Shared base configuration for neural operators. More...

#include <neural_infer.h>

Collaboration diagram for SimNeuralBaseConfig:

[legend]

Public Attributes
size_t	input_field

size_t	output_field

char	model_id [SIM_NEURAL_MODEL_ID_MAX+1U]

bool	accumulate

bool	scale_by_dt

bool	normalize_input

double	input_scale

double	input_bias

double	output_scale

double	output_bias

SimNeuralDeterminismPolicy	determinism_policy

SimNeuralDeviceRequirement	device_requirement

SimNeuralPrecisionMode	precision_mode

SimNeuralShapeConstraints	shape_constraints

SimNeuralInferenceFn	inference_fn

void *	inference_userdata

Detailed Description

Shared base configuration for neural operators.

Member Data Documentation

◆ accumulate

bool SimNeuralBaseConfig::accumulate

Add predictions into output when true.

◆ determinism_policy

SimNeuralDeterminismPolicy SimNeuralBaseConfig::determinism_policy

Determinism requested from backend.

◆ device_requirement

SimNeuralDeviceRequirement SimNeuralBaseConfig::device_requirement

Device requested from backend.

◆ inference_fn

SimNeuralInferenceFn SimNeuralBaseConfig::inference_fn

Optional backend inference callback.

◆ inference_userdata

void* SimNeuralBaseConfig::inference_userdata

Opaque pointer passed to inference_fn.

◆ input_bias

double SimNeuralBaseConfig::input_bias

Input normalization bias.

◆ input_field

size_t SimNeuralBaseConfig::input_field

Source field index.

◆ input_scale

double SimNeuralBaseConfig::input_scale

Input normalization scale.

◆ model_id

char SimNeuralBaseConfig::model_id[SIM_NEURAL_MODEL_ID_MAX+1U]

Backend model identifier.

◆ normalize_input

bool SimNeuralBaseConfig::normalize_input

Apply input_scale and input_bias before inference.

◆ output_bias

double SimNeuralBaseConfig::output_bias

Output denormalization bias.

◆ output_field

size_t SimNeuralBaseConfig::output_field

Destination field index.

◆ output_scale

double SimNeuralBaseConfig::output_scale

Output denormalization scale.

◆ precision_mode

SimNeuralPrecisionMode SimNeuralBaseConfig::precision_mode

Precision requested from backend.

◆ scale_by_dt

bool SimNeuralBaseConfig::scale_by_dt

Scale writes by substep dt when true.

◆ shape_constraints

SimNeuralShapeConstraints SimNeuralBaseConfig::shape_constraints

Accepted input shape constraints.

The documentation for this struct was generated from the following file:

neural_infer.h

Public Attributes