Class Acts::TorchMetricLearning

class TorchMetricLearning : public Acts::GraphConstructionBase

Public Functions

TorchMetricLearning(const Config &cfg, std::unique_ptr<const Logger> logger)
~TorchMetricLearning()
inline Config config() const
virtual std::tuple<std::any, std::any> operator()(std::vector<float> &inputValues, std::size_t numNodes, int deviceHint = -1) override

Perform the graph construction.

Parameters

  • inputValues – Flattened input data

  • numNodes – Number of nodes. inputValues.size() / numNodes then gives the number of features

  • deviceHint – Which GPU to pick. Not relevant for CPU-only builds

Returns

(node_tensor, edge_tensore)

struct Config

Public Members

int embeddingDim = 8
int knnVal = 500
std::string modelPath
int numFeatures = 3
float rVal = 1.6
bool shuffleDirections = false