Movatterモバイル変換

[0]ホーム
URL:

Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

 Sign up
Appearance settings

First documentation of the light API#44

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to ourterms of service andprivacy statement. We’ll occasionally send you account related emails.

Already on GitHub?Sign in to your account

Merged
sdpython merged 9 commits intomainfromdoc5
Nov 7, 2023
Merged
Show file tree
Hide file tree
Changes fromall commits
Commits
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
27 changes: 0 additions & 27 deletions.github/workflows/rstcheck.yml
View file
Open in desktop

This file was deleted.

2 changes: 1 addition & 1 deletionCHANGELOGS.rst
View file
Open in desktop
Original file line numberDiff line numberDiff line change
@@ -1,7 +1,7 @@
Change Logs
===========

0.2.0
0.1.2
+++++

* :pr:`42`: first sketch for a very simple API to create onnx graph in one or two lines
Expand Down
31 changes: 29 additions & 2 deletionsREADME.rst
View file
Open in desktop
Original file line numberDiff line numberDiff line change
Expand Up@@ -2,7 +2,7 @@
.. image:: https://github.com/sdpython/onnx-array-api/raw/main/_doc/_static/logo.png
:width: 120

onnx-array-api:(Numpy) Array API forONNX
onnx-array-api:APIs to createONNX Graphs
==========================================

.. image:: https://dev.azure.com/xavierdupre3/onnx-array-api/_apis/build/status/sdpython.onnx-array-api
Expand All@@ -29,7 +29,9 @@ onnx-array-api: (Numpy) Array API for ONNX
.. image:: https://codecov.io/gh/sdpython/onnx-array-api/branch/main/graph/badge.svg?token=Wb9ZGDta8J
:target: https://codecov.io/gh/sdpython/onnx-array-api

**onnx-array-api** implements a numpy API for ONNX.
**onnx-array-api** implements APIs to create custom ONNX graphs.
The objective is to speed up the implementation of converter libraries.
The first one matches **numpy API**.
It gives the user the ability to convert functions written
following the numpy API to convert that function into ONNX as
well as to execute it.
Expand DownExpand Up@@ -111,6 +113,31 @@ It supports eager mode as well:
l2_loss=[0.002]
[0.042]

The second API ir **Light API** tends to do every thing in one line.
The euclidean distance looks like the following:

::

import numpy as np
from onnx_array_api.light_api import start
from onnx_array_api.plotting.text_plot import onnx_simple_text_plot

model = (
start()
.vin("X")
.vin("Y")
.bring("X", "Y")
.Sub()
.rename("dxy")
.cst(np.array([2], dtype=np.int64), "two")
.bring("dxy", "two")
.Pow()
.ReduceSum()
.rename("Z")
.vout()
.to_onnx()
)

The library is released on
`pypi/onnx-array-api <https://pypi.org/project/onnx-array-api/>`_
and its documentation is published at
Expand Down
7 changes: 7 additions & 0 deletions_doc/api/docs.rst
View file
Open in desktop
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,7 @@
validation.docs
===============

make_euclidean
++++++++++++++

.. autofunction:: onnx_array_api.validation.docs.make_euclidean
1 change: 1 addition & 0 deletions_doc/api/index.rst
View file
Open in desktop
Original file line numberDiff line numberDiff line change
Expand Up@@ -22,3 +22,4 @@ API
tools
profiling
f8
docs
2 changes: 2 additions & 0 deletions_doc/api/light_api.rst
View file
Open in desktop
Original file line numberDiff line numberDiff line change
Expand Up@@ -24,9 +24,11 @@ Var

.. autoclass:: onnx_array_api.light_api.Var
:members:
:inherited-members:

Vars
====

.. autoclass:: onnx_array_api.light_api.Vars
:members:
:inherited-members:
13 changes: 13 additions & 0 deletions_doc/conf.py
View file
Open in desktop
Original file line numberDiff line numberDiff line change
Expand Up@@ -114,21 +114,34 @@
"https://data-apis.org/array-api/",
("2022.12/API_specification/generated/array_api.{0}.html", 1),
),
"ast": "https://docs.python.org/3/library/ast.html",
"cProfile.Profile": "https://docs.python.org/3/library/profile.html#profile.Profile",
"DOT": "https://graphviz.org/doc/info/lang.html",
"inner API": "https://onnx.ai/onnx/intro/python.html",
"JIT": "https://en.wikipedia.org/wiki/Just-in-time_compilation",
"onnx": "https://onnx.ai/onnx/",
"onnx.helper": "https://onnx.ai/onnx/api/helper.html",
"ONNX": "https://onnx.ai/",
"ONNX Operators": "https://onnx.ai/onnx/operators/",
"onnxruntime": "https://onnxruntime.ai/",
"onnxruntime-training": "https://onnxruntime.ai/docs/get-started/training-on-device.html",
"numpy": "https://numpy.org/",
"numba": "https://numba.pydata.org/",
"onnx-array-api": ("https://sdpython.github.io/doc/onnx-array-api/dev/"),
"onnxscript": "https://github.com/microsoft/onnxscript",
"pyinstrument": "https://github.com/joerick/pyinstrument",
"python": "https://www.python.org/",
"pytorch": "https://pytorch.org/",
"reverse Polish notation": "https://en.wikipedia.org/wiki/Reverse_Polish_notation",
"scikit-learn": "https://scikit-learn.org/stable/",
"scipy": "https://scipy.org/",
"sklearn-onnx": "https://onnx.ai/sklearn-onnx/",
"spox": "https://github.com/Quantco/spox",
"sphinx-gallery": "https://github.com/sphinx-gallery/sphinx-gallery",
"tensorflow": "https://www.tensorflow.org/",
"tensorflow-onnx": "https://github.com/onnx/tensorflow-onnx",
"torch": "https://pytorch.org/docs/stable/torch.html",
"torch.onnx": "https://pytorch.org/docs/stable/onnx.html",
#
"C_OrtValue": (
"http://www.xavierdupre.fr/app/onnxcustom/helpsphinx/"
Expand Down
41 changes: 35 additions & 6 deletions_doc/index.rst
View file
Open in desktop
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,5 @@

onnx-array-api:(Numpy) Array API forONNX
onnx-array-api:APIs to createONNX Graphs
==========================================

.. image:: https://dev.azure.com/xavierdupre3/onnx-array-api/_apis/build/status/sdpython.onnx-array-api
Expand All@@ -26,10 +26,8 @@ onnx-array-api: (Numpy) Array API for ONNX
.. image:: https://codecov.io/gh/sdpython/onnx-array-api/branch/main/graph/badge.svg?token=Wb9ZGDta8J
:target: https://codecov.io/gh/sdpython/onnx-array-api

**onnx-array-api** implements a numpy API for ONNX.
It gives the user the ability to convert functions written
following the numpy API to convert that function into ONNX as
well as to execute it.
**onnx-array-api** implements APIs to create custom ONNX graphs.
The objective is to speed up the implementation of converter libraries.

.. toctree::
:maxdepth: 1
Expand All@@ -47,6 +45,8 @@ well as to execute it.
CHANGELOGS
license

**Numpy API**

Sources available on
`github/onnx-array-api <https://github.com/sdpython/onnx-array-api>`_.

Expand All@@ -57,7 +57,7 @@ Sources available on

import numpy as np # A
from onnx_array_api.npx import absolute, jit_onnx
from onnx_array_api.plotting.dot_plot importto_dot
from onnx_array_api.plotting.text_plot importonnx_simple_text_plot

def l1_loss(x, y):
return absolute(x - y).sum()
Expand All@@ -78,6 +78,8 @@ Sources available on
res = jitted_myloss(x, y)
print(res)

print(onnx_simple_text_plot(jitted_myloss.get_onnx()))

.. gdot::
:script: DOT-SECTION
:process:
Expand DownExpand Up@@ -106,3 +108,30 @@ Sources available on
y = np.array([[0.11, 0.22], [0.33, 0.44]], dtype=np.float32)
res = jitted_myloss(x, y)
print(to_dot(jitted_myloss.get_onnx()))

**Light API**

.. runpython::
:showcode:

import numpy as np
from onnx_array_api.light_api import start
from onnx_array_api.plotting.text_plot import onnx_simple_text_plot

model = (
start()
.vin("X")
.vin("Y")
.bring("X", "Y")
.Sub()
.rename("dxy")
.cst(np.array([2], dtype=np.int64), "two")
.bring("dxy", "two")
.Pow()
.ReduceSum()
.rename("Z")
.vout()
.to_onnx()
)

print(onnx_simple_text_plot(model))
5 changes: 3 additions & 2 deletions_doc/tech/aapi.rst
View file
Open in desktop
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,7 @@
.. _l-array-api-painpoint:

Difficulty to implement ananArray API for ONNX
================================================
Difficulty to implement an Array API for ONNX
=============================================

Implementing the full array API is not always easy with :epkg:`onnx`.
Python is not strongly typed and many different types can be used
Expand Down
4 changes: 3 additions & 1 deletion_doc/tutorial/index.rst
View file
Open in desktop
Original file line numberDiff line numberDiff line change
Expand Up@@ -6,5 +6,7 @@ Tutorial
.. toctree::
:maxdepth: 1

overview
onnx_api
light_api
numpy_api
benchmarks
78 changes: 78 additions & 0 deletions_doc/tutorial/light_api.rst
View file
Open in desktop
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,78 @@
.. _l-light-api:

==========================================
Light API for ONNX: everything in one line
==========================================

It is inspired from the :epkg:`reverse Polish notation`.
Following example implements the euclidean distance.
This API tries to keep it simple and intuitive to short functions.

.. runpython::
:showcode:

import numpy as np
from onnx_array_api.light_api import start
from onnx_array_api.plotting.text_plot import onnx_simple_text_plot

model = (
start()
.vin("X")
.vin("Y")
.bring("X", "Y")
.Sub()
.rename("dxy")
.cst(np.array([2], dtype=np.int64), "two")
.bring("dxy", "two")
.Pow()
.ReduceSum()
.rename("Z")
.vout()
.to_onnx()
)

print(onnx_simple_text_plot(model))

There are two kinds of methods, the graph methods, playing with the graph structure,
and the methods for operators starting with an upper letter.

Graph methods
=============

Any graph must start with function :func:`start <onnx_array_api.light_api.start>`.
It is usually following by `vin` to add an input.

* bring (:meth:`Var.bring <onnx_array_api.light_api.Var.bring>`,
:meth:`Vars.bring <onnx_array_api.light_api.Vars.bring>`):
assembles multiple results into a set before calling an operator taking mulitple inputs,
* cst (:meth:`Var.cst <onnx_array_api.light_api.Var.cst>`,
:meth:`Vars.cst <onnx_array_api.light_api.Vars.cst>`):
adds a constant tensor to the graph,
* rename (:meth:`Var.rename <onnx_array_api.light_api.Var.rename>`,
:meth:`Vars.rename <onnx_array_api.light_api.Vars.rename>`):
renames or give a name to a variable in order to call it later.
* vin (:meth:`Var.vin <onnx_array_api.light_api.Var.vin>`,
:meth:`Vars.vin <onnx_array_api.light_api.Vars.vin>`):
adds an input to the graph,
* vout (:meth:`Var.vout <onnx_array_api.light_api.Var.vout>`,
:meth:`Vars.vout <onnx_array_api.light_api.Vars.vout>`):
declares an existing result as an output.

These methods are implemented in class :class:`onnx_array_api.light_api.var.BaseVar`

Operator methods
================

They are described in :epkg:`ONNX Operators` and redefined in a stable API
so that the definition should not change depending on this opset.
:class:`onnx_array_api.light_api.Var` defines all operators taking only one input.
:class:`onnx_array_api.light_api.Vars` defines all other operators.

Numpy methods
=============

Numpy users expect methods such as `reshape`, property `shape` or
operator `+` to be available as well and that the case. They are
defined in class :class:`Var <onnx_array_api.light_api.Var>` or
:class:`Vars <onnx_array_api.light_api.Vars>` depending on the number of
inputs they require. Their name starts with a lower letter.
View file
Open in desktop
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,5 @@
.. _l-numpy-api-onnx:

==================
Numpy API for ONNX
==================
Expand All@@ -19,6 +21,8 @@ loss functions for example without knowing too much about ONNX.
The first version (onnx==1.15) does not support control flow yet (test and loops).
There is no easy syntax for that yet and the main challenge is to deal with local context.

You read :ref:`l-array-api-painpoint` as well.

Overview
========

Expand Down
Loading
[8]ページ先頭
©2009-2025 Movatter.jp