LLaMA-style transformer (decoder-only) for autoregressive token prediction. More...

Inheritance diagram for Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >:

[legend]

Collaboration diagram for Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >:

[legend]

Public Types
using	ComponentPtr = typename NetworkBase::ComponentPtr
using	LinearType = Linear<TDeviceType, TPrecision, TWeightQuantization>
using	LmHeadLinearType = Linear<TDeviceType, TPrecision>
using	MR = typename DeviceTypeTraits<TDeviceType>::memory_resource
using	NetworkBase = LanguageNetwork<TDeviceType, TPrecision>
using	RmsNormType = RmsNorm<TDeviceType, TPrecision>
using	TensorType = Tensor<TPrecision, MR>
using	TokenEmbeddingType = TokenEmbedding<TDeviceType, dtype_t::INT32, TPrecision>
using	TokenIndexType = Tensor<dtype_t::INT32, MR>
using	TransformerBlockType = LlamaBlock<TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy>
Public Types inherited from Mila::Dnn::LanguageNetwork< TDeviceType, TPrecision >
using	MR = typename DeviceTypeTraits<TDeviceType>::memory_resource
using	NetworkBase = Network<TDeviceType, TPrecision>
using	TensorType = Tensor<TPrecision, MR>
using	TokenIndexType = Tensor<TensorDataType::INT32, MR>
Public Types inherited from Mila::Dnn::Network< TDeviceType, TPrecision >
using	ComponentPtr = typename CompositeBase::ComponentPtr
using	CompositeBase = CompositeComponent<TDeviceType, TPrecision>
using	MR = typename DeviceTypeTraits<TDeviceType>::memory_resource
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
	LlamaTransformer (const std::string &name, const LlamaConfig &config, DeviceId device_id)
	~LlamaTransformer () override=default
TokenIndexType &	backward (const TokenIndexType &input, const TensorType &output_grad) override
TensorType &	decode (const TokenIndexType &input, int position) override
TensorType &	forward (const TokenIndexType &input) override
IExecutionContext *	getExecutionContext () const
MemoryStats	getMemoryStats () const override
	Return the current memory allocation breakdown for this component.
const ComponentType	getType () const override
	Get the component type identifier.
void	loadParameters (PretrainedModelReader &reader)
TensorType &	prefill (const TokenIndexType &input) override
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::LanguageNetwork< TDeviceType, TPrecision >
	LanguageNetwork (const std::string &name)
	~LanguageNetwork () override=default
virtual TokenIndexType &	backward (const TokenIndexType &input, const TensorType &output_grad)=0
	Full backward pass (training).
virtual TensorType &	decode (const TokenIndexType &input, int position)=0
	Inference decode — single-token autoregressive step.
virtual TensorType &	forward (const TokenIndexType &input)=0
	Full-sequence forward pass.
virtual TensorType &	prefill (const TokenIndexType &input)=0
	Inference prefill — process full prompt and populate the KV cache.
Public Member Functions inherited from Mila::Dnn::Network< TDeviceType, TPrecision >
	Network (const std::string &name)
	Construct network (context managed by derived class).
	~Network () override=default
template<typename TOptimizer, typename TConfig>
std::shared_ptr< TOptimizer >	createOptimizer (const TConfig &config)
	Create and configure an optimizer for this network's parameters.
DeviceId	getDeviceId () const noexcept
	Get the compute device for this composite.
const ComponentType	getType () const override
	Get the component type identifier.
void	save (ModelArchive &archive, SerializationMode mode) const
	Save network to archive.
void	synchronize () override
	Synchronize all child components.
std::string	toString () const override
	Generate a human-readable description.
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.
virtual	~CompositeComponent ()=default
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.
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.
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.
CompositeComponent &	operator= (CompositeComponent &&) noexcept=default
CompositeComponent &	operator= (const CompositeComponent &)=delete
size_t	parameterCount () const override
	Count parameters across all children.
bool	removeComponent (const std::string &name)
	Get the named child components map.
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	~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
virtual void	loadParameter (const std::string &name, const Serialization::ITensorBlob &blob)
	Load a parameter from serialized tensor data.
void	setTrainingMode (TrainingMode mode)
	Set the runtime behavioral mode for this Component.

