#include <nnet-simple-component.h>

Inheritance diagram for SigmoidComponent:

Collaboration diagram for SigmoidComponent:

[legend]

Public Member Functions
	SigmoidComponent (const SigmoidComponent &other)

	SigmoidComponent ()

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 Component *	Copy () const
	Copies component (deep copy). 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 > &, const CuMatrixBase< BaseFloat > &out_value, 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	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...

Public Member Functions inherited from NonlinearComponent
	NonlinearComponent ()

	NonlinearComponent (const NonlinearComponent &other)

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

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

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

virtual void	Read (std::istream &is, bool binary)
	We implement Read at this level as it just needs the Type(). 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 std::string	Info () const
	Returns some text-form information about this component, for diagnostics. More...

virtual void	Write (std::ostream &os, bool binary) const
	Write component to stream. 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	ConsolidateMemory ()
	This virtual function relates to memory management, and avoiding fragmentation. More...

const CuVector< double > &	ValueSum () const

const CuVector< double > &	DerivSum () const

double	Count () const

Public Member Functions inherited from Component
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
void	RepairGradients (const CuMatrixBase< BaseFloat > &out_value, CuMatrixBase< BaseFloat > in_deriv, SigmoidComponent to_update) const

SigmoidComponent &	operator= (const SigmoidComponent &other)

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 Types inherited from NonlinearComponent
enum	{ kUnsetThreshold = -1000 }

Protected Member Functions inherited from NonlinearComponent
void	StoreStatsInternal (const CuMatrixBase< BaseFloat > &out_value, const CuMatrixBase< BaseFloat > *deriv=NULL)

void	StoreBackpropStats (const CuMatrixBase< BaseFloat > &out_deriv)

const NonlinearComponent &	operator= (const NonlinearComponent &other)

Protected Attributes inherited from NonlinearComponent
int32	dim_

int32	block_dim_

CuVector< double >	value_sum_

CuVector< double >	deriv_sum_

double	count_

CuVector< double >	oderiv_sumsq_

double	oderiv_count_

double	num_dims_self_repaired_

double	num_dims_processed_

BaseFloat	self_repair_lower_threshold_

BaseFloat	self_repair_upper_threshold_

BaseFloat	self_repair_scale_

Detailed Description

Definition at line 222 of file nnet-simple-component.h.

Constructor & Destructor Documentation

◆ SigmoidComponent() [1/2]

SigmoidComponent ( const SigmoidComponent & other )

inlineexplicit

Definition at line 224 of file nnet-simple-component.h.

224 : NonlinearComponent(other) { }

kaldi::nnet3::NonlinearComponent::NonlinearComponent

NonlinearComponent()

Definition: nnet-component-itf.cc:605

◆ SigmoidComponent() [2/2]

SigmoidComponent ( )

inline

Definition at line 225 of file nnet-simple-component.h.

225 { }

Member Function Documentation

◆ 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 328 of file nnet-simple-component.cc.

References CuMatrixBase< Real >::DiffSigmoid(), NVTX_RANGE, and NonlinearComponent::StoreBackpropStats().

                                                                          {
   NVTX_RANGE("SigmoidComponent::Backprop");
   if (in_deriv != NULL) {
     in_deriv->DiffSigmoid(out_value, out_deriv);
     SigmoidComponent *to_update = dynamic_cast<SigmoidComponent*>(to_update_in);
     if (to_update != NULL) {
       RepairGradients(out_value, in_deriv, to_update);
       to_update->StoreBackpropStats(out_deriv);
     }
   }
 }

◆ Copy()

virtual Component* Copy ( ) const

inlinevirtual

Copies component (deep copy).

Implements Component.

Definition at line 230 of file nnet-simple-component.h.

References PnormComponent::Backprop(), PnormComponent::Propagate(), and Component::StoreStats().

230 { return new SigmoidComponent(*this); }

kaldi::nnet3::SigmoidComponent::SigmoidComponent

SigmoidComponent()

Definition: nnet-simple-component.h:225

◆ operator=()

SigmoidComponent& operator= ( const SigmoidComponent & other )

private

◆ 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 321 of file nnet-simple-component.cc.

References CuMatrixBase< Real >::Sigmoid().

                                                                      {
   out->Sigmoid(in);
   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 227 of file nnet-simple-component.h.

References kaldi::nnet3::kBackpropNeedsOutput, kaldi::nnet3::kPropagateInPlace, kaldi::nnet3::kSimpleComponent, and kaldi::nnet3::kStoresStats.

                                    {
     return kSimpleComponent|kBackpropNeedsOutput|kPropagateInPlace|kStoresStats;
   }

◆ RepairGradients()

void RepairGradients	(	const CuMatrixBase< BaseFloat > &	out_value,
		CuMatrixBase< BaseFloat > *	in_deriv,
		SigmoidComponent *	to_update
	)		const

private

Definition at line 347 of file nnet-simple-component.cc.

References CuVectorBase< Real >::Add(), CuMatrixBase< Real >::AddMatDiagVec(), CuVectorBase< Real >::AddVec(), CuMatrixBase< Real >::AddVecToRows(), CuMatrixBase< Real >::ApplyHeaviside(), DropoutComponent::dim_, KALDI_ASSERT, KALDI_ERR, kaldi::kNoTrans, NonlinearComponent::num_dims_processed_, NonlinearComponent::num_dims_self_repaired_, kaldi::RandUniform(), and CuVectorBase< Real >::Sum().

