{
    "info": {
        "author": "Shkarupa Alex",
        "author_email": "shkarupa.alex@gmail.com",
        "bugtrack_url": null,
        "classifiers": [
            "Development Status :: 5 - Production/Stable",
            "Intended Audience :: Developers",
            "Intended Audience :: Education",
            "Intended Audience :: Science/Research",
            "License :: OSI Approved :: MIT License",
            "Programming Language :: Python :: 2",
            "Programming Language :: Python :: 3",
            "Topic :: Scientific/Engineering",
            "Topic :: Scientific/Engineering :: Artificial Intelligence",
            "Topic :: Software Development",
            "Topic :: Software Development :: Libraries",
            "Topic :: Software Development :: Libraries :: Python Modules"
        ],
        "description": "# tfwn\n\n[Weight Normalization](https://arxiv.org/abs/1602.07868) layer wrapper for TensorFlow-Keras API.\n\nInspired by [Sean Morgan](https://github.com/tensorflow/tensorflow/pull/21276) implementation, but:\n- No data initialization (only eager mode was implemented in original pull request).\n- Code refactoring\n- More tests\n- CIFAR10 example from original paper reimplemented\n\n## Examples\nUnfortunately I couldn't reproduce parer results on CIFAR10 with batch size 100.\nAs you can see there is no much difference in accuracy.\n\n<img src=\"https://github.com/shkarupa-alex/tfwn/raw/master/examples/cifar10_accuracy_100.png\">\n<img src=\"https://github.com/shkarupa-alex/tfwn/raw/master/examples/cifar10_loss_100.png\">\n\n\nBut with much smaller batch size model with weight normalization is much better then regular one.\n\n<img src=\"https://github.com/shkarupa-alex/tfwn/raw/master/examples/cifar10_accuracy_16.png\">\n<img src=\"https://github.com/shkarupa-alex/tfwn/raw/master/examples/cifar10_loss_16.png\">\n\n\n## How to use\n```python\nimport tensorflow as tf\nfrom tfwn import WeightNorm\n\n\ndense_wn = WeightNorm(tf.keras.layers.Dense(3))\nout = dense_wn(input)\n```\n\n\n## References\n### Weight Normalization: A Simple Reparameterization to Accelerate Training of Deep Neural Networks\nTim Salimans, and Diederik P. Kingma.\n\n```\n@inproceedings{Salimans2016WeightNorm,\n  title={Weight Normalization: A Simple Reparameterization to Accelerate Training of Deep Neural Networks},\n  author={Tim Salimans and Diederik P. Kingma},\n  booktitle={Neural Information Processing Systems 2016},\n  year={2016}\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/shkarupa-alex/tfwn",
        "keywords": "",
        "license": "MIT",
        "maintainer": "",
        "maintainer_email": "",
        "name": "tfwn",
        "package_url": "https://pypi.org/project/tfwn/",
        "platform": "",
        "project_url": "https://pypi.org/project/tfwn/",
        "project_urls": {
            "Homepage": "https://github.com/shkarupa-alex/tfwn"
        },
        "release_url": "https://pypi.org/project/tfwn/1.0.1/",
        "requires_dist": null,
        "requires_python": "",
        "summary": "Weight normalization layer for TensorFlow",
        "version": "1.0.1"
    },
    "last_serial": 4818822,
    "releases": {
        "1.0.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "b7d6b054facd6ad5489ac45f93dce87a",
                    "sha256": "ca5fdead2f4098605f08a93b457c636d30403a6019c70f42e8b16420cfcbfc3e"
                },
                "downloads": -1,
                "filename": "tfwn-1.0.0.tar.gz",
                "has_sig": false,
                "md5_digest": "b7d6b054facd6ad5489ac45f93dce87a",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 5023,
                "upload_time": "2019-02-14T05:09:09",
                "url": "https://files.pythonhosted.org/packages/38/16/b77093c6772feb919c12c34df00611df776c8da1123c213ce1885f4e7df7/tfwn-1.0.0.tar.gz"
            }
        ],
        "1.0.1": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "e6f4c8dc45948c0fc8d2dc9109f8a4cb",
                    "sha256": "4572f44092a7ab4a2a2c452f57b5090de626d500d879b3e4c7cfbdaf5c13e0a5"
                },
                "downloads": -1,
                "filename": "tfwn-1.0.1.tar.gz",
                "has_sig": false,
                "md5_digest": "e6f4c8dc45948c0fc8d2dc9109f8a4cb",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 5009,
                "upload_time": "2019-02-14T05:15:07",
                "url": "https://files.pythonhosted.org/packages/bd/5b/98ac0c3b972f2550c5ce50c320ad65b14d70c38b5ec75f14bb504070ca0c/tfwn-1.0.1.tar.gz"
            }
        ]
    },
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "e6f4c8dc45948c0fc8d2dc9109f8a4cb",
                "sha256": "4572f44092a7ab4a2a2c452f57b5090de626d500d879b3e4c7cfbdaf5c13e0a5"
            },
            "downloads": -1,
            "filename": "tfwn-1.0.1.tar.gz",
            "has_sig": false,
            "md5_digest": "e6f4c8dc45948c0fc8d2dc9109f8a4cb",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": null,
            "size": 5009,
            "upload_time": "2019-02-14T05:15:07",
            "url": "https://files.pythonhosted.org/packages/bd/5b/98ac0c3b972f2550c5ce50c320ad65b14d70c38b5ec75f14bb504070ca0c/tfwn-1.0.1.tar.gz"
        }
    ]
}