#include <nnet-simple-component.h>

Inheritance diagram for LinearComponent:

Collaboration diagram for LinearComponent:

[legend]

Public Member Functions
virtual int32	InputDim () const
	Returns input-dimension of this component. More...

virtual int32	OutputDim () const
	Returns output-dimension of this component. More...

virtual std::string	Type () const
	Returns a string such as "SigmoidComponent", describing the type of the object. More...

virtual int32	Properties () const
	Return bitmask of the component's properties. More...

virtual void *	Propagate (const ComponentPrecomputedIndexes indexes, const CuMatrixBase< BaseFloat > &in, CuMatrixBase< BaseFloat > out) const
	Propagate function. More...

virtual void	Backprop (const std::string &debug_info, const ComponentPrecomputedIndexes indexes, const CuMatrixBase< BaseFloat > &in_value, const CuMatrixBase< BaseFloat > &, const CuMatrixBase< BaseFloat > &out_deriv, void memo, Component to_update, CuMatrixBase< BaseFloat > in_deriv) const
	Backprop function; depending on which of the arguments 'to_update' and 'in_deriv' are non-NULL, this can compute input-data derivatives and/or perform model update. More...

virtual void	Read (std::istream &is, bool binary)
	Read function (used after we know the type of the Component); accepts input that is missing the token that describes the component type, in case it has already been consumed. More...

virtual void	Write (std::ostream &os, bool binary) const
	Write component to stream. More...

	LinearComponent ()

void	InitFromConfig (ConfigLine *cfl)
	Initialize, from a ConfigLine object. More...

virtual std::string	Info () const
	Returns some text-form information about this component, for diagnostics. More...

virtual Component *	Copy () const
	Copies component (deep copy). More...

virtual void	Scale (BaseFloat scale)
	This virtual function when called on – an UpdatableComponent scales the parameters by "scale" when called by an UpdatableComponent. More...

virtual void	Add (BaseFloat alpha, const Component &other)
	This virtual function when called by – an UpdatableComponent adds the parameters of another updatable component, times some constant, to the current parameters. More...

virtual void	PerturbParams (BaseFloat stddev)
	This function is to be used in testing. More...

virtual BaseFloat	DotProduct (const UpdatableComponent &other) const
	Computes dot-product between parameters of two instances of a Component. More...

virtual int32	NumParameters () const
	The following new virtual function returns the total dimension of the parameters in this class. More...

virtual void	Vectorize (VectorBase< BaseFloat > *params) const
	Turns the parameters into vector form. More...

virtual void	UnVectorize (const VectorBase< BaseFloat > &params)
	Converts the parameters from vector form. More...

virtual void	FreezeNaturalGradient (bool freeze)
	freezes/unfreezes NaturalGradient updates, if applicable (to be overriden by components that use Natural Gradient). More...

virtual void	ConsolidateMemory ()
	This virtual function relates to memory management, and avoiding fragmentation. More...

	LinearComponent (const LinearComponent &other)

	LinearComponent (const CuMatrix< BaseFloat > &params)

BaseFloat	OrthonormalConstraint () const

CuMatrixBase< BaseFloat > &	Params ()

const CuMatrixBase< BaseFloat > &	Params () const

Public Member Functions inherited from UpdatableComponent
	UpdatableComponent (const UpdatableComponent &other)

	UpdatableComponent ()

virtual	~UpdatableComponent ()

virtual void	SetUnderlyingLearningRate (BaseFloat lrate)
	Sets the learning rate of gradient descent- gets multiplied by learning_rate_factor_. More...

virtual void	SetActualLearningRate (BaseFloat lrate)
	Sets the learning rate directly, bypassing learning_rate_factor_. More...

virtual void	SetAsGradient ()
	Sets is_gradient_ to true and sets learning_rate_ to 1, ignoring learning_rate_factor_. More...

virtual BaseFloat	LearningRateFactor ()

virtual void	SetLearningRateFactor (BaseFloat lrate_factor)

void	SetUpdatableConfigs (const UpdatableComponent &other)

BaseFloat	LearningRate () const
	Gets the learning rate to be used in gradient descent. More...

