Gated feed-forward (GatedMLP) composite component. More...

Inheritance diagram for Mila::Dnn::GatedMLP< TDeviceType, TPrecision, TGate, TWeightQuantization >:

Public Types
using	ComponentPtr = typename CompositeComponentBase::ComponentPtr
using	CompositeComponentBase = CompositeComponent<TDeviceType, TPrecision>
using	LinearType = Linear<TDeviceType, TPrecision, TWeightQuantization>
using	MR = typename DeviceTypeTraits<TDeviceType>::memory_resource
using	SwigluType = Swiglu<TDeviceType, TPrecision, TGate>
using	TensorType = Tensor<TPrecision, MR>
Public Types inherited from Mila::Dnn::CompositeComponent< TDeviceType, TPrecision >
using	ComponentBase = Component<TDeviceType, TPrecision>
using	ComponentPtr = std::shared_ptr<Component<TDeviceType, TPrecision>>

Public Member Functions
	GatedMLP (const std::string &name, const GatedMLPConfig &config, std::optional< DeviceId > device_id=std::nullopt)
TensorType &	backward (const TensorType &input, const TensorType &output_grad)
TensorType &	decode (const TensorType &input) const
	Single-token inference convenience with no gradient capture.
TensorType &	forward (const TensorType &input)
	Forward pass: fc_gate_up -> gate -> fc_down.
MemoryStats	getMemoryStats () const override
	Return the current memory allocation breakdown for this component.
std::vector< ObservableStage >	getObservableStages () const override
	Stages this component publishes, and the passes each is published on.
MemoryStats	getRequiredMemory (const BuildContext &context) const override
	What onBuilding() would allocate for this context, without allocating.
const ComponentType	getType () const override
	Get the component type identifier.
void	installSharedOutputs (std::shared_ptr< TensorType > gate_up_output, std::shared_ptr< TensorType > gate_output, std::shared_ptr< TensorType > down_output)
	Install shared output slots into the three children (activation pooling).
std::string	toString () const override
	Generate a human-readable description.
void	zeroGradients () override
	Clear all model-owned gradients for this component.
Public Member Functions inherited from Mila::Dnn::CompositeComponent< TDeviceType, TPrecision >
	CompositeComponent (CompositeComponent &&) noexcept=default
	CompositeComponent (const CompositeComponent &)=delete
	CompositeComponent (const std::string &name)
	Construct composite component with name.
CompositeComponent &	addComponent (ComponentPtr component)
	Add a pre-constructed child component (chainable).
size_t	childCount () const noexcept
	Get the number of direct children.
void	clearComponents ()
	Clear all child components.
std::vector< std::string >	componentPaths () const
	Every component path in this subtree, in traversal order.
ComponentPtr	findComponent (const std::string &path) const
	Resolve a dot-separated component path within this composite.
ComponentPtr	getComponent (const std::string &name) const
	Retrieve a direct child component by name.
const std::vector< ComponentPtr > &	getComponents () const
	Get all child components in insertion order.
DeviceId	getDeviceId () const override
	Get the compute device for this composite.
std::vector< ITensor * >	getGradients () const override
	Get all parameter gradients from all children.
std::vector< ITensor * >	getParameters () const override
	Get all parameters from all children.
bool	hasChildren () const noexcept
	Check if this composite has any children.
bool	hasComponent (const std::string &name) const
	Check if a named child component exists.
size_t	observe (std::string_view pattern, Compute::ComputePassMask passes, Compute::ActivationObserver sink)
	Attach sink to every component in this subtree whose path matches.
CompositeComponent &	operator= (CompositeComponent &&) noexcept=default
CompositeComponent &	operator= (const CompositeComponent &)=delete
dim_t	parameterCount () const override
	Count parameters across all children.
bool	removeComponent (const std::string &name)
	Remove a child component by name.
void	saveFlatTensors (Serialization::SafeTensorsWriter &writer, const std::string &prefix, Serialization::TensorSavePass pass) const override
	Recurse into children, extending the flat dotted prefix.
void	stopObserving ()
	Detach every observer in this subtree and clear the shared sink.
void	synchronize () override
	Synchronize all child components.
ComponentPtr	tryFindComponent (const std::string &path) const
	Try to resolve a dot-separated component path within this composite.
Public Member Functions inherited from Mila::Dnn::Component< TDeviceType, TPrecision >
	Component (const std::string &name)
	Construct component with required name identifier.
virtual void	build (const BuildContext &context) final
	Build the component with the provided BuildContext (canonical overload).
IExecutionContext *	getExecutionContext () const
	Get the shared execution context.
const std::string	getName () const
	Get the component's name identifier.
Compute::ComputePassMask	getObservedPasses () const noexcept
virtual std::vector< const ITensor * >	getOutputs () const
	Return non-owning pointers to the tensors this component produces.
virtual std::vector< std::string >	getParameterNames () const
	List all available parameter names for this component.
RuntimeMode	getRuntimeMode () const noexcept
TrainingMode	getTrainingMode () const noexcept
	The current runtime behavioral mode of this Component.
virtual bool	isBuilt () const final
	Returns true if build() has completed successfully.
bool	isInferenceMode () const noexcept
bool	isTrainingMode () const noexcept
virtual void	loadParameter (const std::string &, const Serialization::ITensorBlob &)
	Load a parameter from serialized tensor data.
void	setObservedPasses (Compute::ComputePassMask passes) noexcept
	Select which passes this component publishes on.
void	setTrainingMode (TrainingMode mode)
	Set the runtime behavioral mode for this Component.

