dlprim::TensorSpecs Class Reference

Definition of Tensor without actual memory/object. More...

#include <include/dlprim/tensor.hpp>

Public Member Functions
	TensorSpecs (Shape const &s=Shape(), DataType d=float_data, bool trainable=true)
	Specifications defined by shape, data type,. More...
bool	operator== (TensorSpecs const &other) const
bool	operator!= (TensorSpecs const &other) const
Shape const &	shape () const
	get tensor shape
void	shape (Shape const &s)
bool	is_trainable () const
	return if tensor need to participate in gradient decent
void	freeze ()
	Mark tensor as one that does not participate in gradients calculations.
void	is_trainable (bool v)
	set - non-trainable property
size_t	memory_size () const
	Get reuired memory size for the tensor.
DataType	dtype () const

Friends
class	Tensor

Detailed Description

Definition of Tensor without actual memory/object.

Constructor & Destructor Documentation

dlprim::TensorSpecs::TensorSpecs ( Shape const &  s = Shape(), DataType  d = float_data, bool  trainable = true  )

inline

Specifications defined by shape, data type,.

flag trainable marks that specific tensor must not participate in gradient decent calculations it is non-trainable parameter - for example Batch Normalization's running_var/running_mean

References dlprim::is_floating_point_data_type().

---

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

• include/dlprim/tensor.hpp