Configuration class for MLP block. More...

Inheritance diagram for Mila::Dnn::MLPConfig:

Collaboration diagram for Mila::Dnn::MLPConfig:

Public Member Functions
	MLPConfig (const std::vector< size_t > &input_shape, size_t hidden_size)
	Constructor with required parameters.

	MLPConfig (size_t input_features, size_t hidden_size)
	Alternative constructor with direct input features specification.

ActivationType	getActivationType () const

float	getDropout () const

size_t	getHiddenSize () const

size_t	getInputFeatures () const

const std::vector< size_t > &	getInputShape () const

bool	hasBias () const

bool	useFusedOperations () const

bool	useLayerNorm () const

bool	useResidual () const

void	validate () const
	Validate configuration parameters.

MLPConfig &	withActivation (ActivationType activation)
	Set the activation function type.

MLPConfig &	withBias (bool has_bias)
	Configure whether the linear layers use bias.

MLPConfig &	withDropout (float dropout)
	Set the dropout rate.

MLPConfig &	withFusedOperations (bool fuse_ops)
	Configure whether to fuse operations for inference.

MLPConfig &	withLayerNorm (bool use_layer_norm)
	Configure whether to use layer normalization.

MLPConfig &	withResidual (bool use_residual)
	Configure whether to use a residual connection.

Public Member Functions inherited from Mila::Dnn::ComponentConfig
virtual	~ComponentConfig ()=default
	Virtual destructor to support proper polymorphic destruction.

const std::string &	getName () const
	Gets the configured component name.

ComputePrecision::Policy	getPrecision () const
	Gets the configured precision policy.

bool	isTraining () const
	Gets the configured training mode.

template<typename Self >
auto &	withName (this Self &&self, std::string name)
	Sets the name of the component with fluent interface.

template<typename Self >
auto &	withPrecision (this Self &&self, ComputePrecision::Policy policy)
	Sets the compute precision policy with fluent interface.

template<typename Self >
auto &	withTraining (this Self &&self, bool is_training)
	Sets the training mode with fluent interface.

Private Attributes
ActivationType	activation_type_ { ActivationType::Gelu}

float	dropout_ { 0.0f }

bool	fuse_operations_ { false }

bool	has_bias_ { true }

size_t	hidden_size_ { 0 }

size_t	input_features_ { 0 }

std::vector< size_t >	input_shape_

bool	use_layer_norm_ { false }

bool	use_residual_ { false }

Additional Inherited Members
Protected Attributes inherited from Mila::Dnn::ComponentConfig
bool	is_training_ = false
	Training mode flag, defaults to false (inference mode)

std::string	name_ = "unnamed"
	Component name, defaults to "unnamed" if not explicitly set.

ComputePrecision::Policy	precision_ = ComputePrecision::Policy::Auto
	Precision policy for computation, defaults to Auto.

Detailed Description

Configuration class for MLP block.

Constructor & Destructor Documentation

◆ MLPConfig() [1/2]

Mila::Dnn::MLPConfig::MLPConfig	(	const std::vector< size_t > &	input_shape,
		size_t	hidden_size
	)

inline

Constructor with required parameters.

Parameters

input_shape	The shape of the input tensor
hidden_size	The size of the hidden layer (intermediate dimension)

◆ MLPConfig() [2/2]

Mila::Dnn::MLPConfig::MLPConfig	(	size_t	input_features,
		size_t	hidden_size
	)

inline

Alternative constructor with direct input features specification.

Parameters

input_features	The number of input features
hidden_size	The size of the hidden layer (intermediate dimension)

Member Function Documentation

◆ getActivationType()

ActivationType Mila::Dnn::MLPConfig::getActivationType ( ) const

inline

Here is the caller graph for this function:

◆ getDropout()

float Mila::Dnn::MLPConfig::getDropout ( ) const

inline

Here is the caller graph for this function:

◆ getHiddenSize()

size_t Mila::Dnn::MLPConfig::getHiddenSize ( ) const

inline

Here is the caller graph for this function:

◆ getInputFeatures()

size_t Mila::Dnn::MLPConfig::getInputFeatures ( ) const

inline

Here is the caller graph for this function:

◆ getInputShape()

const std::vector< size_t > & Mila::Dnn::MLPConfig::getInputShape ( ) const

inline

Here is the caller graph for this function:

◆ hasBias()

bool Mila::Dnn::MLPConfig::hasBias ( ) const

inline

Here is the caller graph for this function:

◆ useFusedOperations()

bool Mila::Dnn::MLPConfig::useFusedOperations ( ) const

inline

Here is the caller graph for this function:

◆ useLayerNorm()

bool Mila::Dnn::MLPConfig::useLayerNorm ( ) const

inline

Here is the caller graph for this function:

◆ useResidual()

bool Mila::Dnn::MLPConfig::useResidual ( ) const

inline

Here is the caller graph for this function:

◆ validate()

void Mila::Dnn::MLPConfig::validate ( ) const

inlinevirtual

Validate configuration parameters.

Exceptions

std::invalid_argument If validation fails

Reimplemented from Mila::Dnn::ComponentConfig.

Here is the call graph for this function:

Here is the caller graph for this function:

◆ withActivation()

MLPConfig & Mila::Dnn::MLPConfig::withActivation ( ActivationType activation )

inline

Set the activation function type.

Parameters

activation The activation function to use

Returns: MLPConfig& Reference to this for method chaining

◆ withBias()

MLPConfig & Mila::Dnn::MLPConfig::withBias ( bool has_bias )

inline

Configure whether the linear layers use bias.

Parameters

has_bias Whether to include bias terms

Returns: MLPConfig& Reference to this for method chaining

◆ withDropout()

MLPConfig & Mila::Dnn::MLPConfig::withDropout ( float dropout )

inline

Set the dropout rate.

Parameters

dropout Dropout probability (0.0 to 1.0)

Returns: MLPConfig& Reference to this for method chaining

◆ withFusedOperations()

MLPConfig & Mila::Dnn::MLPConfig::withFusedOperations ( bool fuse_ops )

inline

Configure whether to fuse operations for inference.

Parameters

fuse_ops Whether to fuse operations when possible

Returns: MLPConfig& Reference to this for method chaining

◆ withLayerNorm()

MLPConfig & Mila::Dnn::MLPConfig::withLayerNorm ( bool use_layer_norm )

inline

Configure whether to use layer normalization.

Parameters

use_layer_norm Whether to use layer normalization

Returns: MLPConfig& Reference to this for method chaining

◆ withResidual()

MLPConfig & Mila::Dnn::MLPConfig::withResidual ( bool use_residual )

inline

Configure whether to use a residual connection.

Parameters

use_residual Whether to add a residual connection

Returns: MLPConfig& Reference to this for method chaining

Member Data Documentation

◆ activation_type_

ActivationType Mila::Dnn::MLPConfig::activation_type_ { ActivationType::Gelu}

private

◆ dropout_

float Mila::Dnn::MLPConfig::dropout_ { 0.0f }

private

◆ fuse_operations_

bool Mila::Dnn::MLPConfig::fuse_operations_ { false }

private

◆ has_bias_

bool Mila::Dnn::MLPConfig::has_bias_ { true }

private

◆ hidden_size_

size_t Mila::Dnn::MLPConfig::hidden_size_ { 0 }

private

◆ input_features_

size_t Mila::Dnn::MLPConfig::input_features_ { 0 }

private

◆ input_shape_

std::vector<size_t> Mila::Dnn::MLPConfig::input_shape_

private

◆ use_layer_norm_

bool Mila::Dnn::MLPConfig::use_layer_norm_ { false }

private

◆ use_residual_

bool Mila::Dnn::MLPConfig::use_residual_ { false }

private

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

/home/runner/work/Mila/Mila/Mila/Src/Dnn/Modules/Blocks/MLPConfig.ixx

Public Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ MLPConfig() [1/2]

◆ MLPConfig() [2/2]

Member Function Documentation

◆ getActivationType()

◆ getDropout()

◆ getHiddenSize()

◆ getInputFeatures()

◆ getInputShape()

◆ hasBias()

◆ useFusedOperations()

◆ useLayerNorm()

◆ useResidual()

◆ validate()

◆ withActivation()

◆ withBias()

◆ withDropout()

◆ withFusedOperations()

◆ withLayerNorm()

◆ withResidual()

Member Data Documentation

◆ activation_type_

◆ dropout_

◆ fuse_operations_

◆ has_bias_

◆ hidden_size_

◆ input_features_

◆ input_shape_

◆ use_layer_norm_

◆ use_residual_