Protected Member Functions
void	onBuilding (const BuildContext &context) override
	Build child graph with the gated shape contract.
void	onTrainingModeChanging (TrainingMode training_mode) override
	Hook invoked when training mode is about to change.
Protected Member Functions inherited from Mila::Dnn::CompositeComponent< TDeviceType, TPrecision >
template<typename TComponent>
std::shared_ptr< TComponent >	getComponentAs (const std::string &name) const
	Retrieve a typed child component by name.
void	load_ (ModelArchive &archive, SerializationMode mode) override
	Restore children from their nested scopes, mirroring save_().
void	onExecutionContextSet () override
	Hook invoked after ExecutionContext is set.
virtual void	optimize ()
	Virtual hook for graph optimization after construction.
void	requireSerializableParameters () const override
	No-op override: a composite names no parameters of its own.
void	save_ (ModelArchive &archive, SerializationMode mode) const override
Protected Member Functions inherited from Mila::Dnn::Component< TDeviceType, TPrecision >
bool	hasExecutionContext () const noexcept
	Check if execution context has been set.
template<TensorDataType TParameterPrecision, typename TMemoryResource>
void	loadParameterFromBlob (const std::string &param_name, const Serialization::ITensorBlob &blob, Tensor< TParameterPrecision, TMemoryResource > &target, const shape_t &expected_shape)
	Load a tensor blob into a parameter tensor with validation.
void	publish (Compute::ComputePass pass, std::string_view stage, const ITensor &value) const
	Publish one activation to the observer, if this pass is being observed.
template<TensorDataType TParameterPrecision, typename TMemoryResource>
void	saveParameterToArchive (ModelArchive &archive, const std::string &parameter_name, const Tensor< TParameterPrecision, TMemoryResource > &parameter) const
	Write one parameter tensor into the archive under "tensors/<name>".
template<TensorDataType TParameterPrecision, typename TMemoryResource>
void	saveParameterToWriter (Serialization::SafeTensorsWriter &writer, const std::string &flat_name, const Tensor< TParameterPrecision, TMemoryResource > &parameter, Serialization::TensorSavePass pass) const
	Drive one parameter through one pass of a flat safetensors save.
void	setExecutionContext (IExecutionContext *context)
	Set the execution context for this component.

Additional Inherited Members
Static Public Member Functions inherited from Mila::Dnn::Component< TDeviceType, TPrecision >
static constexpr DeviceType	getDeviceType ()
	Compile-time device type for this component instance.
static constexpr TensorDataType	getPrecision () noexcept
	Compile-time tensor precision for this component instance.
Protected Types inherited from Mila::Dnn::Component< TDeviceType, TPrecision >
using	HostStagingMemoryResource = typename DeviceTypeTraits<TDeviceType>::host_staging_memory_resource
	Host memory a device-resident parameter stages through.
Protected Attributes inherited from Mila::Dnn::Component< TDeviceType, TPrecision >
BuildContext	build_context_ { shape_t{ 1 }, RuntimeMode::Training }
	The BuildContext stored at build time.

Detailed Description

template<DeviceType TDeviceType, TensorDataType TPrecision, ActivationType TGate = ActivationType::Silu, WeightQuantPolicy TWeightQuantization = NoWeightQuant>
requires PrecisionSupportedOnDevice<TPrecision, TDeviceType>
class Mila::Dnn::GatedMLP< TDeviceType, TPrecision, TGate, TWeightQuantization >

Gated feed-forward (GatedMLP) composite component.