BaseFloat	MaxChange () const
	Returns the per-component max-change value, which is interpreted as the maximum change (in l2 norm) in parameters that is allowed per minibatch for this component. More...

void	SetMaxChange (BaseFloat max_change)

BaseFloat	L2Regularization () const
	Returns the l2 regularization constant, which may be set in any updatable component (usually from the config file). More...

void	SetL2Regularization (BaseFloat a)

Public Member Functions inherited from Component
virtual void	StoreStats (const CuMatrixBase< BaseFloat > &in_value, const CuMatrixBase< BaseFloat > &out_value, void *memo)
	This function may store stats on average activation values, and for some component types, the average value of the derivative of the nonlinearity. More...

virtual void	ZeroStats ()
	Components that provide an implementation of StoreStats should also provide an implementation of ZeroStats(), to set those stats to zero. More...

virtual void	GetInputIndexes (const MiscComputationInfo &misc_info, const Index &output_index, std::vector< Index > *desired_indexes) const
	This function only does something interesting for non-simple Components. More...

virtual bool	IsComputable (const MiscComputationInfo &misc_info, const Index &output_index, const IndexSet &input_index_set, std::vector< Index > *used_inputs) const
	This function only does something interesting for non-simple Components, and it exists to make it possible to manage optionally-required inputs. More...

virtual void	ReorderIndexes (std::vector< Index > input_indexes, std::vector< Index > output_indexes) const
	This function only does something interesting for non-simple Components. More...

virtual ComponentPrecomputedIndexes *	PrecomputeIndexes (const MiscComputationInfo &misc_info, const std::vector< Index > &input_indexes, const std::vector< Index > &output_indexes, bool need_backprop) const
	This function must return NULL for simple Components. More...

virtual void	DeleteMemo (void *memo) const
	This virtual function only needs to be overwritten by Components that return a non-NULL memo from their Propagate() function. More...

	Component ()

virtual	~Component ()

Private Member Functions
LinearComponent &	operator= (const LinearComponent &)

Private Attributes
CuMatrix< BaseFloat >	params_

BaseFloat	orthonormal_constraint_

bool	use_natural_gradient_

OnlineNaturalGradient	preconditioner_in_

OnlineNaturalGradient	preconditioner_out_

Additional Inherited Members
Static Public Member Functions inherited from Component
static Component *	ReadNew (std::istream &is, bool binary)
	Read component from stream (works out its type). Dies on error. More...

static Component *	NewComponentOfType (const std::string &type)
	Returns a new Component of the given type e.g. More...

Protected Member Functions inherited from UpdatableComponent
void	InitLearningRatesFromConfig (ConfigLine *cfl)

std::string	ReadUpdatableCommon (std::istream &is, bool binary)

void	WriteUpdatableCommon (std::ostream &is, bool binary) const

Protected Attributes inherited from UpdatableComponent
BaseFloat	learning_rate_
	learning rate (typically 0.0..0.01) More...

BaseFloat	learning_rate_factor_
	learning rate factor (normally 1.0, but can be set to another < value so that when < you call SetLearningRate(), that value will be scaled by this factor. More...

BaseFloat	l2_regularize_
	L2 regularization constant. More...

bool	is_gradient_
	True if this component is to be treated as a gradient rather than as parameters. More...

BaseFloat	max_change_
	configuration value for imposing max-change More...

Detailed Description

Definition at line 927 of file nnet-simple-component.h.

Constructor & Destructor Documentation

◆ LinearComponent() [1/3]

LinearComponent ( )

inline

Definition at line 952 of file nnet-simple-component.h.

References Component::Add(), Component::ConsolidateMemory(), PnormComponent::Copy(), kaldi::nnet3::DotProduct(), kaldi::nnet3::FreezeNaturalGradient(), Component::Info(), PnormComponent::InitFromConfig(), kaldi::nnet3::NumParameters(), kaldi::nnet3::PerturbParams(), and Component::Scale().

952 { }

◆ LinearComponent() [2/3]

LinearComponent ( const LinearComponent & other )

explicit

Definition at line 3275 of file nnet-simple-component.cc.

                                  :
     UpdatableComponent(other),
     params_(other.params_),
     orthonormal_constraint_(other.orthonormal_constraint_),
     use_natural_gradient_(other.use_natural_gradient_),
     preconditioner_in_(other.preconditioner_in_),
     preconditioner_out_(other.preconditioner_out_) { }

◆ LinearComponent() [3/3]

LinearComponent ( const CuMatrix< BaseFloat > & params )

explicit

Definition at line 3284 of file nnet-simple-component.cc.

References LinearComponent::preconditioner_in_, LinearComponent::preconditioner_out_, OnlineNaturalGradient::SetRank(), and OnlineNaturalGradient::SetUpdatePeriod().

