returnn


Namereturnn JSON
Version 1.20250709.140629 PyPI version JSON
download
home_pagehttps://github.com/rwth-i6/returnn/
SummaryThe RWTH extensible training framework for universal recurrent neural networks
upload_time2025-07-09 12:20:40
maintainerNone
docs_urlNone
authorAlbert Zeyer
requires_pythonNone
licenseRETURNN license
keywords
VCS
bugtrack_url
requirements numpy h5py dm-tree
Travis-CI No Travis.
coveralls test coverage No coveralls.
            ==================
Welcome to RETURNN
==================

`GitHub repository <https://github.com/rwth-i6/returnn>`__.
`RETURNN paper 2016 <https://arxiv.org/abs/1608.00895>`_,
`RETURNN paper 2018 <https://arxiv.org/abs/1805.05225>`_.

RETURNN - RWTH extensible training framework for universal recurrent neural networks,
is a Theano/TensorFlow-based implementation of modern recurrent neural network architectures.
It is optimized for fast and reliable training of recurrent neural networks in a multi-GPU environment.

The high-level features and goals of RETURNN are:

* **Simplicity**

  * Writing config / code is simple & straight-forward (setting up experiment, defining model)
  * Debugging in case of problems is simple
  * Reading config / code is simple (defined model, training, decoding all becomes clear)

* **Flexibility**

  * Allow for many different kinds of experiments / models

* **Efficiency**

  * Training speed
  * Decoding speed

All items are important for research, decoding speed is esp. important for production.

See our `Interspeech 2020 tutorial "Efficient and Flexible Implementation of Machine Learning for ASR and MT" video <https://www.youtube.com/watch?v=wPKdYqSOlAY>`__
(`slides <https://www-i6.informatik.rwth-aachen.de/publications/download/1154/Zeyer--2020.pdf>`__)
with an introduction of the core concepts.

More specific features include:

- Mini-batch training of feed-forward neural networks
- Sequence-chunking based batch training for recurrent neural networks
- Long short-term memory recurrent neural networks
  including our own fast CUDA kernel
- Multidimensional LSTM (GPU only, there is no CPU version)
- Memory management for large data sets
- Work distribution across multiple devices
- Flexible and fast architecture which allows all kinds of encoder-attention-decoder models

See `documentation <https://returnn.readthedocs.io/>`__.
See `basic usage <https://returnn.readthedocs.io/en/latest/basic_usage.html>`__
and `technological overview <https://returnn.readthedocs.io/en/latest/tech_overview.html>`__.

`Here is the video recording of a RETURNN overview talk <https://www-i6.informatik.rwth-aachen.de/web/Software/returnn/downloads/workshop-2019-01-29/01.recording.cut.mp4>`_
(`slides <https://www-i6.informatik.rwth-aachen.de/web/Software/returnn/downloads/workshop-2019-01-29/01.returnn-overview.session1.handout.v1.pdf>`__,
`exercise sheet <https://www-i6.informatik.rwth-aachen.de/web/Software/returnn/downloads/workshop-2019-01-29/01.exercise_sheet.pdf>`__;
hosted by eBay).

There are `many example demos <https://github.com/rwth-i6/returnn/blob/master/demos/>`_
which work on artificially generated data,
i.e. they should work as-is.

There are `some real-world examples <https://github.com/rwth-i6/returnn-experiments>`_
such as setups for speech recognition on the Switchboard or LibriSpeech corpus.

Some benchmark setups against other frameworks
can be found `here <https://github.com/rwth-i6/returnn-benchmarks>`_.
The results are in the `RETURNN paper 2016 <https://arxiv.org/abs/1608.00895>`_.
Performance benchmarks of our LSTM kernel vs CuDNN and other TensorFlow kernels
are in `TensorFlow LSTM benchmark <https://returnn.readthedocs.io/en/latest/tf_lstm_benchmark.html>`__.

There is also `a wiki <https://github.com/rwth-i6/returnn/wiki>`_.
Questions can also be asked on
`StackOverflow using the RETURNN tag <https://stackoverflow.com/questions/tagged/returnn>`_.

.. image:: https://github.com/rwth-i6/returnn/workflows/CI/badge.svg
    :target: https://github.com/rwth-i6/returnn/actions

Dependencies
============

pip dependencies are listed in ``requirements.txt`` and ``requirements-dev``,
although some parts of the code may require additional dependencies (e.g. ``librosa``, ``resampy``) on-demand.