                                        {
   KALDI_ASSERT(to_update != NULL);
   // maximum possible derivative of SigmoidComponent is 0.25.
   // the default lower-threshold on the derivative, below which we
   // add a term to the derivative to encourage the inputs to the sigmoid
   // to be closer to zero, is 0.05, which means the derivative is on average
   // 5 times smaller than its maximum possible value.
   BaseFloat default_lower_threshold = 0.05;
 
   // we use this 'repair_probability' (hardcoded for now) to limit
   // this code to running on about half of the minibatches.
   BaseFloat repair_probability = 0.5;
 
   to_update->num_dims_processed_ += dim_;
 
   if (self_repair_scale_ == 0.0 || count_ == 0.0 || deriv_sum_.Dim() != dim_ ||
       RandUniform() > repair_probability)
     return;
 
   // check that the self-repair scale is in a reasonable range.
   KALDI_ASSERT(self_repair_scale_ > 0.0 && self_repair_scale_ < 0.1);
   BaseFloat unset = kUnsetThreshold; // -1000.0
   BaseFloat lower_threshold = (self_repair_lower_threshold_ == unset ?
                                default_lower_threshold :
                                self_repair_lower_threshold_) *
       count_;
   if (self_repair_upper_threshold_ != unset) {
     KALDI_ERR << "Do not set the self-repair-upper-threshold for sigmoid "
               << "components, it does nothing.";
   }
 
   // thresholds_vec is actually a 1-row matrix.  (the ApplyHeaviside
   // function isn't defined for vectors).
   CuMatrix<BaseFloat> thresholds(1, dim_);
   CuSubVector<BaseFloat> thresholds_vec(thresholds, 0);
   thresholds_vec.AddVec(-1.0, deriv_sum_);
   thresholds_vec.Add(lower_threshold);
   thresholds.ApplyHeaviside();
   to_update->num_dims_self_repaired_ += thresholds_vec.Sum();
 
   // At this point, 'thresholds_vec' contains a 1 for each dimension of
   // the output that is 'problematic', i.e. for which the avg-deriv
   // is less than the self-repair lower threshold, and a 0 for
   // each dimension that is not problematic.
 
   // what we want to do is to add
   // -self_repair_scale_ / repair_probability times (2 * output-valiue - 1.0)
   // to the input derivative for each problematic dimension.
 
   // Here, 2 * output - 1.0 is a version of the sigmoid that goes from -1.0 to
   // 1.0, like a tanh.  the negative sign is so that for inputs <0, we push them
   // up towards 0, and for inputs >0, we push them down towards 0.
   // Our use of this sigmoid-type function here is just a convenience since
   // we have it available.  We could use just about any function that is positive
   // for inputs < 0 and negative for inputs > 0.
 
   // We can rearrange the above as: for only the problematic columns,
   //   input-deriv -= 2 * self-repair-scale / repair-probabilty * output
   //   input-deriv +=  self-repair-scale / repair-probabilty
   // which we can write as:
   //   input-deriv -= 2 * self-repair-scale / repair-probabilty * output * thresholds-vec
   //   input-deriv +=  self-repair-scale / repair-probabilty * thresholds-vec
 
   in_deriv->AddMatDiagVec(-2.0 * self_repair_scale_ / repair_probability,
                           out_value, kNoTrans, thresholds_vec);
   in_deriv->AddVecToRows(self_repair_scale_ / repair_probability,
                          thresholds_vec);
 }

◆ StoreStats()

void StoreStats	(	const CuMatrixBase< BaseFloat > &	in_value,
		const CuMatrixBase< BaseFloat > &	out_value,
		void *	memo
	)

virtual

This function may store stats on average activation values, and for some component types, the average value of the derivative of the nonlinearity.

It only does something for those components that have nonzero Properties()&kStoresStats.

Parameters

[in]	in_value	The input to the Propagate() function. Note: if the component sets the flag kPropagateInPlace, this should not be used; the empty matrix will be provided here if in-place propagation was used.
[in]	out_value	The output of the Propagate() function.
[in]	memo	The 'memo' returned by the Propagate() function; this will usually be NULL.

Reimplemented from Component.

Definition at line 421 of file nnet-simple-component.cc.

References kaldi::kUndefined, CuMatrixBase< Real >::NumCols(), CuMatrixBase< Real >::NumRows(), kaldi::RandInt(), and CuMatrixBase< Real >::Set().

                                               {
   // Only store stats about every other minibatch (but on the first minibatch,
   // always store it, which is necessary for the ConsolidateMemory() operation
   // to work correctly.
   if (RandInt(0, 1) == 0 && count_ != 0)
     return;
   // derivative of the nonlinearity is out_value * (1.0 - out_value);
   CuMatrix<BaseFloat> temp_deriv(out_value.NumRows(), out_value.NumCols(),
                                  kUndefined);
   temp_deriv.Set(1.0);
   temp_deriv.AddMat(-1.0, out_value);
   temp_deriv.MulElements(out_value);
   StoreStatsInternal(out_value, &temp_deriv);
 }

◆ Type()

virtual std::string Type ( ) const

inlinevirtual

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

Implements Component.

Definition at line 226 of file nnet-simple-component.h.

226 { return "SigmoidComponent"; }

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

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ SigmoidComponent() [1/2]

◆ SigmoidComponent() [2/2]

Member Function Documentation

◆ Backprop()

◆ Copy()

◆ operator=()

◆ Propagate()

◆ Properties()

◆ RepairGradients()

◆ StoreStats()

◆ Type()