Device-templated composite implementing the GLU (Gated Linear Unit) family of gated FFNs – SwiGLU (Llama, Qwen) and GeGLU (Gemma) – and the single-expert reference for a future MoE layer: Input -> fc_gate_up Linear(in -> 2H, fused) -> Swiglu gate (2H -> H) -> fc_down Linear(H -> in) -> Output

The gate multiplies TGate(gate half) by the up half: SiLU gives SwiGLU, GELU gives GeGLU. Both projections carry TWeightQuantization; the gate has no weights.

MoE-readiness seams (FfnAndMoE.md s9): the injected-context path is the norm (owned-context construction is a standalone convenience); forward operates on the trailing feature dimension so it is valid for both [B, T, in] and gathered [num_tokens, in] layouts.

Template Parameters

TDeviceType	Device type for execution.
TPrecision	Tensor data precision. Must be supported on the device.
TGate	Gate activation: Silu (SwiGLU) or Gelu (GeGLU, Gemma).
TWeightQuantization	Weight quantization policy of both projections.

Member Function Documentation

◆ forward()

template<DeviceType TDeviceType, TensorDataType TPrecision, ActivationType TGate = ActivationType::Silu, WeightQuantPolicy TWeightQuantization = NoWeightQuant>

TensorType & Mila::Dnn::GatedMLP< TDeviceType, TPrecision, TGate, TWeightQuantization >::forward ( const TensorType & input )

inline

Forward pass: fc_gate_up -> gate -> fc_down.

Captures non-owning pointers to child-owned intermediates for backward().

Parameters

input Input tensor [..., in_features].

Returns: Reference to the final fc_down output (owned by that child).

◆ getMemoryStats()

template<DeviceType TDeviceType, TensorDataType TPrecision, ActivationType TGate = ActivationType::Silu, WeightQuantPolicy TWeightQuantization = NoWeightQuant>

MemoryStats Mila::Dnn::GatedMLP< TDeviceType, TPrecision, TGate, TWeightQuantization >::getMemoryStats ( ) const

inlineoverridevirtual

Return the current memory allocation breakdown for this component.

Reflects allocations at the moment of the call. The returned stats naturally track the component lifecycle:

After construction – nothing; construction allocates none After build( Inference ) – parameters + T=1 state buffers After build( Training ) – parameters + T=full state buffers After setTrainingMode( Train ) – parameters + state + gradients

For CompositeComponent and Network, the returned stats are the recursive aggregate of all child components.

May be called at any time – no lifecycle preconditions.

Returns: MemoryStats reflecting current allocations.

Implements Mila::Dnn::Component< TDeviceType, TPrecision >.

◆ getObservableStages()

template<DeviceType TDeviceType, TensorDataType TPrecision, ActivationType TGate = ActivationType::Silu, WeightQuantPolicy TWeightQuantization = NoWeightQuant>

std::vector< ObservableStage > Mila::Dnn::GatedMLP< TDeviceType, TPrecision, TGate, TWeightQuantization >::getObservableStages ( ) const

inlineoverridevirtual

Stages this component publishes, and the passes each is published on.

Lets a caller tell an uninstrumented component from a clean one before running anything. The pass qualifier is load-bearing rather than decorative: the stage set genuinely differs by pass, since a prefill can produce intermediates its decode never materializes.

Returns: Vector of stages; empty for components that publish nothing.

Reimplemented from Mila::Dnn::Component< TDeviceType, TPrecision >.

◆ getRequiredMemory()

template<DeviceType TDeviceType, TensorDataType TPrecision, ActivationType TGate = ActivationType::Silu, WeightQuantPolicy TWeightQuantization = NoWeightQuant>

MemoryStats Mila::Dnn::GatedMLP< TDeviceType, TPrecision, TGate, TWeightQuantization >::getRequiredMemory ( const BuildContext & context ) const

inlineoverridevirtual

What onBuilding() would allocate for this context, without allocating.

Each child is sized at its own width rather than the generic recursion's single shape, exactly as onBuilding() builds it. Installed outputs are excluded by the children themselves, from their own installed flag or the context's.

Reimplemented from Mila::Dnn::Component< TDeviceType, TPrecision >.

