module onnxrt.validate.validate_benchmark_replay

Short summary

module mlprodict.onnxrt.validate.validate_benchmark_replay

Measures time processing for ONNX models.

source on GitHub

Classes

class	truncated documentation
SimplifiedOnnxInference	Simple wrapper around InferenceSession which imitates OnnxInference. It only enable CPUExecutionProvider. …

Functions

function	truncated documentation
enumerate_benchmark_replay	Replays a benchmark stored with function enumerate_validated_operator_opsets …

Properties

property	truncated documentation
input_names	Returns InferenceSession input names.

Methods

method	truncated documentation
__init__
run	Calls InferenceSession.run.

Documentation

Measures time processing for ONNX models.

source on GitHub

class mlprodict.onnxrt.validate.validate_benchmark_replay.SimplifiedOnnxInference(ort, runtime='onnxruntime')

Bases: object

Simple wrapper around InferenceSession which imitates OnnxInference. It only enable CPUExecutionProvider.

Parameters:

runtime – see InferenceSession

source on GitHub

__init__(ort, runtime='onnxruntime')

property input_names

Returns InferenceSession input names.

run(input)

Calls InferenceSession.run.

mlprodict.onnxrt.validate.validate_benchmark_replay.enumerate_benchmark_replay(folder, runtime='python', time_kwargs=None, skip_long_test=True, time_kwargs_fact=None, time_limit=4, verbose=1, fLOG=None)

Replays a benchmark stored with function enumerate_validated_operator_opsets or command line validate_runtime. Enumerates the results.

Parameters:

• folder – folder where to find pickled files, all files must have pkl or pickle extension

• runtime – runtime or runtimes

• time_kwargs – to define a more precise way to measure a model

• skip_long_test – skips tests for high values of N if they seem too long

• time_kwargs_fact – see _multiply_time_kwargs

• time_limit – to skip the rest of the test after this limit (in second)

• verbose – if >= 1, uses tqdm

• fLOG – logging function

Returns:

iterator on results

source on GitHub