Dropout regularization module for neural networks. More...

Inheritance diagram for Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >:

Collaboration diagram for Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >:

[legend]

Public Types
using	ModuleBase = Module< TDeviceType, TInput, TOutput >
	Alias for base module type.

using	MR = std::conditional_t< TDeviceType==DeviceType::Cuda, CudaMemoryResource, CpuMemoryResource >
	Memory resource type used for tensors, selected based on device type.

Public Types inherited from Mila::Dnn::Module< TDeviceType, TInput, TOutput >
using	MR = std::conditional_t< TDeviceType==DeviceType::Cuda, CudaMemoryResource, CpuMemoryResource >

Public Member Functions
	Dropout (const std::string &device_name, const DropoutConfig &config)
	Constructs a new Dropout module with a device name.

	Dropout (std::shared_ptr< DeviceContext > device_context, const DropoutConfig &config)
	Constructs a new Dropout module with a provided device context.

void	backward (const Tensor< TInput, MR > &input, const Tensor< TOutput, MR > &output_grad, Tensor< TInput, MR > &input_grad)
	Performs the backward pass of the Dropout operation.

void	forward (const Tensor< TInput, MR > &input, Tensor< TOutput, MR > &output)
	Performs the forward pass of the Dropout operation.

float	getProbability () const
	Gets the dropout probability used by this module.

unsigned int	getSeed () const
	Gets the current random seed.

void	load (ModelArchive &archive) override
	Deserializes the module state from a ZIP archive.

size_t	parameterCount () const override
	Gets the number of trainable parameters in this module.

void	save (ModelArchive &archive) const override
	Serializes the module state to a ZIP archive.

void	setSeed (unsigned int seed)
	Sets the random seed for dropout mask generation.

std::string	toString () const override
	Generates a string representation of this module's configuration.

Public Member Functions inherited from Mila::Dnn::Module< TDeviceType, TInput, TOutput >
	Module (const std::string &device_name, const ComponentConfig &config)
	Constructor with device name.

	Module (std::shared_ptr< DeviceContext > context, const ComponentConfig &config)
	Constructor with a specific device context.

virtual	~Module ()=default
	Virtual destructor for proper cleanup in derived classes.

std::shared_ptr< Compute::DeviceContext >	getDeviceContext () const
	Get the device context for this module.

Compute::DeviceType	getDeviceType () const
	Get the device type of the current device context.

std::string	getName () const
	Get the name of the module.

const auto &	getParameterTensors () const
	Get the parameter tensors of this module.

const ComputePrecision::Policy &	getPrecision () const

const auto &	getStateTensors () const
	Get the state tensors of this module.

bool	isTraining () const
	Check if the module is in training mode.

virtual void	setTraining (bool is_training)
	Set the training mode of this module.

Private Member Functions
void	createOperation ()
	Creates the appropriate Dropout operation based on the current device context.

void	generateMask (Tensor< TOutput, MR > &mask, const std::vector< size_t > &shape)
	Generates a new dropout mask for the given shape.

Private Attributes
DropoutConfig	config_
	Configuration for the Dropout module.

std::shared_ptr< Tensor< TOutput, MR > >	mask_ { nullptr }
	The binary mask tensor for element selection.

std::shared_ptr< UnaryOperation< TDeviceType, TInput, TOutput > >	operation_ { nullptr }
	The operation that implements the dropout calculation.

std::vector< std::shared_ptr< Tensor< TOutput, MR > > >	output_state_
	Collection of output state tensors for caching.

std::vector< std::shared_ptr< Tensor< TOutput, MR > > >	parameters_
	Collection of parameters for this module (empty for Dropout).

OperationAttributes	properties_
	Operation attributes and configuration.

std::mt19937	rng_
	Random number generator for mask generation.

unsigned int	seed_ { 0 }
	Random seed for reproducible dropout patterns.

Additional Inherited Members
Protected Member Functions inherited from Mila::Dnn::Module< TDeviceType, TInput, TOutput >
const std::string	parametersToString () const
	Helper method to convert parameters to string representation.

const std::string	stateToString () const
	Helper method to convert state tensors to string representation.

