Name | ndonnx JSON |
Version |
0.4.0
JSON |
| download |
home_page | None |
Summary | ONNX backed array library compliant with Array API standard. |
upload_time | 2024-06-15 10:47:52 |
maintainer | None |
docs_url | None |
author | None |
requires_python | >=3.10 |
license | None |
keywords |
numpy
onnx
array-api
|
VCS |
![](/static/img/github-24-000000.png) |
bugtrack_url |
|
requirements |
No requirements were recorded.
|
Travis-CI |
No Travis.
|
coveralls test coverage |
No coveralls.
|
# ndonnx
[![CI](https://img.shields.io/github/actions/workflow/status/quantco/ndonnx/ci.yml?style=flat-square&branch=main)](https://github.com/quantco/ndonnx/actions/workflows/ci.yml)
[![Documentation](https://readthedocs.org/projects/ndonnx/badge/?version=latest)](https://ndonnx.readthedocs.io/en/latest/?badge=latest)
[![conda-forge](https://img.shields.io/conda/pn/conda-forge/ndonnx?style=flat-square&logoColor=white&logo=conda-forge)](https://prefix.dev/channels/conda-forge/packages/ndonnx)
An ONNX-backed array library that is compliant with the [Array API](https://data-apis.org/array-api/) standard.
## Installation
Releases are available on PyPI and conda-forge.
```bash
# using pip
pip install ndonnx
# using conda
conda install ndonnx
# using pixi
pixi add ndonnx
```
## Development
You can install the package in development mode using:
```bash
git clone https://github.com/quantco/ndonnx
cd ndonnx
# For Array API tests
git submodule update --init --recursive
pixi shell
pre-commit run -a
pip install --no-build-isolation --no-deps -e .
pytest tests -n auto
```
## Quick start
`ndonnx` is an ONNX based python array library.
It has a couple of key features:
- It implements the [`Array API`](https://data-apis.org/array-api/) standard. Standard compliant code can be executed without changes across numerous backends such as like `NumPy`, `JAX` and now `ndonnx`.
```python
import numpy.array_api as npx
import ndonnx as ndx
from jax.experimental import array_api as jxp
def mean_drop_outliers(a, low=-5, high=5):
xp = a.__array_namespace__()
return xp.mean(a[(low < a) & (a < high)])
arr = [-12.12, 1.12, 2.12, 2.13, 123.,]
np_result = mean_drop_outliers(npx.asarray(arr))
jax_result = mean_drop_outliers(jxp.asarray(arr))
ndx_result = mean_drop_outliers(ndx.asarray(arr))
print(np_result) # 1.79
print(jax_result) # 1.79
print(ndx_result) # Array(1.79, dtype=ndx.Float64)
assert np_result == ndx_result.to_numpy()
```
- It supports ONNX export. This allows you persist your logic into an ONNX computation graph for convenient and performant inference.
```python
import onnx
import ndonnx as ndx
a = ndx.array(shape=("N",), dtype=ndx.float64)
b = ndx.array(shape=("N",), dtype=ndx.float64)
out = a[:2] + b[:2]
model_proto = ndx.build({"a": a, "b": b}, {"c": out})
onnx.save(model_proto, "model.onnx")
# Having serialised your model to disk, perform
# inference using a runtime of your choosing.
import onnxruntime as ort
import numpy as np
inference_session = ort.InferenceSession("model.onnx")
prediction, = inference_session.run(None, {
"a": np.array([1, 2, 3], dtype=np.float64),
"b": np.array([4, 5, 6], dtype=np.float64),
})
print(prediction) # array([5., 7.])
```
In the future we will be enabling a stable API for an extensible data type system. This will allow users to define their own data types and operations on arrays with these data types.
## Array API coverage
Array API compatibility is tracked in the array-api coverage test suite in `api-coverage-tests`. Missing coverage is tracked in the `skips.txt` file. Contributions are welcome!
