Pure token embedding component (device-templated). More...

Inheritance diagram for Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >:

Collaboration diagram for Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >:

[legend]

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

Public Member Functions
	TokenEmbedding (const std::string &name, const TokenEmbeddingConfig &config, std::optional< DeviceId > device_id=std::nullopt)
	Construct a TokenEmbedding component.
	~TokenEmbedding () override=default
TokenIndexType &	backward (const TokenIndexType &input, const EmbeddingTensorType &output_grad)
	Backward pass — accumulates gradients into wte.
EmbeddingTensorType &	forward (const TokenIndexType &input)
	Forward pass — returns component-owned embeddings tensor.
DeviceId	getDeviceId () const override
	Get the compute device id associated with this component.
int64_t	getEmbeddingDim () const noexcept
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< ITensor * >	getParameters () const override
	Return non-owning pointers to parameter tensors.
const ComponentType	getType () const override
	Get the component type identifier.
int64_t	getVocabSize () const noexcept
EmbeddingTensorType *	getWteGrad () const noexcept
void	loadParameter (const std::string &name, const ITensorBlob &blob) override
	Load a parameter from serialized tensor data.
size_t	parameterCount () const override
	Return number of trainable parameters.
void	save_ (ModelArchive &archive, SerializationMode mode) 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.
void	zeroGradients () override
	Clear all model-owned gradients for this component.
Public Member Functions inherited from Mila::Dnn::Component< TDeviceType, dtype_t::FP32 >
	Component (const std::string &name)
	Construct component with required name identifier.
virtual	~Component ()=default
virtual void	build (const BuildContext &context) final
	Build the component with the provided BuildContext (canonical overload).
const std::string	getName () const
	Get the component's name identifier.
virtual std::vector< std::string >	getParameterNames () const
	List all available parameter names for this component.
RuntimeMode	getRuntimeMode () const noexcept
	Convenience accessor — true if currently in Eval mode.
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
void	setTrainingMode (TrainingMode mode)
	Set the runtime behavioral mode for this Component.

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, dtype_t::FP32 >
IExecutionContext *	getExecutionContext () const
	Get the shared execution context.
bool	hasExecutionContext () const noexcept
	Check if execution context has been set.
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	setExecutionContext (IExecutionContext *context)
	Set the execution context for this component.

Private Types
using	OpType = typename OperationTraits<OperationType::TokenEmbeddingOp, TDeviceType, TPrecision>::type

Private Member Functions
void	createOperation ()
void	initializeParameterGradients ()
void	initializeParameters ()
void	validateBuildContext (const BuildContext &context) const

Private Attributes
TokenEmbeddingConfig	config_
std::unique_ptr< EmbeddingTensorType >	current_output_view_ { nullptr }
std::unique_ptr< TokenIndexType >	input_grad_ { nullptr }
int64_t	max_batch_size_ { 0 }
int64_t	max_seq_len_ { 0 }
std::shared_ptr< OpType >	operation_ { nullptr }
std::unique_ptr< EmbeddingTensorType >	output_ { nullptr }
std::unique_ptr< IExecutionContext >	owned_exec_context_ { nullptr }
std::unique_ptr< EmbeddingTensorType >	wte_ { nullptr }
std::unique_ptr< EmbeddingTensorType >	wte_grad_ { nullptr }

Additional Inherited Members
Static Public Member Functions inherited from Mila::Dnn::Component< TDeviceType, dtype_t::FP32 >
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 Attributes inherited from Mila::Dnn::Component< TDeviceType, dtype_t::FP32 >
BuildContext	build_context_
	The BuildContext stored at build time.

Detailed Description

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>
requires PrecisionSupportedOnDevice<TPrecision, TDeviceType>
class Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >

Pure token embedding component (device-templated).

Transforms input token indices into continuous vector representations by looking up each index in the vocabulary embedding table (wte). No positional information is added here.

