module testing.onnx_backend

Short summary

module mlprodict.testing.onnx_backend

Tests with onnx backend.

source on GitHub

Classes

class	truncated documentation
OnnxBackendTest	Definition of a backend test. It starts with a folder, in this folder, one onnx file must be there, then a subfolder …

Functions

function	truncated documentation
assert_almost_equal_string	Compares two arrays knowing they contain strings. Raises an exception if the test fails.
enumerate_onnx_tests	Collects test from a sub folder of onnx/backend/test. Works as an enumerator to start processing them without …

Properties

property	truncated documentation
name	Returns the test name.

Static Methods

staticmethod	truncated documentation
_load
_read_proto_from_file
_sort

Methods

method	truncated documentation
__init__
__len__	Returns the number of tests.
__repr__	usual
_compare_results	Compares the expected output and the output produced by the runtime. Raises an exception if not equal.
is_random	Tells if a test is random or not.
run	Executes a tests or all tests if index is None. The function crashes if the tests fails.
to_python	Returns a python code equivalent to the ONNX test.

Documentation

Tests with onnx backend.

source on GitHub

class mlprodict.testing.onnx_backend.OnnxBackendTest(folder)

Bases: object

Definition of a backend test. It starts with a folder, in this folder, one onnx file must be there, then a subfolder for each test to run with this model.

Parameters:

• folder – test folder

• onnx_path – onnx file

• onnx_model – loaded onnx file

• tests – list of test

source on GitHub

__init__(folder)

__len__()

Returns the number of tests.

__repr__()

usual

_compare_results(index, i, e, o, decimal=None)

Compares the expected output and the output produced by the runtime. Raises an exception if not equal.

Parameters:

• index – test index

• i – output index

• e – expected output

• o – output

• decimal – precision

source on GitHub

static _load(folder, names)

static _read_proto_from_file(full)

static _sort(filenames)

is_random()

Tells if a test is random or not.

property name

Returns the test name.

run(load_fct, run_fct, index=None, decimal=None)

Executes a tests or all tests if index is None. The function crashes if the tests fails.

Parameters:

• load_fct – loading function, takes a loaded onnx graph, and returns an object

• run_fct – running function, takes the result of previous function, the inputs, and returns the outputs

• index – index of the test to run or all.

• decimal – requested precision to compare results

source on GitHub

to_python()

Returns a python code equivalent to the ONNX test.

Returns:

code

source on GitHub

mlprodict.testing.onnx_backend.assert_almost_equal_string(expected, value)

Compares two arrays knowing they contain strings. Raises an exception if the test fails.

Parameters:

• expected – expected array

• value – value

source on GitHub

mlprodict.testing.onnx_backend.enumerate_onnx_tests(series, fct_filter=None)

Collects test from a sub folder of onnx/backend/test. Works as an enumerator to start processing them without waiting or storing too much of them.

Parameters:

• series – which subfolder to load, possible values: (‘node’, …)

• fct_filter – function lambda testname: boolean to load or skip the test, None for all

Returns:

list of OnnxBackendTest

source on GitHub