{ "info": { "author": "TATi Team", "author_email": "", "bugtrack_url": null, "classifiers": [ "Intended Audience :: Science/Research", "License :: OSI Approved :: GNU General Public License v3 (GPLv3)", "Operating System :: POSIX", "Programming Language :: Python :: 3", "Programming Language :: Python :: 3.4", "Programming Language :: Python :: 3.5", "Programming Language :: Python :: 3.6", "Topic :: Scientific/Engineering :: Artificial Intelligence", "Topic :: Scientific/Engineering :: Mathematics", "Topic :: Scientific/Engineering :: Physics" ], "description": "Thermodynamic Analytics ToolkIt (TATi)\n======================================\n\nThermodynamic Analytics Toolkit is a sampling-based approach to understand the\neffectiveness of neural networks training and investigate their loss manifolds.\n\nIt uses Tensorflow (https://www.tensorflow.org/) as neural network\nframework and implements advanced sampling algorithms on top of it. It contains\nboth a rapid prototyping platform for new sampling methods and also an analysis\nframework to understand the intricacies of the loss manifold in terms of\naverages, covariance, diffusion maps, and free energy.\n\nPlease take a look at the extensive [userguide](https://alan-turing-institute.github.io/ThermodynamicAnalyticsToolkit/).\n\nDependencies\n------------\n\nIn total, we depend on the following python packages:\n\n * tensorflow (1.4.1, 1.6-1.10; 1.5 is not recommended)\n * numpy\n * pandas\n * scipy\n * scikit-learn\n * acor (see the userguide for installation instructions)\n\nFurthermore, for installation from a cloned git repository or a pure source\ntarball, the following non-python packages are required for creating all \nuserguides,\n\n * doxygen,\n * asciidoc, dblatex\n * pdflatex,\n\nand for running all tests,\n\n * awk, sqlite3.\n\nFinally, there are some optional python packages:\n\n * pydiffmap: allows diffusion map analysis through pydiffmap package\n * tqdm: allows displaying progress bar during training and sampling\n\nInstallation\n------------\n\nUse one of the following ways:\n\n- `pip install tati`\n- Grab latest [release](https://github.com/alan-turing-institute/ThermodynamicAnalyticsToolkit/releases), extract and `configure --prefix=`, `make`, `make install`\n- `git clone https://github.com/alan-turing-institute/ThermodynamicAnalyticsToolkit.git` and `configure --prefix=`, `make`, `make install`.\n\nFor more information please refer to the userguide (see the \n[releases](https://github.com/alan-turing-institute/ThermodynamicAnalyticsToolkit/releases) on github or [as html version](https://alan-turing-institute.github.io/ThermodynamicAnalyticsToolkit/)) \nfor installation instructions.\n\nAlternatively, the userguide PDF is also contained in the release tarballs in \nfolder **doc/userguide**.\nAs a fall-back the asciidoc userguide files reside in **doc/userguide** \nand are perfectly human-readable, see **doc/userguide/introduction.txt**.\nAs a last fall-back have a look at INSTALL for general instructions on how to\ninstalling a package maintained by autotools, automake.\n\nWhen cloning from github please call the `./bootstrap.sh` script (requiring\ninstalled autotools and automake packages).\n\nNOTE: If you only want to *use* the package and *do not plan to submit code*, \nit is strongly advised to *use the PyPI package (using `pip`) or the \"release\" \ntarballs* instead of cloning the repository directly.\n\nDocumentation\n-------------\n\nIn general, the documentation is maintained in the folder **doc**. The asciidoc\nuserguide files reside in **doc/userguide** and are human-readable in your \npreferred editor if every other option fails.