Elementwise gate over two same-shaped inputs: out = TGate(gate) * value. More...

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

Public Types
using	ComponentBase = Component<TDeviceType, TPrecision>
using	MR = typename DeviceTypeTraits<TDeviceType>::memory_resource
using	TensorType = Tensor<TPrecision, MR>

Public Member Functions
	AttentionOutputGate (const std::string &name, const AttentionOutputGateConfig &config=AttentionOutputGateConfig(), std::optional< DeviceId > device_id=std::nullopt)
TensorType &	forward (const TensorType &gate, const TensorType &value)
	out = TGate(gate) * value.
DeviceId	getDeviceId () const override
	Get the compute device id associated with this component.
std::vector< ITensor * >	getGradients () const override
	Return non-owning pointers to parameter gradient tensors.
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.
std::vector< const ITensor * >	getOutputs () const override
	Return non-owning pointers to the tensors this component produces.
std::vector< ITensor * >	getParameters () const override
	Return non-owning pointers to parameter tensors.
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	installSharedOutput (std::shared_ptr< TensorType > output)
	Install a shared output slot (activation pooling).
dim_t	parameterCount () const override
	Return number of trainable parameters.
void	save_ (ModelArchive &archive, SerializationMode) const override
void	synchronize () override
	Wait for outstanding device work submitted by this component.
std::string	toString () const override
	Produce a short, human-readable description of the component.
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< 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	load_ (ModelArchive &archive, SerializationMode mode)
	Restore this component's parameters from its archive scope.
virtual void	loadParameter (const std::string &, const Serialization::ITensorBlob &)
	Load a parameter from serialized tensor data.
virtual void	requireSerializableParameters () const
	Verify this component can serialize whatever parameters it owns.
virtual void	saveFlatTensors (Serialization::SafeTensorsWriter &writer, const std::string &prefix, Serialization::TensorSavePass pass) const
	Drive this component's tensors through one pass of a flat safetensors save.
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.
virtual void	zeroGradients ()
	Clear all model-owned gradients for this component.

Static Public Member Functions
static constexpr ActivationType	getGateType () noexcept
	The compile-time gate function realized by this component.
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 Member Functions
void	onBuilding (const BuildContext &build_context) override
	Hook invoked by build() to allocate component buffers.
void	onExecutionContextSet () override
	Lifecycle hook: Called immediately after ExecutionContext is set.
void	onTrainingModeChanging (TrainingMode training_mode) override
	Hook called before TrainingMode transitions.
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
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>
requires PrecisionSupportedOnDevice<TPrecision, TDeviceType>
class Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >

Elementwise gate over two same-shaped inputs: out = TGate(gate) * value.

Template Parameters

TDeviceType	Compile-time device.
TPrecision	Activation/compute precision.
TGate	Gate function, fixed at compile time. SiLU is the default because it is the common case across gated architectures – NOT because of Qwen 3.8, whose output_gate_type: "swish" is a dead config key its reference implementation never reads. That block passes Sigmoid.

Stateless: no trainable parameters. The gate WEIGHTS are the second half of the query projection and belong to that Linear, not here – which is why this component owns nothing and the parameter budget in Qwen3.8.md section 2 counts the gate under "q (+gate)".

Two launches, not one: the activation runs through the shared ElementwiseActivationOp and the product through the TensorOps multiply, in place over this component's own output. Fusing them into a single kernel would need a new operation type and a dispatch row for a shape that runs on 16 of 64 layers; worth doing only if it ever measures, and the composition is the correct starting point either way.

Member Function Documentation

◆ forward()

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

TensorType & Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >::forward	(	const TensorType &	gate,
		const TensorType &	value )

inline

out = TGate(gate) * value.

Parameters

gate	The gate half of the query projection, [B, T, num_heads * head_dim].
value	The attention output, same shape.

The activation writes into this component's output and the product then runs in place over it, which is safe because both are elementwise. Doing it the other way – activating in place over gate – would corrupt a caller-owned buffer that on the pooled path is shared with every other layer.

◆ getDeviceId()

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

DeviceId Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >::getDeviceId ( ) const

inlineoverridevirtual

Get the compute device id associated with this component.

Must return the device on which parameters and operations execute.

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

◆ getGradients()

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

std::vector< ITensor * > Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >::getGradients ( ) const

inlineoverridevirtual

Return non-owning pointers to parameter gradient tensors.

Gradient buffers are allocated only when the component is built in training mode, so a component built for inference returns an empty vector. Stateless components return empty in either mode. This is the accessor counterpart to getParameters() and does not throw on mode.

Returns: Vector of gradient pointers; empty when built for inference.

Exceptions

std::runtime_error if called before the component has been built.

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

◆ getMemoryStats()

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

MemoryStats Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >::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>

std::vector< ObservableStage > Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >::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 >.

◆ getOutputs()

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

std::vector< const ITensor * > Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >::getOutputs ( ) const

inlineoverridevirtual

Return non-owning pointers to the tensors this component produces.

