstTensorHandle managedCellToInputWeights(m_PeepholeParameters.m_CellToInputWeights);
    ManagedConstTensorHandle managedCellToForgetWeights(m_PeepholeParameters.m_CellToForgetWeights);
    ManagedConstTensorHandle managedCellToOutputWeights(m_PeepholeParameters.m_CellToOutputWeights);
 
    // Layer normalisation parameters
    ManagedConstTensorHandle managedInputLayerNormWeights(m_LayerNormParameters.m_InputLayerNormWeights);
    ManagedConstTensorHandle managedForgetLayerNormWeights(m_LayerNormParameters.m_ForgetLayerNormWeights);
    ManagedConstTensorHandle managedCellLayerNormWeights(m_LayerNormParameters.m_CellLayerNormWeights);
    ManagedConstTensorHandle managedOutputLayerNormWeights(m_LayerNormParameters.m_OutputLayerNormWeights);
 
    // First add mandatory/basic parameters
    if (m_BasicParameters.m_InputToForgetWeights != nullptr)
    {
        constTensors.emplace_back(ConstTensor(managedInputToForgetWeights.GetTensorInfo(),
                                              managedInputToForgetWeights.Map()));
    }
    if (m_BasicParameters.m_InputToCellWeights != nullptr)
    {
        constTensors.emplace_back(ConstTensor(managedInputToCellWeights.GetTensorInfo(),
                                              managedInputToCellWeights.Map()));
    }
    if (m_BasicParameters.m_InputToOutputWeights != nullptr)
    {
        constTensors.emplace_back(ConstTensor(managedInputToOutputWeights.GetTensorInfo(),
                                              managedInputToOutputWeights.Map()));
    }
    if (m_BasicParameters.m_RecurrentToForgetWeights != nullptr)
    {
        constTensors.emplace_back(ConstTensor(
                managedRecurrentToForgetWeights.GetTensorInfo(),
                managedRecurrentToForgetWeights.Map()));
    }
    if (m_BasicParameters.m_RecurrentToCellWeights != nullptr)
    {
        constTensors.emplace_back(ConstTensor(
                managedRecurrentToCellWeights.GetTensorInfo(),
                managedRecurrentToCellWeights.Map()));
    }
    if (m_BasicParameters.m_RecurrentToOutputWeights != nullptr)
    {
        constTensors.emplace_back(ConstTensor(
                managedRecurrentToOutputWeights.GetTensorInfo(),
                managedRecurrentToOutputWeights.Map()));
    }
    if (m_BasicParameters.m_ForgetGateBias != nullptr)
    {
        constTensors.emplace_back(ConstTensor(managedForgetGateBias.GetTensorInfo(),
                                              managedForgetGateBias.Map()));
    }
    if (m_BasicParameters.m_CellBias != nullptr)
    {
        constTensors.emplace_back(ConstTensor(managedCellBias.GetTensorInfo(),
                                              managedCellBias.Map()));
    }
    if (m_BasicParameters.m_OutputGateBias != nullptr)
    {
        constTensors.emplace_back(ConstTensor(managedOutputGateBias.GetTensorInfo(),
                                              managedOutputGateBias.Map()));
    }
 
    // Add cifg parameters
    if (!descriptor.m_CifgEnabled)
    {
        if (m_CifgParameters.m_InputToInputWeights != nullptr)
        {
            constTensors.emplace_back(ConstTensor(managedInputToInputWeights.GetTensorInfo(),
                                                  managedInputToInputWeights.Map()));
        }
        if (m_CifgParameters.m_RecurrentToInputWeights != nullptr)
        {
            constTensors.emplace_back(ConstTensor(
                    managedRecurrentToInputWeights.GetTensorInfo(),
                    managedRecurrentToInputWeights.Map()));
        }
        if (m_CifgParameters.m_InputGateBias != nullptr)
        {
            constTensors.emplace_back(ConstTensor(managedInputGateBias.GetTensorInfo(),
                                                  managedInputGateBias.Map()));
        }
    }
 