                                                                  :
     params_(params),
     orthonormal_constraint_(0.0),
     use_natural_gradient_(true) {
   // Set defaults for natural gradient.
   preconditioner_in_.SetRank(40);
   preconditioner_out_.SetRank(80);
   preconditioner_in_.SetUpdatePeriod(4);
   preconditioner_out_.SetUpdatePeriod(4);
   // the component-level defaults of alpha and num_samples_history, at 4.0 and
   // 2000.0, are the same as in the NaturalGradientOnline code, so there is no
   // need to set those here.
 }

Member Function Documentation

◆ Add()

void Add	(	BaseFloat	alpha,
		const Component &	other
	)

virtual

This virtual function when called by – an UpdatableComponent adds the parameters of another updatable component, times some constant, to the current parameters.

– a NonlinearComponent (or another component that stores stats, like BatchNormComponent)– it relates to adding stats. Otherwise it will normally do nothing.

Reimplemented from Component.

Definition at line 3303 of file nnet-simple-component.cc.

References KALDI_ASSERT, and LinearComponent::params_.

                                                                     {
   const LinearComponent *other =
       dynamic_cast<const LinearComponent*>(&other_in);
   KALDI_ASSERT(other != NULL);
   params_.AddMat(alpha, other->params_);
 }

◆ Backprop()

void Backprop	(	const std::string &	debug_info,
		const ComponentPrecomputedIndexes *	indexes,
		const CuMatrixBase< BaseFloat > &	in_value,
		const CuMatrixBase< BaseFloat > &	out_value,
		const CuMatrixBase< BaseFloat > &	out_deriv,
		void *	memo,
		Component *	to_update,
		CuMatrixBase< BaseFloat > *	in_deriv
	)		const

virtual

Backprop function; depending on which of the arguments 'to_update' and 'in_deriv' are non-NULL, this can compute input-data derivatives and/or perform model update.

Parameters

[in]	debug_info	The component name, to be printed out in any warning messages.
[in]	indexes	A pointer to some information output by this class's PrecomputeIndexes function (will be NULL for simple components, i.e. those that don't do things like splicing).
[in]	in_value	The matrix that was given as input to the Propagate function. Will be ignored (and may be empty) if Properties()&kBackpropNeedsInput == 0.
[in]	out_value	The matrix that was output from the Propagate function. Will be ignored (and may be empty) if Properties()&kBackpropNeedsOutput == 0
[in]	out_deriv	The derivative at the output of this component.
[in]	memo	This will normally be NULL, but for component types that set the flag kUsesMemo, this will be the return value of the Propagate() function that corresponds to this Backprop() function. Ownership of any pointers is not transferred to the Backprop function; DeleteMemo() will be called to delete it.
[out]	to_update	If model update is desired, the Component to be updated, else NULL. Does not have to be identical to this. If supplied, you can assume that to_update->Properties() & kUpdatableComponent is nonzero.
[out]	in_deriv	The derivative at the input of this component, if needed (else NULL). If Properties()&kBackpropInPlace, may be the same matrix as out_deriv. If Properties()&kBackpropAdds, this is added to by the Backprop routine, else it is set. The component code chooses which mode to work in, based on convenience.

Implements Component.

Definition at line 3231 of file nnet-simple-component.cc.

References CuMatrixBase< Real >::AddMatMat(), UpdatableComponent::is_gradient_, kaldi::kNoTrans, kaldi::kTrans, UpdatableComponent::learning_rate_, NVTX_RANGE, LinearComponent::params_, OnlineNaturalGradient::PreconditionDirections(), LinearComponent::preconditioner_in_, and LinearComponent::preconditioner_out_.

                                                                         {
   NVTX_RANGE("LinearComponent::Backprop");
   LinearComponent *to_update = dynamic_cast<LinearComponent*>(to_update_in);
 
   // Propagate the derivative back to the input.  add with coefficient 1.0 since
   // property kBackpropAdds is true.  If we wanted to add with coefficient 0.0
   // we'd need to zero the in_deriv, in case of infinities.
   if (in_deriv)
     in_deriv->AddMatMat(1.0, out_deriv, kNoTrans, params_, kNoTrans, 1.0);
 
   if (to_update != NULL) {
     if (!to_update->is_gradient_) {
       CuMatrix<BaseFloat> in_value_temp(in_value), out_deriv_temp(out_deriv);
       // These "scale" values get will get multiplied into the learning rate (faster
       // than having the matrices scaled inside the preconditioning code).
       BaseFloat in_scale, out_scale;
       to_update->preconditioner_in_.PreconditionDirections(&in_value_temp,
                                                            &in_scale);
       to_update->preconditioner_out_.PreconditionDirections(&out_deriv_temp,
                                                             &out_scale);
       BaseFloat local_lrate = in_scale * out_scale * to_update->learning_rate_;
 
       to_update->params_.AddMatMat(local_lrate, out_deriv_temp, kTrans,
                                    in_value_temp, kNoTrans, 1.0);
     } else {
       to_update->params_.AddMatMat(to_update->learning_rate_,
                                    out_deriv, kTrans,
                                    in_value, kNoTrans, 1.0);
     }
   }
 }

◆ ConsolidateMemory()

void ConsolidateMemory ( )

virtual

This virtual function relates to memory management, and avoiding fragmentation.

