{ "info": { "author": "Intel", "author_email": "intelnervana@intel.com", "bugtrack_url": null, "classifiers": [], "description": "# ngraph-onnx [![Build Status](https://travis-ci.org/NervanaSystems/ngraph-onnx.svg?branch=master)](https://travis-ci.org/NervanaSystems/ngraph-onnx/branches)\n\nnGraph Backend for ONNX.\n\nThis repository contains tools to run [ONNX][onnx] models using the [Intel nGraph library][ngraph_github] as a backend.\n\n## Installation\n\nFollow our [build][building] instructions to install nGraph-ONNX from sources.\n\n\n\n## Usage example\n\n### Importing an ONNX model\n\nYou can download models from the [ONNX model zoo][onnx_model_zoo]. For example ResNet-50:\n\n```\n$ wget https://s3.amazonaws.com/download.onnx/models/opset_8/resnet50.tar.gz\n$ tar -xzvf resnet50.tar.gz\n```\n\nUse the following Python commands to convert the downloaded model to an nGraph model:\n\n```python\n# Import ONNX and load an ONNX file from disk\n>>> import onnx\n>>> onnx_protobuf = onnx.load('resnet50/model.onnx')\n\n# Convert ONNX model to an ngraph model\n>>> from ngraph_onnx.onnx_importer.importer import import_onnx_model\n>>> ng_function = import_onnx_model(onnx_protobuf)\n\n# The importer returns a list of ngraph models for every ONNX graph output:\n>>> print(ng_function)\n\n```\n\nThis creates an nGraph `Function` object, which can be used to execute a computation on a chosen backend.\n\n### Running a computation\n\nAfter importing an ONNX model, you will have an nGraph `Function` object. \nNow you can create an nGraph `Runtime` backend and use it to compile your `Function` to a backend-specific `Computation` object.\nFinally, you can execute your model by calling the created `Computation` object with input data.\n\n```python\n# Using an ngraph runtime (CPU backend) create a callable computation object\n>>> import ngraph as ng\n>>> runtime = ng.runtime(backend_name='CPU')\n>>> resnet_on_cpu = runtime.computation(ng_function)\n\n# Load an image (or create a mock as in this example)\n>>> import numpy as np\n>>> picture = np.ones([1, 3, 224, 224], dtype=np.float32)\n\n# Run computation on the picture:\n>>> resnet_on_cpu(picture)\n[array([[2.16105007e-04, 5.58412226e-04, 9.70510227e-05, 5.76671446e-05,\n 7.45318757e-05, 4.80892748e-04, 5.67404088e-04, 9.48728994e-05,\n ...\n```\n\n[onnx]: http://onnx.ai/\n[onnx_model_zoo]: https://github.com/onnx/models\n[ngraph_github]: https://github.com/NervanaSystems/ngraph\n[building]: https://github.com/NervanaSystems/ngraph-onnx/blob/master/BUILDING.md\n\n\n", "description_content_type": "text/markdown", "docs_url": null, "download_url": "", "downloads": { "last_day": -1, "last_month": -1, "last_week": -1 }, "home_page": "https://github.com/NervanaSystems/ngraph-onnx", "keywords": "", "license": "License :: OSI Approved :: Apache Software License", "maintainer": "", "maintainer_email": "", "name": "ngraph-onnx", "package_url": "https://pypi.org/project/ngraph-onnx/", "platform": "", "project_url": "https://pypi.org/project/ngraph-onnx/", "project_urls": { "Homepage": "https://github.com/NervanaSystems/ngraph-onnx" }, "release_url": "https://pypi.org/project/ngraph-onnx/0.24.0/", "requires_dist": [ "cachetools", "ngraph-core", "numpy", "onnx", "setuptools" ], "requires_python": "", "summary": "nGraph Backend for ONNX", "version": "0.24.0" }, "last_serial": 5621083, "releases": { "0.10.0": [ { "comment_text": "", "digests": { "md5": "133de2e09af2baddf49cb5feed176654", "sha256": "cb4fe0726f235329f2ccb2c913f24f0d2e1550a3ca7bbe3c3754d11c281cb5e8" }, "downloads": -1, "filename": "ngraph_onnx-0.10.0-py3-none-any.whl", "has_sig": false, "md5_digest": "133de2e09af2baddf49cb5feed176654", "packagetype": "bdist_wheel", "python_version": "py3", "requires_python": null, "size": 55378, "upload_time": "2018-12-01T00:23:29", "url": "https://files.pythonhosted.org/packages/20/08/7fa2dd4533e303f1c896d47237701093b1eabce779bb69e0cf60a8804453/ngraph_onnx-0.10.0-py3-none-any.whl" } ], "0.10.0rc5": [ { "comment_text": "", "digests": { "md5": "87898b70fda24e87f1a3d96323f7b734", "sha256": "351f426e42b05cbc9274febb0f26a1e34b64116e35cafb85e1a4bd41c04ba8df" }, "downloads": -1, "filename": "ngraph_onnx-0.10.0rc5-py3-none-any.whl", "has_sig": false, "md5_digest": "87898b70fda24e87f1a3d96323f7b734", "packagetype": "bdist_wheel", "python_version": "py3", "requires_python": null, "size": 51419, "upload_time": "2018-11-30T05:57:23", "url": "https://files.pythonhosted.org/packages/f0/6d/da0f658a366be0dd992b136dd8a7b67dd1781ef295a67308d9b0e89c8798/ngraph_onnx-0.10.0rc5-py3-none-any.whl" } ], "0.24.0": [ { "comment_text": "", "digests": { "md5": "100e29027837a495b2fd1f8fff844280", "sha256": "74610fe1883df36c3dd8e6071dde4b8833759db6aabcdcb3db8f3c1b1cacdd96" }, "downloads": -1, "filename": "ngraph_onnx-0.24.0-py3-none-any.whl", "has_sig": false, "md5_digest": "100e29027837a495b2fd1f8fff844280", "packagetype": "bdist_wheel", "python_version": "py3", "requires_python": null, "size": 20416, "upload_time": "2019-08-01T22:17:44", "url": "https://files.pythonhosted.org/packages/53/82/8440d4bf3e0de3fbea9cdc4fe627dfb66757f2d63d8b342532f7067ce09e/ngraph_onnx-0.24.0-py3-none-any.whl" } ] }, "urls": [ { "comment_text": "", "digests": { "md5": "100e29027837a495b2fd1f8fff844280", "sha256": "74610fe1883df36c3dd8e6071dde4b8833759db6aabcdcb3db8f3c1b1cacdd96" }, "downloads": -1, "filename": "ngraph_onnx-0.24.0-py3-none-any.whl", "has_sig": false, "md5_digest": "100e29027837a495b2fd1f8fff844280", "packagetype": "bdist_wheel", "python_version": "py3", "requires_python": null, "size": 20416, "upload_time": "2019-08-01T22:17:44", "url": "https://files.pythonhosted.org/packages/53/82/8440d4bf3e0de3fbea9cdc4fe627dfb66757f2d63d8b342532f7067ce09e/ngraph_onnx-0.24.0-py3-none-any.whl" } ] }