Protected Attributes inherited from Mila::Dnn::Module< TDeviceType, TInput, TOutput >
std::unordered_map< std::string, std::shared_ptr< Tensor< TOutput, MR > > >	parameter_map_ = {}
	Map of parameter names to parameter tensors.

std::unordered_map< std::string, std::shared_ptr< Tensor< TOutput, MR > > >	state_map_ = {}
	Map of state names to state tensors.

Detailed Description

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>
requires ValidFloatTensorTypes<TInput, TOutput>
class Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >

Dropout regularization module for neural networks.

Dropout is a regularization technique that randomly zeroes elements of an input tensor during training with probability p, and optionally scales the remaining elements by 1/(1-p). During inference, dropout is typically disabled.

Dropout helps prevent overfitting by preventing co-adaptation of feature detectors. The technique effectively trains an ensemble of multiple networks sharing parameters, which improves generalization.

Template Parameters

TDeviceType	The device type (CPU or CUDA) on which the module will operate.
TInput	The data type of the input tensor elements.
TOutput	The data type of the output tensor elements, defaults to TInput.

Member Typedef Documentation

◆ ModuleBase

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

using Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::ModuleBase = Module<TDeviceType, TInput, TOutput>

export

Alias for base module type.

◆ MR

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

using Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::MR = std::conditional_t<TDeviceType == DeviceType::Cuda, CudaMemoryResource, CpuMemoryResource>

export

Memory resource type used for tensors, selected based on device type.

Constructor & Destructor Documentation

◆ Dropout() [1/2]

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::Dropout	(	const std::string &	device_name,
		const DropoutConfig &	config
	)

inlineexplicitexport

Constructs a new Dropout module with a device name.

Creates a new DeviceContext internally using the provided device name. This constructor is useful for creating standalone modules without pre-existing device contexts.

Parameters

device_name	The name of the device to use (e.g., "CPU", "CUDA:0").
config	Configuration parameters for the Dropout module.

Exceptions

std::invalid_argument	If the device name is invalid or the configuration is invalid
std::runtime_error	If device type doesn't match template parameter TDeviceType

Here is the call graph for this function:

◆ Dropout() [2/2]

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::Dropout	(	std::shared_ptr< DeviceContext >	device_context,
		const DropoutConfig &	config
	)

inlineexplicitexport

Constructs a new Dropout module with a provided device context.

Uses a pre-existing DeviceContext instance. This constructor is useful when integrating the module into a larger network that shares device contexts across modules.

Parameters

device_context	The device context to use for this module.
config	Configuration parameters for the Dropout module.

Exceptions

std::invalid_argument	If device_context is null or configuration is invalid
std::runtime_error	If device context type doesn't match template parameter TDeviceType

Here is the call graph for this function:

Member Function Documentation

◆ backward()

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

void Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::backward	(	const Tensor< TInput, MR > &	input,
		const Tensor< TOutput, MR > &	output_grad,
		Tensor< TInput, MR > &	input_grad
	)

inlineexport

Performs the backward pass of the Dropout operation.

Computes the gradient of the Dropout function with respect to its input. During training, gradients are only propagated for non-zeroed elements. The gradient computation is straightforward: multiply the output gradient by the same mask used in the forward pass.

Parameters

input	The input tensor from the forward pass.
output_grad	The gradient of loss with respect to the output.
input_grad	The tensor to store gradients with respect to input.

◆ createOperation()

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

void Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::createOperation ( )

inlineexportprivate

Creates the appropriate Dropout operation based on the current device context.

This method initializes the operation_ member with the appropriate implementation of the Dropout operation for either CPU or CUDA, based on the current device context. It also passes the config object to the operation.

Here is the call graph for this function:

Here is the caller graph for this function:

◆ forward()

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

void Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::forward	(	const Tensor< TInput, MR > &	input,
		Tensor< TOutput, MR > &	output
	)

inlineexport

Performs the forward pass of the Dropout operation.

During training, randomly zeroes elements with probability p and scales remaining elements by 1/(1-p). During inference, performs identity operation or scaling based on configuration.

Parameters

input	The input tensor.
output	The output tensor where the results will be stored.

Here is the call graph for this function:

◆ generateMask()

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

void Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::generateMask	(	Tensor< TOutput, MR > &	mask,
		const std::vector< size_t > &	shape
	)

inlineexportprivate

Generates a new dropout mask for the given shape.

Creates a binary mask with elements set to 0 (drop) with probability p, and scale factor 1/(1-p) (keep) with probability (1-p).