It is called only once per model, after we do the first minibatch of training. The default implementation does nothing, but it can be overridden by child classes, where it may re-initialize certain quantities that may possibly have been allocated during the forward pass (e.g. certain statistics; OnlineNaturalGradient objects). We use our own CPU-based allocator (see cu-allocator.h) and since it can't do paging since we're not in control of the GPU page table, fragmentation can be a problem. The allocator always tries to put things in 'low-address memory' (i.e. at smaller memory addresses) near the beginning of the block it allocated, to avoid fragmentation; but if permanent things (belonging to the model) are allocated in the forward pass, they can permanently stay in high memory. This function helps to prevent that, by re-allocating those things into low-address memory (It's important that it's called after all the temporary buffers for the forward-backward have been freed, so that there is low-address memory available)).

Reimplemented from Component.

Definition at line 3337 of file nnet-simple-component.cc.

References LinearComponent::preconditioner_in_, LinearComponent::preconditioner_out_, and OnlineNaturalGradient::Swap().

                                         {
   OnlineNaturalGradient temp_in(preconditioner_in_);
   preconditioner_in_.Swap(&temp_in);
   OnlineNaturalGradient temp_out(preconditioner_out_);
   preconditioner_out_.Swap(&temp_out);
 }

◆ Copy()

Component * Copy ( ) const

virtual

Copies component (deep copy).

Implements Component.

Definition at line 3271 of file nnet-simple-component.cc.

                                        {
   return new LinearComponent(*this);
 }

◆ DotProduct()

BaseFloat DotProduct ( const UpdatableComponent & other ) const

virtual

Computes dot-product between parameters of two instances of a Component.

Can be used for computing parameter-norm of an UpdatableComponent.

Implements UpdatableComponent.

Definition at line 3326 of file nnet-simple-component.cc.

References kaldi::kTrans, LinearComponent::params_, and kaldi::TraceMatMat().

                                                                               {
   const LinearComponent *other =
       dynamic_cast<const LinearComponent*>(&other_in);
   return TraceMatMat(params_, other->params_, kTrans);
 }

◆ FreezeNaturalGradient()

void FreezeNaturalGradient ( bool freeze )

virtual

freezes/unfreezes NaturalGradient updates, if applicable (to be overriden by components that use Natural Gradient).

Reimplemented from UpdatableComponent.

Definition at line 3332 of file nnet-simple-component.cc.

References OnlineNaturalGradient::Freeze(), LinearComponent::preconditioner_in_, and LinearComponent::preconditioner_out_.

                                                        {
   preconditioner_in_.Freeze(freeze);
   preconditioner_out_.Freeze(freeze);
 }

◆ Info()

std::string Info ( ) const

virtual

Returns some text-form information about this component, for diagnostics.

Starts with the type of the component. E.g. "SigmoidComponent dim=900", although most components will have much more info.

Reimplemented from UpdatableComponent.

Definition at line 3203 of file nnet-simple-component.cc.

References OnlineNaturalGradient::GetAlpha(), OnlineNaturalGradient::GetNumSamplesHistory(), OnlineNaturalGradient::GetRank(), OnlineNaturalGradient::GetUpdatePeriod(), kaldi::GetVerboseLevel(), UpdatableComponent::Info(), AffineComponent::orthonormal_constraint_, NaturalGradientAffineComponent::preconditioner_in_, NaturalGradientAffineComponent::preconditioner_out_, and kaldi::nnet3::PrintParameterStats().

                                       {
   std::ostringstream stream;
   stream << UpdatableComponent::Info();
   PrintParameterStats(stream, "params", params_,
                       false, // include_mean
                       true, // include_row_norms
                       true, // include_column_norms
                       GetVerboseLevel() >= 2); // include_singular_values
   if (orthonormal_constraint_ != 0.0)
     stream << ", orthonormal-constraint=" << orthonormal_constraint_;
   stream << ", use-natural-gradient="
          << (use_natural_gradient_ ? "true" : "false")
          << ", rank-in=" << preconditioner_in_.GetRank()
          << ", rank-out=" << preconditioner_out_.GetRank()
          << ", num-samples-history="
          << preconditioner_in_.GetNumSamplesHistory()
          << ", update-period=" << preconditioner_in_.GetUpdatePeriod()
          << ", alpha=" << preconditioner_in_.GetAlpha();
   return stream.str();
 }

◆ InitFromConfig()

void InitFromConfig ( ConfigLine * cfl )

virtual

Initialize, from a ConfigLine object.