Protected Member Functions
void	onBuilding (const BuildContext &context) override
	Hook invoked by build() to allocate component buffers.
void	onTrainingModeChanging (TrainingMode training_mode) override
	Hook invoked when training mode is about to change.
void	save_ (ModelArchive &archive, SerializationMode) const override
	Hook for concrete classes to save type-specific state.
Protected Member Functions inherited from Mila::Dnn::Network< TDeviceType, TPrecision >
void	verifyArchitectureCompatibility (const PretrainedMetadata &metadata)
	Verify that imported model is compatible with network architecture.
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	onExecutionContextSet () override
	Hook invoked after ExecutionContext is set.
virtual void	optimize ()
	Virtual hook for graph optimization after construction.
Protected Member Functions inherited from Mila::Dnn::Component< TDeviceType, TPrecision >
IExecutionContext *	getExecutionContext () const
	Get the shared execution context.
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	setExecutionContext (IExecutionContext *context)
	Set the execution context for this component.

Private Member Functions
void	createGraph ()
std::pair< std::string, std::string >	parseParameterPath (const std::string &full_name) const
void	validateBuildContext (const BuildContext &context) const
void	validateLeadingShape (const shape_t &leading_shape) const

Static Private Member Functions
static LlamaConfig	createConfigFromMetadata (const PretrainedMetadata &metadata)

Private Attributes
int64_t	batch_size_ { 0 }
std::vector< TensorType * >	block_input_ptrs_
std::vector< TensorType * >	block_output_ptrs_
LlamaConfig	config_
shape_t	embedding_shape_
std::unique_ptr< IExecutionContext >	exec_context_ { nullptr }
std::shared_ptr< RmsNormType >	final_rmsnorm_ { nullptr }
std::unique_ptr< TensorType >	gqa_att_ { nullptr }
std::unique_ptr< TensorType >	gqa_att_decode_ { nullptr }
std::unique_ptr< TensorType >	gqa_preatt_ { nullptr }
std::unique_ptr< TensorType >	gqa_preatt_decode_ { nullptr }
std::unique_ptr< TensorType >	gqa_q_permute_ { nullptr }
std::unique_ptr< TensorType >	gqa_v_out_ { nullptr }
std::unique_ptr< TensorType >	gqa_v_out_decode_ { nullptr }
shape_t	input_shape_
std::shared_ptr< LmHeadLinearType >	lm_head_ { nullptr }
TensorType *	logits_ptr_ { nullptr }
TensorType *	normalized_ptr_ { nullptr }
shape_t	output_shape_
std::unique_ptr< TensorType >	prefill_ { nullptr }
int64_t	prefill_chunk_size_ { 0 }
int64_t	seq_length_ { 0 }
TensorType *	token_embed_out_ptr_ { nullptr }
std::shared_ptr< TokenEmbeddingType >	token_embedding_ { nullptr }
std::vector< std::shared_ptr< TransformerBlockType > >	transformer_blocks_

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 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, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>
requires PrecisionSupportedOnDevice<TPrecision, TDeviceType>
class Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >

LLaMA-style transformer (decoder-only) for autoregressive token prediction.

Graph: TokenEmbedding → RoPE → LlamaBlock × N → RmsNorm → Linear (lm_head). RoPE is applied to the full embedding stream after the token lookup; each LlamaBlock receives rotary-encoded embeddings as input.

Template parameters:

TDeviceType: device type (Cpu/Cuda)
TPrecision: tensor precision

Constructor & Destructor Documentation

◆ LlamaTransformer()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::LlamaTransformer	(	const std::string &	name,
		const LlamaConfig &	config,
		DeviceId	device_id )

inlineexplicitexport

Here is the call graph for this function:

◆ ~LlamaTransformer()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::~LlamaTransformer ( )

overrideexportdefault

Member Function Documentation

◆ backward()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

TokenIndexType & Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::backward	(	const TokenIndexType &	input,
		const TensorType &	output_grad )

inlineoverrideexport

Here is the call graph for this function:

Here is the caller graph for this function:

◆ createConfigFromMetadata()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

LlamaConfig Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::createConfigFromMetadata ( const PretrainedMetadata & metadata )

inlinestaticexportprivate

Here is the call graph for this function:

◆ createGraph()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

void Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::createGraph ( )

inlineexportprivate

Here is the call graph for this function:

Here is the caller graph for this function:

◆ decode()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

TensorType & Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::decode	(	const TokenIndexType &	input,
		int	position )

inlineoverrideexport

◆ forward()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

TensorType & Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::forward ( const TokenIndexType & input )

inlineoverrideexport

Here is the call graph for this function:

◆ getExecutionContext()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

IExecutionContext * Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::getExecutionContext ( ) const

inlineexport

Here is the call graph for this function:

Here is the caller graph for this function:

◆ getMemoryStats()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

MemoryStats Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::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, TPrecision >.

Here is the call graph for this function:

◆ getType()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

const ComponentType Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::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, TPrecision >.

◆ loadParameters()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

void Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::loadParameters ( PretrainedModelReader & reader )

inlineexport

Here is the call graph for this function:

◆ onBuilding()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

void Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::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, TPrecision >.

Here is the call graph for this function:

◆ onTrainingModeChanging()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

void Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::onTrainingModeChanging ( TrainingMode training_mode )

inlineoverrideexportprotectedvirtual

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 setTraining().

Parameters

is_training New training mode (true = training, false = eval)

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

Here is the call graph for this function:

◆ parseParameterPath()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

std::pair< std::string, std::string > Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::parseParameterPath ( const std::string & full_name ) const

inlineexportprivate

Here is the caller graph for this function:

◆ prefill()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

TensorType & Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::prefill ( const TokenIndexType & input )

inlineoverrideexport

Here is the call graph for this function:

◆ save_()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

void Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::save_	(	ModelArchive &	archive,
		SerializationMode	mode ) const

inlineoverrideexportprotectedvirtual

Hook for concrete classes to save type-specific state.

REQUIRED override for concrete networks. Must write:

Type identifier (e.g., "type": "MnistClassifier")
Configuration parameters (batch_size, architecture constants)
Shape metadata (for validation during Load())

This metadata enables the concrete class's Load() method to reconstruct the network.

Example implementation:

void save_(ModelArchive& archive, SerializationMode mode) const override
{
    json meta;
    meta["type"] = "MnistClassifier";  // Type identifier for runtime dispatch
    meta["batch_size"] = batch_size_;
    meta["input_shape"] = leading_shape_;
    // ... other configuration
    archive.writeJson("network/classifier_meta.json", meta);
}

Parameters

archive	Archive to write to
mode	Serialization mode (passed from save())

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

Here is the call graph for this function:

◆ toString()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

std::string Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::toString ( ) const

inlineoverrideexportvirtual

Generate a human-readable description.

Returns: String representation showing children

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

Here is the call graph for this function:

◆ validateBuildContext()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

void Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::validateBuildContext ( const BuildContext & context ) const

inlineexportprivate

Here is the call graph for this function:

Here is the caller graph for this function:

◆ validateLeadingShape()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

void Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::validateLeadingShape ( const shape_t & leading_shape ) const

inlineexportprivate

Here is the call graph for this function:

◆ zeroGradients()

template<DeviceType TDeviceType, TensorDataType TPrecision, WeightQuantPolicy TWeightQuantization = NoWeightQuant, KvCachePolicy TKvCachePolicy = NoKvCompression>

void Mila::Dnn::LlamaTransformer< TDeviceType, TPrecision, TWeightQuantization, TKvCachePolicy >::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, TPrecision >.

Here is the call graph for this function:

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

/__w/Mila/Mila/Mila/Src/Dnn/Components/Transformers/LlaMa/Llama.ixx

Public Types

Public Member Functions

Protected Member Functions

Private Member Functions

Static Private Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ LlamaTransformer()

◆ ~LlamaTransformer()

Member Function Documentation

◆ backward()

◆ createConfigFromMetadata()

◆ createGraph()

◆ decode()

◆ forward()

◆ getExecutionContext()

◆ getMemoryStats()

◆ getType()

◆ loadParameters()

◆ onBuilding()

◆ onTrainingModeChanging()

◆ parseParameterPath()

◆ prefill()

◆ save_()

◆ toString()

◆ validateBuildContext()

◆ validateLeadingShape()

◆ zeroGradients()