    // Add peephole parameters
    if (descriptor.m_PeepholeEnabled)
    {
        if (!descriptor.m_CifgEnabled)
        {
            if (m_PeepholeParameters.m_CellToInputWeights != nullptr)
            {
                constTensors.emplace_back(ConstTensor(managedCellToInputWeights.GetTensorInfo(),
                                                      managedCellToInputWeights.Map()));
            }
        }
        if (m_PeepholeParameters.m_CellToForgetWeights != nullptr)
        {
            constTensors.emplace_back(ConstTensor(managedCellToForgetWeights.GetTensorInfo(),
                                                  managedCellToForgetWeights.Map()));
        }
        if (m_PeepholeParameters.m_CellToOutputWeights != nullptr)
        {
            constTensors.emplace_back(ConstTensor(managedCellToOutputWeights.GetTensorInfo(),
                                                  managedCellToOutputWeights.Map()));
        }
    }
 
    // Add projection parameters
    if (descriptor.m_ProjectionEnabled)
    {
        if (m_ProjectionParameters.m_ProjectionWeights != nullptr)
        {
            constTensors.emplace_back(ConstTensor(managedProjectionWeights.GetTensorInfo(),
                                                  managedProjectionWeights.Map()));
        }
        if (m_ProjectionParameters.m_ProjectionBias != nullptr)
        {
            constTensors.emplace_back(ConstTensor(managedProjectionBias.GetTensorInfo(),
                                                  managedProjectionBias.Map()));
        }
    }
 
    // Add norm parameters
    if (descriptor.m_LayerNormEnabled)
    {
        if (!descriptor.m_CifgEnabled)
        {
            if (m_LayerNormParameters.m_InputLayerNormWeights != nullptr)
            {
                constTensors.emplace_back(ConstTensor(managedInputLayerNormWeights.GetTensorInfo(),
                                                      managedInputLayerNormWeights.Map()));
            }
        }
        if (m_LayerNormParameters.m_ForgetLayerNormWeights != nullptr)
        {
            constTensors.emplace_back(ConstTensor(managedForgetLayerNormWeights.GetTensorInfo(),
                                                  managedForgetLayerNormWeights.Map()));
        }
        if (m_LayerNormParameters.m_CellLayerNormWeights != nullptr)
        {
            constTensors.emplace_back(ConstTensor(managedCellLayerNormWeights.GetTensorInfo(),
                                                  managedCellLayerNormWeights.Map()));
        }
        if (m_LayerNormParameters.m_OutputLayerNormWeights != nullptr)
        {
            constTensors.emplace_back(ConstTensor(managedOutputLayerNormWeights.GetTensorInfo(),
                                                  managedOutputLayerNormWeights.Map()));
        }
    }
 
    strategy.ExecuteStrategy(this, GetParameters(), constTensors, GetName());
}

◆ GetConstantTensorsByRef()

Layer::ImmutableConstantTensors GetConstantTensorsByRef ( ) const

overrideprotectedvirtual

Retrieve the handles to the constant values stored by the layer.

Returns: A vector of the constant tensors stored by this layer.

Reimplemented from Layer.

Definition at line 368 of file LstmLayer.cpp.

{
    // For API stability DO NOT ALTER order and add new members to the end of vector
    return {m_BasicParameters.m_InputToForgetWeights,
            m_BasicParameters.m_InputToCellWeights,
            m_BasicParameters.m_InputToOutputWeights,
            m_BasicParameters.m_RecurrentToForgetWeights,
            m_BasicParameters.m_RecurrentToCellWeights,
            m_BasicParameters.m_RecurrentToOutputWeights,
            m_BasicParameters.m_ForgetGateBias,
            m_BasicParameters.m_CellBias,
            m_BasicParameters.m_OutputGateBias,
 
            // Cifg parameters
            m_CifgParameters.m_InputToInputWeights,
            m_CifgParameters.m_RecurrentToInputWeights,
            m_CifgParameters.m_InputGateBias,
 
            // Projection parameters
            m_ProjectionParameters.m_ProjectionWeights,
            m_ProjectionParameters.m_ProjectionBias,
 
            // Peephole parameters
            m_PeepholeParameters.m_CellToInputWeights,
            m_PeepholeParameters.m_CellToForgetWeights,
            m_PeepholeParameters.m_CellToOutputWeights,
 
            // Layer normalisation parameters
            m_LayerNormParameters.m_InputLayerNormWeights,
            m_LayerNormParameters.m_ForgetLayerNormWeights,
            m_LayerNormParameters.m_CellLayerNormWeights,
            m_LayerNormParameters.m_OutputLayerNormWeights};
}

◆ InferOutputShapes()

std::vector< TensorShape > InferOutputShapes ( const std::vector< TensorShape > & inputShapes ) const

overridevirtual

By default returns inputShapes if the number of inputs are equal to number of outputs, otherwise infers the output shapes from given input shapes and layer properties.

