blace::ml_core::InferenceArgsCollection Class Reference

#include <types.h>

Public Member Functions
ml_core::BlaceHash	hash ()
bool	operator== (const InferenceArgsCollection &other) const
bool	operator!= (const InferenceArgsCollection &other) const

Public Attributes
ModelInferenceArgs	inference_args
ml_core::FORWARD_MODE	mode = FORWARD_MODE::SIMPLE
std::optional< MultisampleInferenceArgs >	multi_args = std::nullopt
std::optional< SlicedInferenceArgs >	sliced_args = std::nullopt

Detailed Description

A collection of arguments holding all parameters for model inference.

Member Function Documentation

hash()

ml_core::BlaceHash blace::ml_core::InferenceArgsCollection::hash

(

)

Hash the struct.

Returns

The hash.

operator!=()

bool blace::ml_core::InferenceArgsCollection::operator!= ( const InferenceArgsCollection &  other ) const

inline

Simple inequality operator.

Parameters

other

The other struct.

Returns

operator==()

bool blace::ml_core::InferenceArgsCollection::operator==

(

const InferenceArgsCollection &

other

)

const

Simple equality operator.

Parameters

other

The other struct.

Returns

Member Data Documentation

inference_args

ModelInferenceArgs blace::ml_core::InferenceArgsCollection::inference_args

The basic inference arguments like backend etc.

mode

ml_core::FORWARD_MODE blace::ml_core::InferenceArgsCollection::mode = FORWARD_MODE::SIMPLE

The mode the inference shall be run in. MULTIPASS will run multiple jittered passes on BCHW tensors and SLICED cuts the inputs in parts to save memory.

multi_args

std::optional<MultisampleInferenceArgs> blace::ml_core::InferenceArgsCollection::multi_args = std::nullopt

In case of MULTIPASS hold the arguments here.

sliced_args

std::optional<SlicedInferenceArgs> blace::ml_core::InferenceArgsCollection::sliced_args = std::nullopt

In case of SLICED hold the arguments here.