RETURNN supports Python >= 3.8. Bumps to the minimum Python version are listed in `CHANGELOG.md <https://github.com/rwth-i6/returnn/blob/master/CHANGELOG.md>`__.

TensorFlow-based setups require TensorFlow >= 2.2.

PyTorch-based setups require Torch >= 1.0.

            

Raw data

            {
    "_id": null,
    "home_page": "https://github.com/rwth-i6/returnn/",
    "name": "returnn",
    "maintainer": null,
    "docs_url": null,
    "requires_python": null,
    "maintainer_email": null,
    "keywords": null,
    "author": "Albert Zeyer",
    "author_email": "albzey@gmail.com",
    "download_url": "https://files.pythonhosted.org/packages/c0/11/75cfa8c3fa9a8ed5a04f1e94f4e9851be6c51aea6caaf3e5888252ef95a1/returnn-1.20250709.140629.tar.gz",
    "platform": null,
    "description": "==================\nWelcome to RETURNN\n==================\n\n`GitHub repository <https://github.com/rwth-i6/returnn>`__.\n`RETURNN paper 2016 <https://arxiv.org/abs/1608.00895>`_,\n`RETURNN paper 2018 <https://arxiv.org/abs/1805.05225>`_.\n\nRETURNN - RWTH extensible training framework for universal recurrent neural networks,\nis a Theano/TensorFlow-based implementation of modern recurrent neural network architectures.\nIt is optimized for fast and reliable training of recurrent neural networks in a multi-GPU environment.\n\nThe high-level features and goals of RETURNN are:\n\n* **Simplicity**\n\n  * Writing config / code is simple & straight-forward (setting up experiment, defining model)\n  * Debugging in case of problems is simple\n  * Reading config / code is simple (defined model, training, decoding all becomes clear)\n\n* **Flexibility**\n\n  * Allow for many different kinds of experiments / models\n\n* **Efficiency**\n\n  * Training speed\n  * Decoding speed\n\nAll items are important for research, decoding speed is esp. important for production.\n\nSee our `Interspeech 2020 tutorial \"Efficient and Flexible Implementation of Machine Learning for ASR and MT\" video <https://www.youtube.com/watch?v=wPKdYqSOlAY>`__\n(`slides <https://www-i6.informatik.rwth-aachen.de/publications/download/1154/Zeyer--2020.pdf>`__)\nwith an introduction of the core concepts.\n\nMore specific features include:\n\n- Mini-batch training of feed-forward neural networks\n- Sequence-chunking based batch training for recurrent neural networks\n- Long short-term memory recurrent neural networks\n  including our own fast CUDA kernel\n- Multidimensional LSTM (GPU only, there is no CPU version)\n- Memory management for large data sets\n- Work distribution across multiple devices\n- Flexible and fast architecture which allows all kinds of encoder-attention-decoder models\n\nSee `documentation <https://returnn.readthedocs.io/>`__.\nSee `basic usage <https://returnn.readthedocs.io/en/latest/basic_usage.html>`__\nand `technological overview <https://returnn.readthedocs.io/en/latest/tech_overview.html>`__.\n\n`Here is the video recording of a RETURNN overview talk <https://www-i6.informatik.rwth-aachen.de/web/Software/returnn/downloads/workshop-2019-01-29/01.recording.cut.mp4>`_\n(`slides <https://www-i6.informatik.rwth-aachen.de/web/Software/returnn/downloads/workshop-2019-01-29/01.returnn-overview.session1.handout.v1.pdf>`__,\n`exercise sheet <https://www-i6.informatik.rwth-aachen.de/web/Software/returnn/downloads/workshop-2019-01-29/01.exercise_sheet.pdf>`__;\nhosted by eBay).\n\nThere are `many example demos <https://github.com/rwth-i6/returnn/blob/master/demos/>`_\nwhich work on artificially generated data,\ni.e. they should work as-is.\n\nThere are `some real-world examples <https://github.com/rwth-i6/returnn-experiments>`_\nsuch as setups for speech recognition on the Switchboard or LibriSpeech corpus.\n\nSome benchmark setups against other frameworks\ncan be found `here <https://github.com/rwth-i6/returnn-benchmarks>`_.\nThe results are in the `RETURNN paper 2016 <https://arxiv.org/abs/1608.00895>`_.