Parameters

[in] inputShapes The input shapes layer has.

Returns: A vector to the inferred output shape.

Reimplemented from Layer.

Definition at line 150 of file LstmLayer.cpp.

{
    if (inputShapes.size() != 3)
    {
        throw armnn::Exception("inputShapes' size is \"" + std::to_string(inputShapes.size()) +
                               "\" - should be \"3\".");
    }
 
    // Get input values for validation
    unsigned int batchSize = inputShapes[0][0];
    unsigned int outputSize = inputShapes[1][1];
    unsigned int numUnits = inputShapes[2][1];
 
    std::vector<TensorShape> outShapes;
    outShapes.push_back(TensorShape({batchSize, numUnits * (m_Param.m_CifgEnabled ? 3 : 4)}));
    outShapes.push_back(TensorShape({batchSize, outputSize}));
    outShapes.push_back(TensorShape({batchSize, numUnits}));
    outShapes.push_back(TensorShape({batchSize, outputSize}));
 
    return outShapes;
}

References LstmDescriptor::m_CifgEnabled, and LayerWithParameters< LstmDescriptor >::m_Param.

Referenced by LstmLayer::ValidateTensorShapesFromInputs().

◆ ValidateTensorShapesFromInputs()

void ValidateTensorShapesFromInputs ( )

overridevirtual

Check if the input tensor shape(s) will lead to a valid configuration of LstmLayer.

Parameters

[in] shapeInferenceMethod Indicates if output shape shall be overwritten or just validated.

Implements Layer.

Definition at line 172 of file LstmLayer.cpp.

{
    VerifyLayerConnections(3, CHECK_LOCATION());
 
    const TensorShape& outputShape = GetOutputSlot(0).GetTensorInfo().GetShape();
 
    VerifyShapeInferenceType(outputShape, m_ShapeInferenceMethod);
 
    auto inferredShapes = InferOutputShapes( {
        GetInputSlot(0).GetTensorInfo().GetShape(),
        GetInputSlot(1).GetTensorInfo().GetShape(),
        GetInputSlot(2).GetTensorInfo().GetShape()
    });
 
    if (inferredShapes.size() != 4)
    {
        throw armnn::Exception("inferredShapes has "
                               + std::to_string(inferredShapes.size()) +
                               " element(s) - should only have 4.");
    }
 
    // Check if the weights are nullptr
    if (!m_BasicParameters.m_InputToForgetWeights)
    {
        throw armnn::NullPointerException("LstmLayer: "
                                          "m_BasicParameters.m_InputToForgetWeights should not be null.");
    }
 
    if (!m_BasicParameters.m_InputToCellWeights)
    {
        throw armnn::NullPointerException("LstmLayer: "
                                          "m_BasicParameters.m_InputToCellWeights should not be null.");
    }
 
    if (!m_BasicParameters.m_InputToOutputWeights)
    {
        throw armnn::NullPointerException("LstmLayer: "
                                          "m_BasicParameters.m_InputToOutputWeights should not be null.");
    }
 
    if (!m_BasicParameters.m_RecurrentToForgetWeights)
    {
        throw armnn::NullPointerException("LstmLayer: "
                                           "m_BasicParameters.m_RecurrentToForgetWeights should not be null.");
    }
 
    if (!m_BasicParameters.m_RecurrentToCellWeights)
    {
        throw armnn::NullPointerException("LstmLayer: "
                                          "m_BasicParameters.m_RecurrentToCellWeights should not be null.");
    }
 
    if (!m_BasicParameters.m_RecurrentToOutputWeights)
    {
        throw armnn::NullPointerException("LstmLayer: "
                                          "m_BasicParameters.m_RecurrentToOutputWeights should not be null.");
    }
 
    if (!m_BasicParameters.m_ForgetGateBias)
    {
        throw armnn::NullPointerException("LstmLayer: "
                                          "m_BasicParameters.m_ForgetGateBias should not be null.");
    }
 
    if (!m_BasicParameters.m_CellBias)
    {
        throw armnn::NullPointerException("LstmLayer: "
                                          "m_BasicParameters.m_CellBias should not be null.");
    }
 
    if (!m_BasicParameters.m_OutputGateBias)
    {
        throw armnn::NullPointerException("LstmLayer: "
                                          "m_BasicParameters.m_OutputGateBias should not be null.");
    }
 
    if (!m_Param.m_CifgEnabled)
    {
        if (!m_CifgParameters.m_InputToInputWeights)
        {
            throw armnn::NullPointerException("LstmLayer: "
                                              "m_CifgParameters.m_InputToInputWeights should not be null.");
        }
 
        if (!m_CifgParameters.m_RecurrentToInputWeights)
        {
            throw armnn::NullPointerException("LstmLayer: "
                                              "m_CifgParameters.m_RecurrentToInputWeights should not be null.");
        }
 
        if (!m_CifgParameters.m_InputGateBias)
        {
            throw armnn::NullPointerException("LstmLayer: "
                                              "m_CifgParameters.m_InputGateBias should not be null.");
        }
 