Parameters

[in] cfl A ConfigLine containing any parameters that are needed for initialization. For example: "dim=100 param-stddev=0.1"

Implements Component.

Definition at line 3108 of file nnet-simple-component.cc.

References ConfigLine::GetValue(), ConfigLine::HasUnusedValues(), UpdatableComponent::InitLearningRatesFromConfig(), AffineComponent::InputDim(), UpdatableComponent::is_gradient_, KALDI_ASSERT, KALDI_ERR, AffineComponent::orthonormal_constraint_, AffineComponent::OutputDim(), NaturalGradientAffineComponent::preconditioner_in_, NaturalGradientAffineComponent::preconditioner_out_, kaldi::ReadKaldiObject(), OnlineNaturalGradient::SetAlpha(), OnlineNaturalGradient::SetNumSamplesHistory(), OnlineNaturalGradient::SetRank(), OnlineNaturalGradient::SetUpdatePeriod(), ConfigLine::UnusedValues(), and ConfigLine::WholeLine().

                                                     {
   bool ok = true;
   std::string matrix_filename;
   is_gradient_ = false;  // not configurable; there's no reason you'd want this
 
   InitLearningRatesFromConfig(cfl);
 
   int32 input_dim = -1, output_dim = -1;
   if (cfl->GetValue("matrix", &matrix_filename)) {
     ReadKaldiObject(matrix_filename, &params_); // will abort on failure.
     KALDI_ASSERT(params_.NumRows() != 0);
     if (cfl->GetValue("input-dim", &input_dim))
       KALDI_ASSERT(input_dim == InputDim() &&
                    "input-dim mismatch vs. matrix.");
     if (cfl->GetValue("output-dim", &output_dim))
       KALDI_ASSERT(output_dim == OutputDim() &&
                    "output-dim mismatch vs. matrix.");
   } else {
     ok = ok && cfl->GetValue("input-dim", &input_dim);
     ok = ok && cfl->GetValue("output-dim", &output_dim);
     if (!ok)
       KALDI_ERR << "Bad initializer " << cfl->WholeLine();
     BaseFloat param_stddev = 1.0 / std::sqrt(input_dim);
     cfl->GetValue("param-stddev", &param_stddev);
     params_.Resize(output_dim, input_dim);
     KALDI_ASSERT(output_dim > 0 && input_dim > 0 && param_stddev >= 0.0);
     params_.SetRandn(); // sets to random normally distributed noise.
     params_.Scale(param_stddev);
   }
   // Read various natural-gradient-related configs.
   int32 rank_in = -1, rank_out = -1, update_period = 4;
   BaseFloat alpha = 4.0,
       num_samples_history = 2000.0;
 
   use_natural_gradient_ = true;
 
   cfl->GetValue("num-samples-history", &num_samples_history);
   cfl->GetValue("alpha", &alpha);
   cfl->GetValue("rank-in", &rank_in);
   cfl->GetValue("rank-out", &rank_out);
   cfl->GetValue("update-period", &update_period);
   cfl->GetValue("use-natural-gradient", &use_natural_gradient_);
 
   if (rank_in < 0)
     rank_in = std::min<int32>(20, (InputDim() + 1) / 2);
   if (rank_out < 0)
     rank_out = std::min<int32>(80, (OutputDim() + 1) / 2);
 
   preconditioner_in_.SetAlpha(alpha);
   preconditioner_out_.SetAlpha(alpha);
   preconditioner_in_.SetRank(rank_in);
   preconditioner_out_.SetRank(rank_out);
   preconditioner_in_.SetNumSamplesHistory(num_samples_history);
   preconditioner_out_.SetNumSamplesHistory(num_samples_history);
   preconditioner_in_.SetUpdatePeriod(update_period);
   preconditioner_out_.SetUpdatePeriod(update_period);
 
   orthonormal_constraint_ = 0.0;
   cfl->GetValue("orthonormal-constraint", &orthonormal_constraint_);
 
   if (cfl->HasUnusedValues())
     KALDI_ERR << "Could not process these elements in initializer: "
               << cfl->UnusedValues();
 }

◆ InputDim()

virtual int32 InputDim ( ) const

inlinevirtual

Returns input-dimension of this component.

Implements Component.

Definition at line 929 of file nnet-simple-component.h.

929 { return params_.NumCols(); }

kaldi::nnet3::LinearComponent::params_

CuMatrix< BaseFloat > params_

Definition: nnet-simple-component.h:980

◆ NumParameters()

int32 NumParameters ( ) const

virtual

The following new virtual function returns the total dimension of the parameters in this class.

