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

Commit70d6f64

Browse files
authored
First documentation of the light API (#44)
* documentation* remove rstcheck, too many bugs* add missing link* fix unit test* fix documentation* unstable* windows* update titles* minorchanges
1 parentdaf4402 commit70d6f64

File tree

22 files changed

+1012
-61
lines changed

22 files changed

+1012
-61
lines changed

‎.github/workflows/rstcheck.yml‎

Lines changed: 0 additions & 27 deletions
This file was deleted.

‎CHANGELOGS.rst‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,7 +1,7 @@
11
Change Logs
22
===========
33

4-
0.2.0
4+
0.1.2
55
+++++
66

77
*:pr:`42`: first sketch for a very simple API to create onnx graph in one or two lines

‎README.rst‎

Lines changed: 29 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -2,7 +2,7 @@
22
..image::https://github.com/sdpython/onnx-array-api/raw/main/_doc/_static/logo.png
33
:width:120
44

5-
onnx-array-api:(Numpy) Array API forONNX
5+
onnx-array-api:APIs to createONNX Graphs
66
==========================================
77

88
..image::https://dev.azure.com/xavierdupre3/onnx-array-api/_apis/build/status/sdpython.onnx-array-api
@@ -29,7 +29,9 @@ onnx-array-api: (Numpy) Array API for ONNX
2929
..image::https://codecov.io/gh/sdpython/onnx-array-api/branch/main/graph/badge.svg?token=Wb9ZGDta8J
3030
:target:https://codecov.io/gh/sdpython/onnx-array-api
3131

32-
**onnx-array-api** implements a numpy API for ONNX.
32+
**onnx-array-api** implements APIs to create custom ONNX graphs.
33+
The objective is to speed up the implementation of converter libraries.
34+
The first one matches **numpy API**.
3335
It gives the user the ability to convert functions written
3436
following the numpy API to convert that function into ONNX as
3537
well as to execute it.
@@ -111,6 +113,31 @@ It supports eager mode as well:
111113
l2_loss=[0.002]
112114
[0.042]
113115

116+
The second API ir **Light API** tends to do every thing in one line.
117+
The euclidean distance looks like the following:
118+
119+
::
120+
121+
import numpy as np
122+
from onnx_array_api.light_api import start
123+
from onnx_array_api.plotting.text_plot import onnx_simple_text_plot
124+
125+
model = (
126+
start()
127+
.vin("X")
128+
.vin("Y")
129+
.bring("X", "Y")
130+
.Sub()
131+
.rename("dxy")
132+
.cst(np.array([2], dtype=np.int64), "two")
133+
.bring("dxy", "two")
134+
.Pow()
135+
.ReduceSum()
136+
.rename("Z")
137+
.vout()
138+
.to_onnx()
139+
)
140+
114141
The library is released on
115142
`pypi/onnx-array-api<https://pypi.org/project/onnx-array-api/>`_
116143
and its documentation is published at

‎_doc/api/docs.rst‎

Lines changed: 7 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,7 @@
1+
validation.docs
2+
===============
3+
4+
make_euclidean
5+
++++++++++++++
6+
7+
..autofunction::onnx_array_api.validation.docs.make_euclidean

‎_doc/api/index.rst‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -22,3 +22,4 @@ API
2222
tools
2323
profiling
2424
f8
25+
docs

‎_doc/api/light_api.rst‎

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -24,9 +24,11 @@ Var
2424

2525
..autoclass::onnx_array_api.light_api.Var
2626
:members:
27+
:inherited-members:
2728

2829
Vars
2930
====
3031

3132
..autoclass::onnx_array_api.light_api.Vars
3233
:members:
34+
:inherited-members:

‎_doc/conf.py‎

Lines changed: 13 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -114,21 +114,34 @@
114114
"https://data-apis.org/array-api/",
115115
("2022.12/API_specification/generated/array_api.{0}.html",1),
116116
),
117+
"ast":"https://docs.python.org/3/library/ast.html",
117118
"cProfile.Profile":"https://docs.python.org/3/library/profile.html#profile.Profile",
118119
"DOT":"https://graphviz.org/doc/info/lang.html",
120+
"inner API":"https://onnx.ai/onnx/intro/python.html",
119121
"JIT":"https://en.wikipedia.org/wiki/Just-in-time_compilation",
120122
"onnx":"https://onnx.ai/onnx/",
123+
"onnx.helper":"https://onnx.ai/onnx/api/helper.html",
121124
"ONNX":"https://onnx.ai/",
125+
"ONNX Operators":"https://onnx.ai/onnx/operators/",
122126
"onnxruntime":"https://onnxruntime.ai/",
127+
"onnxruntime-training":"https://onnxruntime.ai/docs/get-started/training-on-device.html",
123128
"numpy":"https://numpy.org/",
124129
"numba":"https://numba.pydata.org/",
125130
"onnx-array-api": ("https://sdpython.github.io/doc/onnx-array-api/dev/"),
131+
"onnxscript":"https://github.com/microsoft/onnxscript",
126132
"pyinstrument":"https://github.com/joerick/pyinstrument",
127133
"python":"https://www.python.org/",
134+
"pytorch":"https://pytorch.org/",
135+
"reverse Polish notation":"https://en.wikipedia.org/wiki/Reverse_Polish_notation",
128136
"scikit-learn":"https://scikit-learn.org/stable/",
129137
"scipy":"https://scipy.org/",
138+
"sklearn-onnx":"https://onnx.ai/sklearn-onnx/",
139+
"spox":"https://github.com/Quantco/spox",
130140
"sphinx-gallery":"https://github.com/sphinx-gallery/sphinx-gallery",
141+
"tensorflow":"https://www.tensorflow.org/",
142+
"tensorflow-onnx":"https://github.com/onnx/tensorflow-onnx",
131143
"torch":"https://pytorch.org/docs/stable/torch.html",
144+
"torch.onnx":"https://pytorch.org/docs/stable/onnx.html",
132145
#
133146
"C_OrtValue": (
134147
"http://www.xavierdupre.fr/app/onnxcustom/helpsphinx/"

