dl::loss::CrossEntropyLoss< T > Class Template Reference

Cross Entropy Loss with autograd support. More...

#include <losses.hpp>

Inheritance diagram for dl::loss::CrossEntropyLoss< T >:

Collaboration diagram for dl::loss::CrossEntropyLoss< T >:

Public Member Functions
	CrossEntropyLoss (const std::string &reduction="mean")
	Constructor.
Variable< T >	forward (const Variable< T > &predictions, const Variable< T > &targets) override
	Forward pass: compute cross entropy loss.
Public Member Functions inherited from dl::loss::AutogradLoss< T >
virtual	~AutogradLoss ()=default
Variable< T >	operator() (const Variable< T > &predictions, const Variable< T > &targets)
	Convenience operator for computing loss.

Detailed Description

template<typename T>
class dl::loss::CrossEntropyLoss< T >

Cross Entropy Loss with autograd support.

CrossEntropy(y_pred, y_true) = -sum(y_true * log(softmax(y_pred)))

Commonly used for multi-class classification.

Definition at line 83 of file losses.hpp.

Constructor & Destructor Documentation

CrossEntropyLoss()

template<typename T >

dl::loss::CrossEntropyLoss< T >::CrossEntropyLoss ( const std::string &  reduction = "mean" )

inlineexplicit

Constructor.

Parameters

reduction

Type of reduction ('mean', 'sum', 'none')

Definition at line 89 of file losses.hpp.

Member Function Documentation

forward()

template<typename T >

Variable< T > dl::loss::CrossEntropyLoss< T >::forward ( const Variable< T > &  predictions, const Variable< T > &  targets  )

overridevirtual

Forward pass: compute cross entropy loss.

Parameters

predictions	Raw logits (before softmax)
targets	Target class indices or one-hot vectors

Returns

Cross entropy loss

Implements dl::loss::AutogradLoss< T >.

Definition at line 61 of file losses.cpp.

---

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

• include/loss/losses.hpp
• src/loss/losses.cpp