Reimplemented from UpdatableComponent.

Definition at line 3315 of file nnet-simple-component.cc.

References LinearComponent::params_.

Referenced by LinearComponent::UnVectorize(), and LinearComponent::Vectorize().

                                            {
   return params_.NumRows() * params_.NumCols();
 }

◆ operator=()

LinearComponent& operator= ( const LinearComponent & )

private

◆ OrthonormalConstraint()

BaseFloat OrthonormalConstraint ( ) const

inline

Definition at line 971 of file nnet-simple-component.h.

Referenced by kaldi::nnet3::ConstrainOrthonormal().

971 { return orthonormal_constraint_; }

kaldi::nnet3::LinearComponent::orthonormal_constraint_

BaseFloat orthonormal_constraint_

Definition: nnet-simple-component.h:982

◆ OutputDim()

virtual int32 OutputDim ( ) const

inlinevirtual

Returns output-dimension of this component.

Implements Component.

Definition at line 930 of file nnet-simple-component.h.

Referenced by ModelCollapser::GetDiagonallyPreModifiedComponentIndex().

930 { return params_.NumRows(); }

kaldi::nnet3::LinearComponent::params_

CuMatrix< BaseFloat > params_

Definition: nnet-simple-component.h:980

◆ Params() [1/2]

CuMatrixBase<BaseFloat>& Params ( )

inline

Definition at line 972 of file nnet-simple-component.h.

Referenced by kaldi::nnet3::ConstrainOrthonormal(), and ModelCollapser::GetDiagonallyPreModifiedComponentIndex().

972 { return params_; }

kaldi::nnet3::LinearComponent::params_

CuMatrix< BaseFloat > params_

Definition: nnet-simple-component.h:980

◆ Params() [2/2]

const CuMatrixBase<BaseFloat>& Params ( ) const

inline

Definition at line 973 of file nnet-simple-component.h.

973 { return params_; }

kaldi::nnet3::LinearComponent::params_

CuMatrix< BaseFloat > params_

Definition: nnet-simple-component.h:980

◆ PerturbParams()

void PerturbParams ( BaseFloat stddev )

virtual

This function is to be used in testing.

It adds unit noise times "stddev" to the parameters of the component.

Implements UpdatableComponent.

Definition at line 3310 of file nnet-simple-component.cc.

References LinearComponent::params_, and CuMatrixBase< Real >::SetRandn().

                                                     {
   CuMatrix<BaseFloat> temp_params(params_);
   temp_params.SetRandn();
   params_.AddMat(stddev, temp_params);
 }

◆ Propagate()

void * Propagate	(	const ComponentPrecomputedIndexes *	indexes,
		const CuMatrixBase< BaseFloat > &	in,
		CuMatrixBase< BaseFloat > *	out
	)		const

virtual

Propagate function.

Parameters

[in]	indexes	A pointer to some information output by this class's PrecomputeIndexes function (will be NULL for simple components, i.e. those that don't do things like splicing).
[in]	in	The input to this component. Num-columns == InputDim().
[out]	out	The output of this component. Num-columns == OutputDim(). Note: output of this component will be added to the initial value of "out" if Properties()&kPropagateAdds != 0; otherwise the output will be set and the initial value ignored. Each Component chooses whether it is more convenient implementation-wise to add or set, and the calling code has to deal with it.

Returns: Normally returns NULL, but may return a non-NULL value for components which have the flag kUsesMemo set. This value will be passed into the corresponding Backprop routine.

Implements Component.

Definition at line 3224 of file nnet-simple-component.cc.

References CuMatrixBase< Real >::AddMatMat(), kaldi::kNoTrans, and kaldi::kTrans.

                                                                      {
   out->AddMatMat(1.0, in, kNoTrans, params_, kTrans, 1.0);
   return NULL;
 }

◆ Properties()

virtual int32 Properties ( ) const

inlinevirtual

Return bitmask of the component's properties.

These properties depend only on the component's type. See enum ComponentProperties.

Implements Component.

Definition at line 933 of file nnet-simple-component.h.

References PnormComponent::Backprop(), kaldi::nnet3::kBackpropAdds, kaldi::nnet3::kBackpropNeedsInput, kaldi::nnet3::kPropagateAdds, kaldi::nnet3::kSimpleComponent, kaldi::nnet3::kUpdatableComponent, PnormComponent::Propagate(), PnormComponent::Read(), and PnormComponent::Write().