\n\nThere are multiple guides to help you:\n\n- Userguide: user manual on how to install and use TATi\n- Programmer's guide: manual on basic programming with Tensorflow and TATi\n- API reference: doxygen-generated API reference\n\nAfter installation (configure, make, make doc, make install) these guides\ncan be found in the typical documentation directory (e.g., \n**share/doc/thermodynamicanalyticstoolkit/** depending on your OS).\n\nNote that all of the above guides are also available as *html* versions after\ninstallation.\n\nAcknowledgments\n---------------\n\nTATi has received financial support from a seed funding grant and through a \nRutherford fellowship from the Alan Turing Institute in London (R-SIS-003, \nR-RUT-001), from an EPSRC grant no. EP/P006175/1 (Data Driven Coarse Graining\nusing Space-Time Diffusion Maps, B. Leimkuhler PI), and also from a Microsoft\nAzure Sponsorship (MS-AZR-0143P).\n\n\n\n", "description_content_type": "", "docs_url": null, "download_url": "", "downloads": { "last_day": -1, "last_month": -1, "last_week": -1 }, "home_page": "https://github.com/alan-turing-institute/ThermodynamicAnalyticsToolkit", "keywords": "", "license": "GPL-3", "maintainer": "Frederik Heber", "maintainer_email": "frederik.heber@gmail.com", "name": "tati", "package_url": "https://pypi.org/project/tati/", "platform": "", "project_url": "https://pypi.org/project/tati/", "project_urls": { "Bug Tracker": "https://github.com/alan-turing-institute/ThermodynamicAnalyticsToolkit/issues", "Documentation": "https://thermodynamicanalyticstoolkit.github.io/", "Homepage": "https://github.com/alan-turing-institute/ThermodynamicAnalyticsToolkit", "Source Code": "https://github.com/alan-turing-institute/ThermodynamicAnalyticsToolkit/tree" }, "release_url": "https://pypi.org/project/tati/0.9.5/", "requires_dist": [ "matplotlib", "numpy (>=1.7)", "pandas", "scipy", "scikit-learn", "tensorflow (>=1.4.1)", "acor; extra == 'IAT'" ], "requires_python": "", "summary": "ThermodynamicAnalyticsToolkit - analyze loss manifolds of neural networks", "version": "0.9.5" }, "last_serial": 5004995, "releases": { "0.9.4": [ { "comment_text": "", "digests": { "md5": "6cc0b4afcceef3f6fe603fc3be133618", "sha256": "1494a09c1b09701e75051b1138c83596a4c422351b1360ce5b920c8dd03cbd00" }, "downloads": -1, "filename": "tati-0.9.4-py3-none-any.whl", "has_sig": false, "md5_digest": "6cc0b4afcceef3f6fe603fc3be133618", "packagetype": "bdist_wheel", "python_version": "py3", "requires_python": null, "size": 230208, "upload_time": "2019-01-28T08:50:03", "url": "https://files.pythonhosted.org/packages/97/06/8a8eac58a66c7097abe150f0eb4f15039eef9f01b2a9f4b7c4f9353acafa/tati-0.9.4-py3-none-any.whl" } ], "0.9.5": [ { "comment_text": "", "digests": { "md5": "cab87b869a59c08fcf648258c7a8afb3", "sha256": "35f06b2822820ac5382c469ed97b410f6a00df5dcd5dea1f2f77bd8815a41bca" }, "downloads": -1, "filename": "tati-0.9.5-py3-none-any.whl", "has_sig": false, "md5_digest": "cab87b869a59c08fcf648258c7a8afb3", "packagetype": "bdist_wheel", "python_version": "py3", "requires_python": null, "size": 235811, "upload_time": "2019-03-29T20:37:36", "url": "https://files.pythonhosted.org/packages/4e/11/9376fda32ad0ff6686a10779984f22b6d4b73ae3b2edb2889ac7ba59d947/tati-0.9.5-py3-none-any.whl" } ] }, "urls": [ { "comment_text": "", "digests": { "md5": "cab87b869a59c08fcf648258c7a8afb3", "sha256": "35f06b2822820ac5382c469ed97b410f6a00df5dcd5dea1f2f77bd8815a41bca" }, "downloads": -1, "filename": "tati-0.9.5-py3-none-any.whl", "has_sig": false, "md5_digest": "cab87b869a59c08fcf648258c7a8afb3", "packagetype": "bdist_wheel", "python_version": "py3", "requires_python": null, "size": 235811, "upload_time": "2019-03-29T20:37:36", "url": "https://files.pythonhosted.org/packages/4e/11/9376fda32ad0ff6686a10779984f22b6d4b73ae3b2edb2889ac7ba59d947/tati-0.9.5-py3-none-any.whl" } ] }