Collaboration diagram for Compilation Parameters:

Typedefs
typedef std::tuple< int, int, int >	celantur::MinOptMaxDims
	A tuple representing minimum, optimal, and maximum dimensions. This tuple is used to specify the range of dimensions for model optimization. The three integers represent the minimum, optimal, and maximum dimensions respectively. More...

Enumerations
enum class	celantur::OptimisationLevel { celantur::Low , celantur::Medium , celantur::Full }
	Enumeration for different levels of optimization. This enum is used to specify the desired level of optimization during model compilation. More...

enum class	celantur::CompilePrecision { celantur::FP32 , celantur::FP16 , celantur::INT8 }
	Enumeration for different precision levels during model compilation. This enum is used to specify the desired precision for model weights and computations. Most engines support only FP32. Currently, only TensorRT supports FP16. INT8 is supported by TensorRT and OpenVINO, but requires special model provided to you by Celantur. More...

Detailed Description

Typedef Documentation

◆ MinOptMaxDims

typedef std::tuple<int, int, int> celantur::MinOptMaxDims

A tuple representing minimum, optimal, and maximum dimensions. This tuple is used to specify the range of dimensions for model optimization. The three integers represent the minimum, optimal, and maximum dimensions respectively.

Enumeration Type Documentation

◆ CompilePrecision

enum celantur::CompilePrecision

strong

Enumeration for different precision levels during model compilation. This enum is used to specify the desired precision for model weights and computations. Most engines support only FP32. Currently, only TensorRT supports FP16. INT8 is supported by TensorRT and OpenVINO, but requires special model provided to you by Celantur.

Enumerator
FP32
FP16
INT8

◆ OptimisationLevel

enum celantur::OptimisationLevel