Parameters

mask	The tensor to populate with the mask values
shape	The shape of the input/mask tensor

Here is the call graph for this function:

Here is the caller graph for this function:

◆ getProbability()

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

float Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::getProbability ( ) const

inlineexport

Gets the dropout probability used by this module.

Returns: float The probability of zeroing elements

Here is the call graph for this function:

◆ getSeed()

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

unsigned int Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::getSeed ( ) const

inlineexport

Gets the current random seed.

Returns: unsigned int The current random seed

◆ load()

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

void Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::load ( ModelArchive & archive )

inlineoverrideexportvirtual

Deserializes the module state from a ZIP archive.

Implementation of the Module interface for deserialization. Since Dropout has no learnable parameters, this is a no-op implementation.

Parameters

zip	ZIP archive for deserialization

Implements Mila::Dnn::Module< TDeviceType, TInput, TOutput >.

◆ parameterCount()

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

size_t Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::parameterCount ( ) const

inlineoverrideexportvirtual

Gets the number of trainable parameters in this module.

The Dropout module has no trainable parameters.

Returns: size_t Always returns 0 for Dropout.

Implements Mila::Dnn::Module< TDeviceType, TInput, TOutput >.

◆ save()

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

void Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::save ( ModelArchive & archive ) const

inlineoverrideexportvirtual

Serializes the module state to a ZIP archive.

Implementation of the Module interface for serialization. Since Dropout has no learnable parameters, this is a no-op implementation.

Parameters

zip	ZIP archive for serialization

Implements Mila::Dnn::Module< TDeviceType, TInput, TOutput >.

◆ setSeed()

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

void Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::setSeed ( unsigned int seed )

inlineexport

Sets the random seed for dropout mask generation.

This allows for reproducible dropout patterns when needed.

Parameters

seed	The random seed value

◆ toString()

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

std::string Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::toString ( ) const

inlineoverrideexportvirtual

Generates a string representation of this module's configuration.

Returns: std::string A formatted string with module information

Implements Mila::Dnn::Module< TDeviceType, TInput, TOutput >.

Here is the call graph for this function:

Member Data Documentation

◆ config_

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

DropoutConfig Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::config_

exportprivate

Configuration for the Dropout module.

◆ mask_

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

std::shared_ptr<Tensor<TOutput, MR> > Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::mask_ { nullptr }

exportprivate

The binary mask tensor for element selection.

Contains 0 for dropped elements and scale factor for kept elements.

◆ operation_

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

std::shared_ptr<UnaryOperation<TDeviceType, TInput, TOutput> > Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::operation_ { nullptr }

exportprivate

The operation that implements the dropout calculation.

◆ output_state_

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

std::vector<std::shared_ptr<Tensor<TOutput, MR> > > Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::output_state_

exportprivate

Collection of output state tensors for caching.

◆ parameters_

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

std::vector<std::shared_ptr<Tensor<TOutput, MR> > > Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::parameters_

exportprivate

Collection of parameters for this module (empty for Dropout).

◆ properties_

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

OperationAttributes Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::properties_

exportprivate

Operation attributes and configuration.

◆ rng_

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

std::mt19937 Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::rng_

exportprivate

Random number generator for mask generation.

◆ seed_

template<DeviceType TDeviceType = DeviceType::Cuda, typename TInput = float, typename TOutput = TInput>

unsigned int Mila::Dnn::Dropout< TDeviceType, TInput, TOutput >::seed_ { 0 }

exportprivate

Random seed for reproducible dropout patterns.

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

/home/runner/work/Mila/Mila/Mila/Src/Dnn/Modules/Regularization/Dropout.ixx

Public Types

Public Member Functions

Private Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

Member Typedef Documentation

◆ ModuleBase

◆ MR

Constructor & Destructor Documentation

◆ Dropout() [1/2]

◆ Dropout() [2/2]

Member Function Documentation

◆ backward()

◆ createOperation()

◆ forward()

◆ generateMask()

◆ getProbability()

◆ getSeed()

◆ load()

◆ parameterCount()

◆ save()

◆ setSeed()

◆ toString()

Member Data Documentation

◆ config_

◆ mask_

◆ operation_

◆ output_state_

◆ parameters_

◆ properties_

◆ rng_

◆ seed_