mltk.core.tflite_micro.tflite_micro_accelerator.TfliteMicroAccelerator

class TfliteMicroAccelerator

TF-Lite Micro Accelerator

This class allows for providing hardware-accelerated kernels to the TFLM interpreter.

Properties

accelerator_wrapper	Return the TfliteMicroAcceleratorWrapper instance
active_variant	The name of the variant actively being used by this accelerator
api_version	The API version number this wrapper was built with This number must match the tflite_micro_wrapper's API version
git_hash	Return the GIT hash of the MLTK repo used to compile the wrapper library
name	The name of the accelerator
supports_model_compilation	Return if this accelerator supports model compilation
variants	List of variants supported by this accelerator

Methods

__init__
compile_model	Compile the given .tflite model and return a new TfliteModel instance with the compiled data
enable_data_recorder	Enable the accelerator data recorder
estimate_profiling_results	Update the given ProfilingModelResults with estimated model metrics
set_program_recorder_enabled	Enable the accelerator instruction recorder

__init__(accelerator_wrapper)

property name: str

The name of the accelerator

Return type:

str

property variants: List[str]

List of variants supported by this accelerator

Return type:

List[str]

property active_variant: str

The name of the variant actively being used by this accelerator

Return type:

str

property api_version: int

The API version number this wrapper was built with This number must match the tflite_micro_wrapper’s API version

Return type:

int

property git_hash: str

Return the GIT hash of the MLTK repo used to compile the wrapper library

Return type:

str

property accelerator_wrapper: object

Return the TfliteMicroAcceleratorWrapper instance

Return type:

object

property supports_model_compilation: bool

Return if this accelerator supports model compilation

Return type:

bool

estimate_profiling_results(results, **kwargs)

Update the given ProfilingModelResults with estimated model metrics

set_program_recorder_enabled(enabled)

Enable the accelerator instruction recorder

Parameters:

enabled (bool) –

enable_data_recorder()

Enable the accelerator data recorder

compile_model(model, logger=None, report_path=None, **kwargs)

Compile the given .tflite model and return a new TfliteModel instance with the compiled data

NOTE: The accelerator must support model compilation to use this API

Return type:

TfliteModel

Parameters:

• model (TfliteModel) –

• logger (Logger) –

• report_path (str) –