\nPerformance benchmarks of our LSTM kernel vs CuDNN and other TensorFlow kernels\nare in `TensorFlow LSTM benchmark <https://returnn.readthedocs.io/en/latest/tf_lstm_benchmark.html>`__.\n\nThere is also `a wiki <https://github.com/rwth-i6/returnn/wiki>`_.\nQuestions can also be asked on\n`StackOverflow using the RETURNN tag <https://stackoverflow.com/questions/tagged/returnn>`_.\n\n.. image:: https://github.com/rwth-i6/returnn/workflows/CI/badge.svg\n    :target: https://github.com/rwth-i6/returnn/actions\n\nDependencies\n============\n\npip dependencies are listed in ``requirements.txt`` and ``requirements-dev``,\nalthough some parts of the code may require additional dependencies (e.g. ``librosa``, ``resampy``) on-demand.\n\nRETURNN supports Python >= 3.8. Bumps to the minimum Python version are listed in `CHANGELOG.md <https://github.com/rwth-i6/returnn/blob/master/CHANGELOG.md>`__.\n\nTensorFlow-based setups require TensorFlow >= 2.2.\n\nPyTorch-based setups require Torch >= 1.0.\n",
    "bugtrack_url": null,
    "license": "RETURNN license",
    "summary": "The RWTH extensible training framework for universal recurrent neural networks",
    "version": "1.20250709.140629",
    "project_urls": {
        "Homepage": "https://github.com/rwth-i6/returnn/"
    },
    "split_keywords": [],
    "urls": [
        {
            "comment_text": null,
            "digests": {
                "blake2b_256": "fea9a6f3cc1f6d30f6d5fa394415234dd846af419a4f92b281dbe3de9356efb8",
                "md5": "cea579780c0284b732db47f989a711a3",
                "sha256": "5d3084e8019b66bd2ae05ea070f84a1b8c23c3ef4f0f94ad334d914819a67877"
            },
            "downloads": -1,
            "filename": "returnn-1.20250709.140629-py3-none-any.whl",
            "has_sig": false,
            "md5_digest": "cea579780c0284b732db47f989a711a3",
            "packagetype": "bdist_wheel",
            "python_version": "py3",
            "requires_python": null,
            "size": 1501685,
            "upload_time": "2025-07-09T12:20:37",
            "upload_time_iso_8601": "2025-07-09T12:20:37.644596Z",
            "url": "https://files.pythonhosted.org/packages/fe/a9/a6f3cc1f6d30f6d5fa394415234dd846af419a4f92b281dbe3de9356efb8/returnn-1.20250709.140629-py3-none-any.whl",
            "yanked": false,
            "yanked_reason": null
        },
        {
            "comment_text": null,
            "digests": {
                "blake2b_256": "c01175cfa8c3fa9a8ed5a04f1e94f4e9851be6c51aea6caaf3e5888252ef95a1",
                "md5": "f9ea55bb984f22374a2caf938769027e",
                "sha256": "8674ced813efdf13fe766872f8b1305b266ac6f2ca0abcc49566a20a45d6fc84"
            },
            "downloads": -1,
            "filename": "returnn-1.20250709.140629.tar.gz",
            "has_sig": false,
            "md5_digest": "f9ea55bb984f22374a2caf938769027e",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": null,
            "size": 2353761,
            "upload_time": "2025-07-09T12:20:40",
            "upload_time_iso_8601": "2025-07-09T12:20:40.988919Z",
            "url": "https://files.pythonhosted.org/packages/c0/11/75cfa8c3fa9a8ed5a04f1e94f4e9851be6c51aea6caaf3e5888252ef95a1/returnn-1.20250709.140629.tar.gz",
            "yanked": false,
            "yanked_reason": null
        }
    ],
    "upload_time": "2025-07-09 12:20:40",
    "github": true,
    "gitlab": false,
    "bitbucket": false,
    "codeberg": false,
    "github_user": "rwth-i6",
    "github_project": "returnn",
    "travis_ci": false,
    "coveralls": false,
    "github_actions": true,
    "requirements": [
        {
            "name": "numpy",
            "specs": []
        },
        {
            "name": "h5py",
            "specs": []
        },
        {
            "name": "dm-tree",
            "specs": []
        }
    ],
    "lcname": "returnn"
}
        
Elapsed time: 1.86031s