                                    {
     return kSimpleComponent|kUpdatableComponent|kBackpropNeedsInput|
         kPropagateAdds|kBackpropAdds;
   }

◆ Read()

void Read	(	std::istream &	is,
		bool	binary
	)

virtual

Read function (used after we know the type of the Component); accepts input that is missing the token that describes the component type, in case it has already been consumed.

Implements Component.

Definition at line 3069 of file nnet-simple-component.cc.

References kaldi::nnet3::ExpectToken(), KALDI_ASSERT, AffineComponent::orthonormal_constraint_, kaldi::PeekToken(), NaturalGradientAffineComponent::preconditioner_in_, NaturalGradientAffineComponent::preconditioner_out_, kaldi::ReadBasicType(), UpdatableComponent::ReadUpdatableCommon(), OnlineNaturalGradient::SetAlpha(), OnlineNaturalGradient::SetNumSamplesHistory(), OnlineNaturalGradient::SetRank(), and OnlineNaturalGradient::SetUpdatePeriod().

                                                       {
   std::string token = ReadUpdatableCommon(is, binary);
   KALDI_ASSERT(token == "");
   ExpectToken(is, binary, "<Params>");
   params_.Read(is, binary);
   if (PeekToken(is, binary) == 'O') {
     ExpectToken(is, binary, "<OrthonormalConstraint>");
     ReadBasicType(is, binary, &orthonormal_constraint_);
   } else {
     orthonormal_constraint_ = 0.0;
   }
   ExpectToken(is, binary, "<UseNaturalGradient>");
   ReadBasicType(is, binary, &use_natural_gradient_);
 
   // Read various natural-gradient-related configs.
   int32 rank_in,  rank_out, update_period;
   BaseFloat alpha, num_samples_history;
   ExpectToken(is, binary, "<RankInOut>");
   ReadBasicType(is, binary, &rank_in);
   ReadBasicType(is, binary, &rank_out);
   ExpectToken(is, binary, "<Alpha>");
   ReadBasicType(is, binary, &alpha);
   ExpectToken(is, binary, "<NumSamplesHistory>");
   ReadBasicType(is, binary, &num_samples_history);
   ExpectToken(is, binary, "<UpdatePeriod>");
   ReadBasicType(is, binary, &update_period);
 
   preconditioner_in_.SetAlpha(alpha);
   preconditioner_out_.SetAlpha(alpha);
   preconditioner_in_.SetRank(rank_in);
   preconditioner_out_.SetRank(rank_out);
   preconditioner_in_.SetNumSamplesHistory(num_samples_history);
   preconditioner_out_.SetNumSamplesHistory(num_samples_history);
   preconditioner_in_.SetUpdatePeriod(update_period);
   preconditioner_out_.SetUpdatePeriod(update_period);
 
   ExpectToken(is, binary, "</LinearComponent>");
 }

◆ Scale()

void Scale ( BaseFloat scale )

virtual

This virtual function when called on – an UpdatableComponent scales the parameters by "scale" when called by an UpdatableComponent.

– a Nonlinear component (or another component that stores stats, like BatchNormComponent)– it relates to scaling activation stats, not parameters. Otherwise it will normally do nothing.

Reimplemented from Component.

Definition at line 3298 of file nnet-simple-component.cc.

References LinearComponent::params_.

Referenced by ModelCollapser::GetScaledComponentIndex().

                                            {
   if (scale == 0.0) params_.SetZero();
   else params_.Scale(scale);
 }

◆ Type()

virtual std::string Type ( ) const

inlinevirtual

Returns a string such as "SigmoidComponent", describing the type of the object.

Implements Component.

Definition at line 932 of file nnet-simple-component.h.

Referenced by FixedAffineComponent::InitFromConfig().

932 { return "LinearComponent"; }

◆ UnVectorize()

void UnVectorize ( const VectorBase< BaseFloat > & params )

virtual

Converts the parameters from vector form.

Reimplemented from UpdatableComponent.

Definition at line 3322 of file nnet-simple-component.cc.

References VectorBase< Real >::Dim(), KALDI_ASSERT, LinearComponent::NumParameters(), and LinearComponent::params_.

                                                                      {
   KALDI_ASSERT(params.Dim() == this->NumParameters());
   params_.CopyRowsFromVec(params);
 }

◆ Vectorize()

void Vectorize ( VectorBase< BaseFloat > * params ) const

virtual

Turns the parameters into vector form.

We put the vector form on the CPU, because in the kinds of situations where we do this, we'll tend to use too much memory for the GPU.

Reimplemented from UpdatableComponent.

Definition at line 3318 of file nnet-simple-component.cc.

References VectorBase< Real >::CopyRowsFromMat(), VectorBase< Real >::Dim(), KALDI_ASSERT, LinearComponent::NumParameters(), and LinearComponent::params_.

