ActsPlugins::TorchMetricLearning Class Reference

Graph construction using PyTorch metric learning. More...

#include <ActsPlugins/Gnn/TorchMetricLearning.hpp>

Inheritance diagram for ActsPlugins::TorchMetricLearning:

Collaboration diagram for ActsPlugins::TorchMetricLearning:

Classes
struct	Config
	Configuration struct for Torch metric learning. More...

Public Member Functions
	TorchMetricLearning (const Config &cfg, std::unique_ptr< const Acts::Logger > logger)
	Constructor.
Config	config () const
	Get the configuration.
PipelineTensors	operator() (std::vector< float > &inputValues, std::size_t numNodes, const std::vector< std::uint64_t > &moduleIds, const ExecutionContext &execContext={}) override
	Perform the graph construction.

Detailed Description

Graph construction using PyTorch metric learning.

Constructor & Destructor Documentation

TorchMetricLearning()

ActsPlugins::TorchMetricLearning::TorchMetricLearning	(	const Config &	cfg,
		std::unique_ptr< const Acts::Logger >	logger )

Constructor.

Parameters

cfg	Configuration parameters
logger	Logging instance

Member Function Documentation

config()

Config ActsPlugins::TorchMetricLearning::config

(

)

const

Get the configuration.

Returns

Copy of the configuration struct

operator()()

PipelineTensors ActsPlugins::TorchMetricLearning::operator() ( std::vector< float > & inputValues, std::size_t numNodes, const std::vector< std::uint64_t > & moduleIds, const ExecutionContext & execContext = {} )

overridevirtual

Perform the graph construction.

Parameters

inputValues	Flattened input data
numNodes	Number of nodes. inputValues.size() / numNodes then gives the number of features
moduleIds	Module IDs of the features (used for module-map-like graph construction)
execContext	Device & stream information

Returns

(node_features, edge_features, edge_index)

Implements ActsPlugins::GraphConstructionBase.