        ValidateAndCopyShape(outputShape, inferredShapes[0], m_ShapeInferenceMethod, "LstmLayer");
    }
    else
    {
        if (m_CifgParameters.m_InputToInputWeights)
        {
            throw armnn::Exception("LstmLayer: "
                                   "m_CifgParameters.m_InputToInputWeights should not have a value "
                                   "when CIFG is enabled.");
        }
 
        if (m_CifgParameters.m_RecurrentToInputWeights)
        {
            throw armnn::Exception("LstmLayer: "
                                   "m_CifgParameters.m_RecurrentToInputWeights should not have a value "
                                   "when CIFG is enabled.");
        }
 
        if (m_CifgParameters.m_InputGateBias)
        {
            throw armnn::Exception("LstmLayer: "
                                   "m_CifgParameters.m_InputGateBias should not have a value "
                                   "when CIFG is enabled.");
        }
 
        ValidateAndCopyShape(outputShape, inferredShapes[0], m_ShapeInferenceMethod, "LstmLayer");
    }
 
    if (m_Param.m_ProjectionEnabled)
    {
        if (!m_ProjectionParameters.m_ProjectionWeights)
        {
            throw armnn::NullPointerException("LstmLayer: "
                                              "m_ProjectionParameters.m_ProjectionWeights should not be null.");
        }
    }
 
    if (m_Param.m_PeepholeEnabled)
    {
        if (!m_Param.m_CifgEnabled)
        {
            if (!m_PeepholeParameters.m_CellToInputWeights)
            {
                throw armnn::NullPointerException("LstmLayer: "
                                                  "m_PeepholeParameters.m_CellToInputWeights should not be null "
                                                  "when Peephole is enabled and CIFG is disabled.");
            }
        }
 
        if (!m_PeepholeParameters.m_CellToForgetWeights)
        {
            throw armnn::NullPointerException("LstmLayer: "
                                              "m_PeepholeParameters.m_CellToForgetWeights should not be null.");
        }
 
        if (!m_PeepholeParameters.m_CellToOutputWeights)
        {
            throw armnn::NullPointerException("LstmLayer: "
                                              "m_PeepholeParameters.m_CellToOutputWeights should not be null.");
        }
    }
 
    ValidateAndCopyShape(
            GetOutputSlot(1).GetTensorInfo().GetShape(), inferredShapes[1], m_ShapeInferenceMethod, "LstmLayer", 1);
    ValidateAndCopyShape(
            GetOutputSlot(2).GetTensorInfo().GetShape(), inferredShapes[2], m_ShapeInferenceMethod, "LstmLayer", 2);
    ValidateAndCopyShape(
            GetOutputSlot(3).GetTensorInfo().GetShape(), inferredShapes[3], m_ShapeInferenceMethod, "LstmLayer", 3);
 
    if (m_Param.m_LayerNormEnabled)
    {
        if(!m_Param.m_CifgEnabled)
        {
            if (!m_LayerNormParameters.m_InputLayerNormWeights)
            {
                throw armnn::NullPointerException("LstmLayer: "
                                                  "m_LayerNormParameters.m_inputLayerNormWeights should not be null.");
            }
        }
 
        if (!m_LayerNormParameters.m_ForgetLayerNormWeights)
        {
            throw armnn::NullPointerException("LstmLayer: "
                                              "m_LayerNormParameters.m_forgetLayerNormWeights should not be null.");
        }
 
        if (!m_LayerNormParameters.m_CellLayerNormWeights)
        {
            throw armnn::NullPointerException("LstmLayer: "
                                              "m_LayerNormParameters.m_cellLayerNormWeights should not be null.");
        }
 
        if (!m_LayerNormParameters.m_OutputLayerNormWeights)
        {
            throw armnn::NullPointerException("LstmLayer: "
                                              "m_LayerNormParameters.m_outputLayerNormWeights should not be null.");
        }
    }
}