{ "info": { "author": "Ondrej Biza", "author_email": "ondrej.biza@gmail.com", "bugtrack_url": null, "classifiers": [ "Development Status :: 5 - Production/Stable", "Intended Audience :: Science/Research", "License :: OSI Approved :: MIT License", "Natural Language :: English", "Programming Language :: Python :: 3", "Topic :: Scientific/Engineering :: Artificial Intelligence" ], "description": "tfset\n=====\n\n.. figure:: tfset/images/validation_curve.png\n :alt: Validation Curve\n\n**Change the hyper-parameters** of your Tensorflow training session on\nthe fly. The package allows you to schedule events that change the\nvalues of arbitrary Tensors with a simple command.\n\nRequirements\n~~~~~~~~~~~~\n\n- Python >= 3\n- tensorflow >= 1.0\n\nSet Up\n~~~~~~\n\nInstall the package with pip:\n\n``pip install tfset``\n\nOr clone and install from github:\n\n.. code-block:: bash\n\n git clone https://github.com/ondrejba/tfset.git\n cd tfset\n python setup.py install\n\nUsage\n~~~~~\n\ntfset DEMO\n^^^^^^^^^^^^^^^^^^^^^^^^^^^\n\nCheck\n`MNIST\\_demo.ipynb `__\nfor a demostration of the usage of tfset in a simple\ntraining script.\n\nServer\n^^^^^^\n\nImport tfset server.\n\n::\n\n import tfset.server as server\n\nCreate Tensors for your hyper-parameters.\n\n::\n\n learning_rate = tf.get_variable(\"learning_rate\", initializer=tf.constant(0.1, dtype=tf.float32))\n dropout_prob = tf.get_variable(\"dropout_prob\", initializer=tf.constant(0.9, dtype=tf.float32))\n\nCreate and start a Session Server.\n\n::\n\n # \"session\" is a Tensorflow session\n s, thread = server.run_server([learning_rate, dropout_prob], session)\n\nPeriodically check for events.\n\n::\n\n # \"step\" is the global step of your training procedure\n s.check_events(step)\n\nStop the server.\n\n::\n\n s.shutdown()\n thread.join(timeout=10)\n\nClient\n^^^^^^\n\nGet status.\n\n``tfset -s``\n\nAdd an event (this event sets the learning rate to 0.01 at iteration\n10000).\n\n``tfset -a -n learning_rate:0 -i 10000 --value 0.01``\n\nRemove an event (with index 0 in this case).\n\n``tfset -r -e 0``\n\nEvents\n^^^^^^\n\ntfset schedules hyper-parameter changes based on\n**events**. An event contains the following information:\n\n- **iteration**: when to execute the event\n- **Tensor name**: which Tensor to change\n- **value**: value to set the Tensor to\n\nThe reason for the use of events is that you might want to schedule\nhyper-parameter changes in the future (e.g. lower learning rate to 10e-3\nat 800k iteration). If two events targeting the same Tensor are\nscheduled at the same iteration, the one that was scheduled later is\ngoing to be executed.\n", "description_content_type": null, "docs_url": null, "download_url": "", "downloads": { "last_day": -1, "last_month": -1, "last_week": -1 }, "home_page": "https://github.com/ondrejba/tfset", "keywords": "tensorflow interactive", "license": "MIT", "maintainer": "", "maintainer_email": "", "name": "tfset", "package_url": "https://pypi.org/project/tfset/", "platform": "", "project_url": "https://pypi.org/project/tfset/", "project_urls": { "Homepage": "https://github.com/ondrejba/tfset" }, "release_url": "https://pypi.org/project/tfset/1.2/", "requires_dist": null, "requires_python": "", "summary": "Set Tensor values during training in Tensorflow.", "version": "1.2" }, "last_serial": 3545378, "releases": { "1.0": [ { "comment_text": "", "digests": { "md5": "78d87de7a67223be8f187f57486c3e60", "sha256": "b0fc38555fcd2aff861dc285ec55f7c1bf4f7a5343c7bd647812619141252128" }, "downloads": -1, "filename": "tfset-1.0.tar.gz", "has_sig": false, "md5_digest": "78d87de7a67223be8f187f57486c3e60", "packagetype": "sdist", "python_version": "source", "requires_python": null, "size": 5630, "upload_time": "2018-02-01T22:45:57", "url": "https://files.pythonhosted.org/packages/fd/a9/eb4543c4701024cc2c024088547d4f6410b8e5f7d621e8bf7e8883fb5509/tfset-1.0.tar.gz" } ], "1.1": [ { "comment_text": "", "digests": { "md5": "025093f383e3291f20a9e1524b665bb9", "sha256": "83a5297a21c7c8906cc906da43672afbf3fc6798f47c12a526e1aab0a666ed85" }, "downloads": -1, "filename": "tfset-1.1.tar.gz", "has_sig": false, "md5_digest": "025093f383e3291f20a9e1524b665bb9", "packagetype": "sdist", "python_version": "source", "requires_python": null, "size": 5531, "upload_time": "2018-02-02T11:55:17", "url": "https://files.pythonhosted.org/packages/fb/68/ce8f6af9c9466c4e18ba5a2d28a657e1e55bfd64d9b1abf620507a909119/tfset-1.1.tar.gz" } ], "1.2": [ { "comment_text": "", "digests": { "md5": "d64585466d9ea3a1cec134aaffde6612", "sha256": "4a235151951e9631a93ac6cd484eb510cde93e6b5da7003e032a93fa983b0e65" }, "downloads": -1, "filename": "tfset-1.2.tar.gz", "has_sig": false, "md5_digest": "d64585466d9ea3a1cec134aaffde6612", "packagetype": "sdist", "python_version": "source", "requires_python": null, "size": 5540, "upload_time": "2018-02-02T12:08:03", "url": "https://files.pythonhosted.org/packages/72/d6/7de30d0b6311eab2ab38ee98cf1e02f6fb28b4577bef1e00e415c019fe2c/tfset-1.2.tar.gz" } ] }, "urls": [ { "comment_text": "", "digests": { "md5": "d64585466d9ea3a1cec134aaffde6612", "sha256": "4a235151951e9631a93ac6cd484eb510cde93e6b5da7003e032a93fa983b0e65" }, "downloads": -1, "filename": "tfset-1.2.tar.gz", "has_sig": false, "md5_digest": "d64585466d9ea3a1cec134aaffde6612", "packagetype": "sdist", "python_version": "source", "requires_python": null, "size": 5540, "upload_time": "2018-02-02T12:08:03", "url": "https://files.pythonhosted.org/packages/72/d6/7de30d0b6311eab2ab38ee98cf1e02f6fb28b4577bef1e00e415c019fe2c/tfset-1.2.tar.gz" } ] }