Construction modes:

Standalone: provide a DeviceId to create and own an ExecutionContext.
Child/deferred: omit DeviceId; caller must call setExecutionContext() before build().

Template Parameters

TDeviceType	Device type (DeviceType::Cpu or DeviceType::Cuda).
TIndex	Data type for token indices (typically INT32).
TPrecision	Tensor precision for embeddings (FP32 or FP16).

Constructor & Destructor Documentation

◆ TokenEmbedding()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::TokenEmbedding	(	const std::string &	name,
		const TokenEmbeddingConfig &	config,
		std::optional< DeviceId >	device_id = std::nullopt )

inlineexplicitexport

Construct a TokenEmbedding component.

Parameters

name	Component name identifier.
config	TokenEmbedding configuration.
device_id	Optional DeviceId for standalone (owned context) mode.

◆ ~TokenEmbedding()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::~TokenEmbedding ( )

overrideexportdefault

Member Function Documentation

◆ backward()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

TokenIndexType & Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::backward	(	const TokenIndexType &	input,
		const EmbeddingTensorType &	output_grad )

inlineexport

Backward pass — accumulates gradients into wte.

Token indices are discrete and non-differentiable; the returned input_grad tensor exists for interface consistency but carries no meaningful gradient.

wte_grad buffers use atomicAdd accumulation and must be zeroed before each backward call, which zeroGradients() handles.

Parameters

input	Token indices used in forward [B, T].
output_grad	Upstream gradient w.r.t. embeddings [B, T, C].

Returns: Reference to component-owned (unused) input gradient tensor.

Exceptions

std::runtime_error if not built, not in training mode, or buffers are not initialized.

◆ createOperation()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

void Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::createOperation ( )

inlineexportprivate

◆ forward()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

EmbeddingTensorType & Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::forward ( const TokenIndexType & input )

inlineexport

Forward pass — returns component-owned embeddings tensor.

output[b, t, :] = wte[ X[b, t], : ]

Accepts any sequence length T <= max built T, including T=1 for single-token autoregressive steps.

Parameters

input Token indices [B, T].

Returns: Reference to component-owned embeddings [B, T, C].

Exceptions

std::runtime_error if the component is not built.

◆ getDeviceId()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

DeviceId Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::getDeviceId ( ) const

inlineoverrideexportvirtual

Get the compute device id associated with this component.

Must return the device on which parameters and operations execute.

Implements Mila::Dnn::Component< TDeviceType, dtype_t::FP32 >.

◆ getEmbeddingDim()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

int64_t Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::getEmbeddingDim ( ) const

inlineexportnoexcept

◆ getGradients()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

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

inlineoverrideexportvirtual

Return non-owning pointers to parameter gradient tensors.

Only valid when isTraining() is true.

Exceptions

std::runtime_error if called when not in training mode or before the component has been built.

Implements Mila::Dnn::Component< TDeviceType, dtype_t::FP32 >.

◆ getMemoryStats()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

MemoryStats Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::getMemoryStats ( ) const

inlineoverrideexportvirtual

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 — parameters only After build( Inference ) — parameters + T=1 state buffers After build( Training ) — parameters + T=full state buffers After setEvaluation( false ) — 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, dtype_t::FP32 >.

◆ getParameters()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

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

inlineoverrideexportvirtual

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, dtype_t::FP32 >.

◆ getType()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

const ComponentType Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::getType ( ) const

inlineoverrideexportvirtual

Get the component type identifier.

Used for serialization and runtime type identification.

Returns: Component type enum value.

Implements Mila::Dnn::Component< TDeviceType, dtype_t::FP32 >.