The counterpart to getParameters(): that reports what a component HOLDS, this reports what it PRODUCES. Tensors are the built allocations, so their extents are the ceiling a forward pass may narrow within – an observer wanting the live value of a particular call receives it through publication instead.

An empty result means the component does not describe itself. That absence cannot be mistaken for a finding – a view renders no row for it – so no distinction beyond empty is offered.

Returns: Vector of output pointers; empty before build, and for undescribed components.

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

◆ getParameters()

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

std::vector< ITensor * > Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >::getParameters ( ) const

inlineoverridevirtual

Return non-owning pointers to parameter tensors.

The returned tensor pointers remain valid for the lifetime of the component. Order should be canonical (weights before biases).

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

◆ getRequiredMemory()

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

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

inlineoverridevirtual

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

The build context carries ONE input shape; gate and value are the same shape, so the output matches it directly. See Specifications/MemoryFootprint.md.

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

◆ getType()

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

const ComponentType Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >::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 >.

◆ installSharedOutput()

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

void Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >::installSharedOutput ( std::shared_ptr< TensorType > output )

inline

Install a shared output slot (activation pooling).

Must be called before build(). Mirrors Swiglu::installSharedOutput: forward() always returns a shape-adjusted view, so a wider slot never leaks its geometry. The slot is owned and memory-accounted by the installer.

◆ onBuilding()

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

void Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >::onBuilding ( const BuildContext & build_context )

inlineoverrideprotectedvirtual

Hook invoked by build() to allocate component buffers.

Receives the stored BuildContext. Implementations must use config.allocationSeqLen() when sizing output buffers – this is the single call that makes Inference and Training allocate the correct buffer sizes automatically without per-component logic.

// Example -- Linear component:
shape_t out_shape =
{
    config.batchSize(),
    config.allocationSeqLen(),   // 1 for Inference, T for Training
    config_.getOutputFeatures()
};
output_ = std::make_unique<TensorType>( device, out_shape,
    this->getName() + ".output" );

The default implementation forwards to the legacy onBuilding( const shape_t& ) overload for backwards compatibility. New components should override this overload directly.

Note: Do not call build() or onBuilding() from within this hook.; Implementations should either succeed fully or leave no partial state, as a failed build() may be retried.

Takes the build-time configuration; use its allocationSeqLen() to obtain the correct output buffer sequence dimension.

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

◆ onExecutionContextSet()

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

void Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >::onExecutionContextSet ( )

inlineoverrideprotectedvirtual

Lifecycle hook: Called immediately after ExecutionContext is set.

Override this to perform initialization that requires a valid ExecutionContext. At the time this is called, getExecutionContext() is guaranteed to return a valid context.

Common uses:

Composite components: Create and configure child components.
Device resource allocation: Query device capabilities.

Default implementation does nothing.

Exceptions

Any	exception thrown will cause setExecutionContext() to fail and restore the component to a "context not set" state.

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

◆ onTrainingModeChanging()

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

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

inlineoverrideprotectedvirtual

Hook called before TrainingMode transitions.

Called by setTrainingMode() after validation and lock acquisition, before the internal state is updated. Derived classes override to respond to the transition – e.g. zeroing gradient buffers on transition to Eval, or re-enabling dropout on transition to Training.

The default implementation is a no-op.

Takes the incoming TrainingMode.

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

◆ parameterCount()

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

dim_t Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >::parameterCount ( ) const

inlineoverridevirtual

Return number of trainable parameters.

For leaf components this is the element count of owned parameter tensors. CompositeComponent and Network implementations should return the recursive aggregate across all children.

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

◆ save_()

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

void Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >::save_	(	ModelArchive &	archive,
		SerializationMode	) const

inlineoverridevirtual

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

◆ synchronize()

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

void Mila::Dnn::AttentionOutputGate< TDeviceType, TPrecision, TGate >::synchronize ( )

inlineoverridevirtual

Wait for outstanding device work submitted by this component.

On CPU this may be a no-op. Use to ensure results are visible to the host or to measure synchronous timings.

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

◆ toString()

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

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

inlineoverridevirtual

Produce a short, human-readable description of the component.

Implementations should keep output concise and avoid throwing.

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

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

Mila/Src/Dnn/Components/Attention/OutputGate/AttentionOutputGate.ixx

Public Types

Public Member Functions

Static Public Member Functions

Protected Member Functions

Additional Inherited Members

Detailed Description

Member Function Documentation

◆ forward()

◆ getDeviceId()

◆ getGradients()

◆ getMemoryStats()

◆ getObservableStages()

◆ getOutputs()

◆ getParameters()

◆ getRequiredMemory()

◆ getType()

◆ installSharedOutput()

◆ onBuilding()

◆ onExecutionContextSet()

◆ onTrainingModeChanging()

◆ parameterCount()

◆ save_()

◆ synchronize()

◆ toString()