‎_doc/index.rst‎

Lines changed: 35 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,5 @@
11

2-
onnx-array-api:(Numpy) Array API forONNX
2+
onnx-array-api:APIs to createONNX Graphs
33
==========================================
44

55
..image::https://dev.azure.com/xavierdupre3/onnx-array-api/_apis/build/status/sdpython.onnx-array-api
@@ -26,10 +26,8 @@ onnx-array-api: (Numpy) Array API for ONNX
2626
..image::https://codecov.io/gh/sdpython/onnx-array-api/branch/main/graph/badge.svg?token=Wb9ZGDta8J
2727
:target:https://codecov.io/gh/sdpython/onnx-array-api
2828

29-
**onnx-array-api** implements a numpy API for ONNX.
30-
It gives the user the ability to convert functions written
31-
following the numpy API to convert that function into ONNX as
32-
well as to execute it.
29+
**onnx-array-api** implements APIs to create custom ONNX graphs.
30+
The objective is to speed up the implementation of converter libraries.
3331

3432
..toctree::
3533
:maxdepth:1
@@ -47,6 +45,8 @@ well as to execute it.
4745
CHANGELOGS
4846
license
4947

48+
**Numpy API**
49+
5050
Sources available on
5151
`github/onnx-array-api<https://github.com/sdpython/onnx-array-api>`_.
5252

@@ -57,7 +57,7 @@ Sources available on
5757

5858
import numpy as np # A
5959
from onnx_array_api.npx import absolute, jit_onnx
60-
from onnx_array_api.plotting.dot_plot importto_dot
60+
from onnx_array_api.plotting.text_plot importonnx_simple_text_plot
6161

6262
def l1_loss(x, y):
6363
return absolute(x - y).sum()
@@ -78,6 +78,8 @@ Sources available on
7878
res = jitted_myloss(x, y)
7979
print(res)
8080

81+
print(onnx_simple_text_plot(jitted_myloss.get_onnx()))
82+
8183
..gdot::
8284
:script: DOT-SECTION
8385
:process:
@@ -106,3 +108,30 @@ Sources available on
106108
y = np.array([[0.11, 0.22], [0.33, 0.44]], dtype=np.float32)
107109
res = jitted_myloss(x, y)
108110
print(to_dot(jitted_myloss.get_onnx()))
111+
112+
**Light API**
113+
114+
..runpython::
115+
:showcode:
116+
117+
import numpy as np
118+
from onnx_array_api.light_api import start
119+
from onnx_array_api.plotting.text_plot import onnx_simple_text_plot
120+
121+
model = (
122+
start()
123+
.vin("X")
124+
.vin("Y")
125+
.bring("X", "Y")
126+
.Sub()
127+
.rename("dxy")
128+
.cst(np.array([2], dtype=np.int64), "two")
129+
.bring("dxy", "two")
130+
.Pow()
131+
.ReduceSum()
132+
.rename("Z")
133+
.vout()
134+
.to_onnx()
135+
)
136+
137+
print(onnx_simple_text_plot(model))

‎_doc/tech/aapi.rst‎

Lines changed: 3 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,7 @@
1+
.. _l-array-api-painpoint:
12

2-
Difficulty to implement ananArray API for ONNX
3-
================================================
3+
Difficulty to implement an Array API for ONNX
4+
=============================================
45

56
Implementing the full array API is not always easy with:epkg:`onnx`.
67
Python is not strongly typed and many different types can be used

‎_doc/tutorial/index.rst‎

Lines changed: 3 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -6,5 +6,7 @@ Tutorial
66
..toctree::
77
:maxdepth:1
88

9-
overview
9+
onnx_api
10+
light_api
11+
numpy_api
1012
benchmarks

0 commit comments

Comments
 (0)
[8]ページ先頭
©2009-2025 Movatter.jp