dismec::init::WeightsInitializer Class Reference

Base class for all weight initializers. More...

#include <initializer.h>

Inheritance diagram for dismec::init::WeightsInitializer:

Public Member Functions
virtual	~WeightsInitializer ()=default
virtual void	get_initial_weight (label_id_t label_id, Eigen::Ref< DenseRealVector > target, objective::Objective &objective)=0
	Generate an initial vector for the given label. The result should be placed in target. More...
Public Member Functions inherited from dismec::stats::Tracked
	Tracked ()
	Default constructor, creates the internal stats::StatisticsCollection. More...
void	register_stat (const std::string &name, std::unique_ptr< Statistics > stat)
	Registers a tracker for the statistics name. More...
std::shared_ptr< StatisticsCollection >	get_stats () const
	Gets an ownership-sharing reference to the StatisticsCollection. More...

Additional Inherited Members
Protected Member Functions inherited from dismec::stats::Tracked
	~Tracked ()
	Non-virtual destructor. Declared protected, so we don't accidentally try to do a polymorphic delete. More...
template<class T >
void	record (stat_id_t stat, T &&value)
	Record statistics. This function just forwards all its arguments to the internal StatisticsCollection. More...
void	declare_stat (stat_id_t index, StatisticMetaData meta)
	Declares a new statistics. This function just forwards all its arguments to the internal StatisticsCollection. More...
void	declare_tag (tag_id_t index, std::string name)
	Declares a new tag. This function just forwards all its arguments to the internal StatisticsCollection. More...
template<class... Args>
void	set_tag (tag_id_t tag, long value)
	Set value of tag. This function just forwards all its arguments to the internal StatisticsCollection. More...
template<class... Args>
auto	make_timer (stat_id_t id, Args... args)
	Creates a new ScopeTimer using stats::record_scope_time. More...

Detailed Description

Base class for all weight initializers.

Weight initializers are used by TrainingTaskGenerator to generate the initial weight vector before the actual training process starts. A well-chosen initial vector can significantly decrease training time. Of course, this requires that the calculation of the initial vector itself be fast. For more discussion as well as the list of available initializers, see initvectors . Instances of WeightsInitializer will be created inside the training threads by a WeightInitializationStrategy. This ensure that each training thread has its own initializer, so the get_initial_weight() method does not need to be mutex-protected.

See also

WeightInitializationStrategy

Definition at line 30 of file initializer.h.

Constructor & Destructor Documentation

~WeightsInitializer()

virtual dismec::init::WeightsInitializer::~WeightsInitializer ( )

virtualdefault

Member Function Documentation

get_initial_weight()

virtual void dismec::init::WeightsInitializer::get_initial_weight ( label_id_t  label_id, Eigen::Ref< DenseRealVector >  target, objective::Objective &  objective  )

pure virtual

Generate an initial vector for the given label. The result should be placed in target.

Implemented in dismec::init::ZeroInitializer, dismec::init::PreTrainedInitializer, dismec::init::NumpyInitializer, dismec::init::MultiPosMeanInitializer< Sparse >, dismec::init::MeanOfFeaturesInitializer, dismec::init::ConstantInitializer, and anonymous_namespace{cascade.cpp}::CombinedWeightInitializer.

---

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

• src/training/initializer.h