magmadnn::model::Model< T > Class Template Reference

Inheritance diagram for magmadnn::model::Model< T >:

Collaboration diagram for magmadnn::model::Model< T >:

Public Member Functions
	Model ()
virtual magmadnn_error_t	fit (Tensor< T > *x, Tensor< T > *y, metric_t &metric_out, bool verbose=false)=0
virtual Tensor< T > *	predict (Tensor< T > *sample)=0
virtual unsigned int	predict_class (Tensor< T > *sample)=0
virtual double	get_accuracy ()
virtual double	get_loss ()
virtual double	get_training_time ()
virtual std::string	get_name ()

Protected Attributes
std::string	_name
metric_t	_last_training_metric

Constructor & Destructor Documentation

Model()

template<typename T >

magmadnn::model::Model< T >::Model ( )

inline

Creates a new model.

Member Function Documentation

fit()

template<typename T >

virtual magmadnn_error_t magmadnn::model::Model< T >::fit ( Tensor< T > *  x, Tensor< T > *  y, metric_t &  metric_out, bool  verbose = false  )

pure virtual

train this model on x and y.

Parameters

x	model input data
y	ground truth data
metric_out	[out] a metric_t struct in which the model run metrics are stored.
verbose	if true then training info is printed every couple iterations

Returns

magmadnn_error_t 0 on success; non-zero otherwise

Implemented in magmadnn::model::NeuralNetwork< T >.

get_accuracy()

template<typename T >

virtual double magmadnn::model::Model< T >::get_accuracy ( )

inlinevirtual

Gets the training accuracy of the last run.

Returns

double training accuracy of last run.

get_loss()

template<typename T >

virtual double magmadnn::model::Model< T >::get_loss ( )

inlinevirtual

Gets the loss of the last run.

Returns

double loss of last run.

get_name()

template<typename T >

virtual std::string magmadnn::model::Model< T >::get_name ( )

inlinevirtual

name of this model

Returns

std::string

get_training_time()

template<typename T >

virtual double magmadnn::model::Model< T >::get_training_time ( )

inlinevirtual

Gets training time of last run.

Returns

double training time of last run.

predict()

template<typename T >

virtual Tensor<T>* magmadnn::model::Model< T >::predict ( Tensor< T > *  sample )

pure virtual

return a one-hot encoded output of the network on this sample

Parameters

sample

a single sample

Returns

Tensor<T>* one-hot encoded prediction

Implemented in magmadnn::model::NeuralNetwork< T >.

predict_class()

template<typename T >

virtual unsigned int magmadnn::model::Model< T >::predict_class ( Tensor< T > *  sample )

pure virtual

This is equivalent to math::argmax(predict(sample)), where the one-hot encoded predict value is used.

Parameters

sample

Returns

unsigned int the index of the predicted class.

Implemented in magmadnn::model::NeuralNetwork< T >.

---

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

• include/model/model.h