ActsPlugins::TorchMetricLearning::Config Struct Reference

Configuration struct for Torch metric learning. More...

#include <ActsPlugins/Gnn/TorchMetricLearning.hpp>

Collaboration diagram for ActsPlugins::TorchMetricLearning::Config:

Public Attributes
int	deviceID = 0
	CUDA device ID to use for inference.
int	embeddingDim = 8
	Dimensionality of the embedding space.
int	knnVal = 500
	Number of nearest neighbors.
std::string	modelPath
	Path to the PyTorch model file.
float	phiScale = 3.141592654
	Scaling factor for phi coordinate in edge features.
float	rVal = 1.6
	Radius value for graph construction.
std::vector< int >	selectedFeatures = {}
	Selected feature indices for input.
bool	shuffleDirections = false
	Whether to shuffle edge directions.

Detailed Description

Configuration struct for Torch metric learning.