Summary(1119 total):
- 898 passed
- 210 failed
- 11 deselected
Run the tests with:
```bash
pixi run arrayapitests
```
Raw data
{
"_id": null,
"home_page": null,
"name": "ndonnx",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.10",
"maintainer_email": null,
"keywords": "numpy, onnx, array-api",
"author": null,
"author_email": "Aditya Goel <agoel4512@gmail.com>, Christian Bourjau <christian.bourjau@quantco.com>",
"download_url": "https://files.pythonhosted.org/packages/ae/c1/a17930859988d84f00ecff210ae703170411976b828766584caf00605755/ndonnx-0.4.0.tar.gz",
"platform": null,
"description": "# ndonnx\n\n[![CI](https://img.shields.io/github/actions/workflow/status/quantco/ndonnx/ci.yml?style=flat-square&branch=main)](https://github.com/quantco/ndonnx/actions/workflows/ci.yml)\n[![Documentation](https://readthedocs.org/projects/ndonnx/badge/?version=latest)](https://ndonnx.readthedocs.io/en/latest/?badge=latest)\n[![conda-forge](https://img.shields.io/conda/pn/conda-forge/ndonnx?style=flat-square&logoColor=white&logo=conda-forge)](https://prefix.dev/channels/conda-forge/packages/ndonnx)\n\nAn ONNX-backed array library that is compliant with the [Array API](https://data-apis.org/array-api/) standard.\n\n## Installation\n\nReleases are available on PyPI and conda-forge.\n\n```bash\n# using pip\npip install ndonnx\n# using conda\nconda install ndonnx\n# using pixi\npixi add ndonnx\n```\n\n## Development\n\nYou can install the package in development mode using:\n\n```bash\ngit clone https://github.com/quantco/ndonnx\ncd ndonnx\n\n# For Array API tests\ngit submodule update --init --recursive\n\npixi shell\npre-commit run -a\npip install --no-build-isolation --no-deps -e .\npytest tests -n auto\n```\n\n## Quick start\n\n`ndonnx` is an ONNX based python array library.\n\nIt has a couple of key features:\n\n- It implements the [`Array API`](https://data-apis.org/array-api/) standard. Standard compliant code can be executed without changes across numerous backends such as like `NumPy`, `JAX` and now `ndonnx`.\n\n ```python\n import numpy.array_api as npx\n import ndonnx as ndx\n from jax.experimental import array_api as jxp\n\n def mean_drop_outliers(a, low=-5, high=5):\n xp = a.__array_namespace__()\n return xp.mean(a[(low < a) & (a < high)])\n\n arr = [-12.12, 1.12, 2.12, 2.13, 123.,]\n\n np_result = mean_drop_outliers(npx.asarray(arr))\n jax_result = mean_drop_outliers(jxp.asarray(arr))\n ndx_result = mean_drop_outliers(ndx.asarray(arr))\n print(np_result) # 1.79\n print(jax_result) # 1.79\n print(ndx_result) # Array(1.79, dtype=ndx.Float64)\n assert np_result == ndx_result.to_numpy()\n ```\n\n- It supports ONNX export. This allows you persist your logic into an ONNX computation graph for convenient and performant inference.\n\n ```python\n import onnx\n import ndonnx as ndx\n\n a = ndx.array(shape=(\"N\",), dtype=ndx.float64)\n b = ndx.array(shape=(\"N\",), dtype=ndx.float64)\n out = a[:2] + b[:2]\n model_proto = ndx.build({\"a\": a, \"b\": b}, {\"c\": out})\n onnx.save(model_proto, \"model.onnx\")\n\n # Having serialised your model to disk, perform\n # inference using a runtime of your choosing.\n import onnxruntime as ort\n import numpy as np\n inference_session = ort.InferenceSession(\"model.onnx\")\n prediction, = inference_session.run(None, {\n \"a\": np.array([1, 2, 3], dtype=np.float64),\n \"b\": np.array([4, 5, 6], dtype=np.float64),\n })\n print(prediction) # array([5., 7.])\n ```\n\nIn the future we will be enabling a stable API for an extensible data type system. This will allow users to define their own data types and operations on arrays with these data types.\n\n## Array API coverage\n\nArray API compatibility is tracked in the array-api coverage test suite in `api-coverage-tests`. Missing coverage is tracked in the `skips.txt` file. Contributions are welcome!\n\nSummary(1119 total):\n\n- 898 passed\n- 210 failed\n- 11 deselected\n\nRun the tests with:\n\n```bash\npixi run arrayapitests\n```\n",
"bugtrack_url": null,
"license": null,
"summary": "ONNX backed array library compliant with Array API standard.",
"version": "0.4.0",
"project_urls": {
"repository": "https://github.com/quantco/ndonnx"
},
"split_keywords": [
"numpy",
" onnx",
" array-api"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "74d357db06038f0195065b2dbc014ee06342cb6814f0962797e5527e8c6c67f0",
"md5": "08ec8547ee74e3f4890bb751164b2a06",
"sha256": "03e45e2e2e6e4cad28b2130c7d6d9f8e1cac46c6c216554e8c2537e7bb96d4b9"
},
"downloads": -1,
"filename": "ndonnx-0.4.0-py3-none-any.whl",
"has_sig": false,
"md5_digest": "08ec8547ee74e3f4890bb751164b2a06",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.10",
"size": 45584,
"upload_time": "2024-06-15T10:47:50",
"upload_time_iso_8601": "2024-06-15T10:47:50.324975Z",
"url": "https://files.pythonhosted.org/packages/74/d3/57db06038f0195065b2dbc014ee06342cb6814f0962797e5527e8c6c67f0/ndonnx-0.4.0-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "aec1a17930859988d84f00ecff210ae703170411976b828766584caf00605755",
"md5": "d8a553ff333174b0ba80aea8be976c78",
"sha256": "5bbc7298d4afd209d324bf313704be42a06e90d6f42880cfe321cdf520870cce"
},
"downloads": -1,
"filename": "ndonnx-0.4.0.tar.gz",
"has_sig": false,
"md5_digest": "d8a553ff333174b0ba80aea8be976c78",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.10",
"size": 293828,
"upload_time": "2024-06-15T10:47:52",
"upload_time_iso_8601": "2024-06-15T10:47:52.944208Z",
"url": "https://files.pythonhosted.org/packages/ae/c1/a17930859988d84f00ecff210ae703170411976b828766584caf00605755/ndonnx-0.4.0.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-06-15 10:47:52",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "quantco",
"github_project": "ndonnx",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"lcname": "ndonnx"
}