                                                                    {
   KALDI_ASSERT(params->Dim() == this->NumParameters());
   params->CopyRowsFromMat(params_);
 }

◆ Write()

void Write	(	std::ostream &	os,
		bool	binary
	)		const

virtual

Write component to stream.

Implements Component.

Definition at line 3174 of file nnet-simple-component.cc.

References OnlineNaturalGradient::GetAlpha(), OnlineNaturalGradient::GetNumSamplesHistory(), OnlineNaturalGradient::GetRank(), OnlineNaturalGradient::GetUpdatePeriod(), AffineComponent::orthonormal_constraint_, NaturalGradientAffineComponent::preconditioner_in_, NaturalGradientAffineComponent::preconditioner_out_, kaldi::WriteBasicType(), kaldi::WriteToken(), and UpdatableComponent::WriteUpdatableCommon().

                                                {
   WriteUpdatableCommon(os, binary);  // Write the opening tag and learning rate
   WriteToken(os, binary, "<Params>");
   params_.Write(os, binary);
   if (orthonormal_constraint_ != 0.0) {
     WriteToken(os, binary, "<OrthonormalConstraint>");
     WriteBasicType(os, binary, orthonormal_constraint_);
   }
   WriteToken(os, binary, "<UseNaturalGradient>");
   WriteBasicType(os, binary, use_natural_gradient_);
 
   int32 rank_in = preconditioner_in_.GetRank(),
       rank_out = preconditioner_out_.GetRank(),
       update_period = preconditioner_in_.GetUpdatePeriod();
   BaseFloat alpha = preconditioner_in_.GetAlpha(),
       num_samples_history = preconditioner_in_.GetNumSamplesHistory();
   WriteToken(os, binary, "<RankInOut>");
   WriteBasicType(os, binary, rank_in);
   WriteBasicType(os, binary, rank_out);
   WriteToken(os, binary, "<Alpha>");
   WriteBasicType(os, binary, alpha);
   WriteToken(os, binary, "<NumSamplesHistory>");
   WriteBasicType(os, binary, num_samples_history);
   WriteToken(os, binary, "<UpdatePeriod>");
   WriteBasicType(os, binary, update_period);
   WriteToken(os, binary, "</LinearComponent>");
 }

Member Data Documentation

◆ orthonormal_constraint_

BaseFloat orthonormal_constraint_

private

Definition at line 982 of file nnet-simple-component.h.

◆ params_

CuMatrix<BaseFloat> params_

private

Definition at line 980 of file nnet-simple-component.h.

Referenced by LinearComponent::Add(), LinearComponent::Backprop(), LinearComponent::DotProduct(), LinearComponent::NumParameters(), LinearComponent::PerturbParams(), LinearComponent::Scale(), LinearComponent::UnVectorize(), and LinearComponent::Vectorize().

◆ preconditioner_in_

OnlineNaturalGradient preconditioner_in_

private

Definition at line 985 of file nnet-simple-component.h.

Referenced by LinearComponent::Backprop(), LinearComponent::ConsolidateMemory(), LinearComponent::FreezeNaturalGradient(), and LinearComponent::LinearComponent().

◆ preconditioner_out_

OnlineNaturalGradient preconditioner_out_

private

Definition at line 986 of file nnet-simple-component.h.

Referenced by LinearComponent::Backprop(), LinearComponent::ConsolidateMemory(), LinearComponent::FreezeNaturalGradient(), and LinearComponent::LinearComponent().

◆ use_natural_gradient_

bool use_natural_gradient_

private

Definition at line 984 of file nnet-simple-component.h.

The documentation for this class was generated from the following files:

nnet3/nnet-simple-component.h
nnet3/nnet-simple-component.cc

Public Member Functions

Private Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ LinearComponent() [1/3]

◆ LinearComponent() [2/3]

◆ LinearComponent() [3/3]

Member Function Documentation

◆ Add()

◆ Backprop()

◆ ConsolidateMemory()

◆ Copy()

◆ DotProduct()

◆ FreezeNaturalGradient()

◆ Info()

◆ InitFromConfig()

◆ InputDim()

◆ NumParameters()

◆ operator=()

◆ OrthonormalConstraint()

◆ OutputDim()

◆ Params() [1/2]

◆ Params() [2/2]

◆ PerturbParams()

◆ Propagate()

◆ Properties()

◆ Read()

◆ Scale()

◆ Type()

◆ UnVectorize()

◆ Vectorize()

◆ Write()

Member Data Documentation

◆ orthonormal_constraint_

◆ params_

◆ preconditioner_in_

◆ preconditioner_out_

◆ use_natural_gradient_