◆ getVocabSize()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

int64_t Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::getVocabSize ( ) const

inlineexportnoexcept

◆ getWteGrad()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

EmbeddingTensorType * Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::getWteGrad ( ) const

inlineexportnoexcept

◆ initializeParameterGradients()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

void Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::initializeParameterGradients ( )

inlineexportprivate

◆ initializeParameters()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

void Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::initializeParameters ( )

inlineexportprivate

◆ loadParameter()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

void Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::loadParameter	(	const std::string &	name,
		const ITensorBlob &	blob )

inlineoverrideexportvirtual

Load a parameter from serialized tensor data.

Loads raw tensor bytes directly into an existing parameter tensor, handling precision conversion and device upload as needed.

The component validates that the blob's shape matches the parameter's expected shape, then delegates to the backend to perform:

Precision conversion (blob dtype → parameter dtype)
Device upload (CPU bytes → target device)

Parameters

name	Parameter name used to locate the target tensor.
blob	Serialized tensor metadata and raw bytes.

Exceptions

std::runtime_error	if component has no parameters to load.
std::runtime_error	if blob shape doesn't match parameter shape.

Reimplemented from Mila::Dnn::Component< TDeviceType, dtype_t::FP32 >.

◆ onBuilding()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

void Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::onBuilding ( const BuildContext & config )

inlineoverrideexportprotectedvirtual

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.

Parameters

config Build-time configuration. Use config.allocationSeqLen() to obtain the correct output buffer sequence dimension.

Reimplemented from Mila::Dnn::Component< TDeviceType, dtype_t::FP32 >.

◆ onExecutionContextSet()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

void Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::onExecutionContextSet ( )

inlineoverrideexportprotectedvirtual

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, dtype_t::FP32 >.

◆ onTrainingModeChanging()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

void Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::onTrainingModeChanging ( TrainingMode mode )

inlineoverrideexportprotectedvirtual

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.

Parameters

mode	The incoming TrainingMode.

Reimplemented from Mila::Dnn::Component< TDeviceType, dtype_t::FP32 >.

◆ parameterCount()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

size_t Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::parameterCount ( ) const

inlineoverrideexportvirtual

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, dtype_t::FP32 >.

◆ save_()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

void Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::save_	(	ModelArchive &	archive,
		SerializationMode	mode ) const

inlineoverrideexportvirtual

Implements Mila::Dnn::Component< TDeviceType, dtype_t::FP32 >.

◆ synchronize()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

void Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::synchronize ( )

inlineoverrideexportvirtual

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, dtype_t::FP32 >.

◆ toString()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

std::string Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::toString ( ) const

inlineoverrideexportvirtual

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

Implementations should keep output concise and avoid throwing.

Implements Mila::Dnn::Component< TDeviceType, dtype_t::FP32 >.

◆ validateBuildContext()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

void Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::validateBuildContext ( const BuildContext & context ) const

inlineexportprivate

◆ zeroGradients()

template<DeviceType TDeviceType, TensorDataType TIndex = dtype_t::INT32, TensorDataType TPrecision = dtype_t::FP32>

void Mila::Dnn::TokenEmbedding< TDeviceType, TIndex, TPrecision >::zeroGradients ( )

inlineoverrideexportvirtual

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, dtype_t::FP32 >.

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

/__w/Mila/Mila/Mila/Src/Dnn/Components/Embeddings/TokenEmbedding.ixx

Public Types

Public Member Functions

Protected Member Functions

Private Types

Private Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ TokenEmbedding()

◆ ~TokenEmbedding()

Member Function Documentation

◆ backward()

◆ createOperation()

◆ forward()

◆ getDeviceId()

◆ getEmbeddingDim()

◆ getGradients()

◆ getMemoryStats()

◆ getParameters()

◆ getType()

◆ getVocabSize()

◆ getWteGrad()

◆ initializeParameterGradients()

◆ initializeParameters()

◆ loadParameter()

◆ onBuilding()

◆ onExecutionContextSet()

◆ onTrainingModeChanging()

◆ parameterCount()

◆ save_()

◆ synchronize()

◆ toString()

◆ validateBuildContext()

◆ zeroGradients()