◆ getType()

template<DeviceType TDeviceType, TensorDataType TPrecision, ActivationType TGate = ActivationType::Silu, WeightQuantPolicy TWeightQuantization = NoWeightQuant>

const ComponentType Mila::Dnn::GatedMLP< TDeviceType, TPrecision, TGate, TWeightQuantization >::getType ( ) const

inlineoverridevirtual

Get the component type identifier.

Used for serialization and runtime type identification.

Returns: Component type enum value.

Implements Mila::Dnn::Component< TDeviceType, TPrecision >.

◆ installSharedOutputs()

template<DeviceType TDeviceType, TensorDataType TPrecision, ActivationType TGate = ActivationType::Silu, WeightQuantPolicy TWeightQuantization = NoWeightQuant>

void Mila::Dnn::GatedMLP< TDeviceType, TPrecision, TGate, TWeightQuantization >::installSharedOutputs	(	std::shared_ptr< TensorType >	gate_up_output,
		std::shared_ptr< TensorType >	gate_output,
		std::shared_ptr< TensorType >	down_output )

inline

Install shared output slots into the three children (activation pooling).

Must be called before build(). Each slot is owned and memory-accounted by the installer, and must cover this component's build shape at its own width: [..., 2H] for the gate+up projection, [..., H] for the gate, [..., in] for the down projection. A pooling caller declares the same intent to getRequiredMemory() through BuildContext::withInstalledOutput().

◆ onBuilding()

template<DeviceType TDeviceType, TensorDataType TPrecision, ActivationType TGate = ActivationType::Silu, WeightQuantPolicy TWeightQuantization = NoWeightQuant>

void Mila::Dnn::GatedMLP< TDeviceType, TPrecision, TGate, TWeightQuantization >::onBuilding ( const BuildContext & context )

inlineoverrideprotectedvirtual

Build child graph with the gated shape contract.

fc_gate_up receives the input shape; the gate and fc_down receive the fused 2H and the gated H shapes respectively. The child contexts are derived with withShape so the prefill size, the installed-output declaration and the parameter-initialization choice reach the children unchanged.

Reimplemented from Mila::Dnn::Component< TDeviceType, TPrecision >.

◆ onTrainingModeChanging()

template<DeviceType TDeviceType, TensorDataType TPrecision, ActivationType TGate = ActivationType::Silu, WeightQuantPolicy TWeightQuantization = NoWeightQuant>

void Mila::Dnn::GatedMLP< TDeviceType, TPrecision, TGate, TWeightQuantization >::onTrainingModeChanging ( TrainingMode training_mode )

inlineoverrideprotectedvirtual

Hook invoked when training mode is about to change.

Propagates the new mode to all child components. The hook runs with the Component's training mutex held; it MUST NOT call setTrainingMode().

Parameters

training_mode New training mode (Normal or Eval)

Reimplemented from Mila::Dnn::CompositeComponent< TDeviceType, TPrecision >.

◆ toString()

template<DeviceType TDeviceType, TensorDataType TPrecision, ActivationType TGate = ActivationType::Silu, WeightQuantPolicy TWeightQuantization = NoWeightQuant>

std::string Mila::Dnn::GatedMLP< TDeviceType, TPrecision, TGate, TWeightQuantization >::toString ( ) const

inlineoverridevirtual

Generate a human-readable description.

Returns: String representation showing children

Reimplemented from Mila::Dnn::CompositeComponent< TDeviceType, TPrecision >.

◆ zeroGradients()

template<DeviceType TDeviceType, TensorDataType TPrecision, ActivationType TGate = ActivationType::Silu, WeightQuantPolicy TWeightQuantization = NoWeightQuant>

void Mila::Dnn::GatedMLP< TDeviceType, TPrecision, TGate, TWeightQuantization >::zeroGradients ( )

inlineoverridevirtual

Clear all model-owned gradients for this component.

Default implementation is a no-op. Composite components should override to recurse to children. Leaf components should override to zero their parameter and activation gradients using device-aware helpers.

Reimplemented from Mila::Dnn::Component< TDeviceType, TPrecision >.

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

Mila/Src/Dnn/Components/FFN/GatedMLP/GatedMLP.ixx

Public Types

Public Member Functions

Protected Member Functions

Additional Inherited Members

Detailed Description

Member Function Documentation

◆ forward()

◆ getMemoryStats()

◆ getObservableStages()

◆ getRequiredMemory()

◆ getType()

◆ installSharedOutputs()

◆ onBuilding()

◆ onTrainingModeChanging()

◆ toString()

◆ zeroGradients()