Metadata-Version: 2.1
Name: cockpit-for-pytorch
Version: 1.0.2
Summary: A Practical Debugging Tool for Training Deep Neural Networks.
Home-page: https://github.com/f-dangel/cockpit
Author: Frank Schneider and Felix Dangel
Author-email: f.schneider@uni-tuebingen.de
License: MIT
Keywords: deep-learning,machine-learning,debugging
Platform: any
Classifier: Development Status :: 4 - Beta
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Classifier: Programming Language :: Python :: 3.7
Classifier: Programming Language :: Python :: 3.8
Classifier: Programming Language :: Python :: 3.9
Requires-Python: >=3.7
Description-Content-Type: text/markdown
License-File: LICENSE.txt
License-File: AUTHORS.md

<!-- PROJECT LOGO -->
<br />
<p align="center">
<a href="#"><img src="docs/source/_static/Banner.svg" alt="Logo"/></a>
  <h3 align="center">A Practical Debugging Tool for Training Deep Neural Networks</h3>

  <p align="center">
    A better status screen for deep learning.
  </p>
</p>

<p align="center">
  <a href="#installation">Installation</a> •
  <a href="https://cockpit.readthedocs.io/">Docs</a> •
  <a href="https://github.com/fsschneider/cockpit-experiments">Experiments</a> •
  <a href="#license">License</a> •
  <a href="#citation">Citation</a>
</p>

[![CI](https://github.com/f-dangel/cockpit/actions/workflows/CI.yml/badge.svg)](https://github.com/f-dangel/cockpit/actions/workflows/CI.yml)
[![Lint](https://github.com/f-dangel/cockpit/actions/workflows/Lint.yml/badge.svg)](https://github.com/f-dangel/cockpit/actions/workflows/Lint.yml)
[![Doc](https://img.shields.io/readthedocs/cockpit/latest.svg?logo=read%20the%20docs&logoColor=white&label=Doc)](https://cockpit.readthedocs.io)
[![Coverage](https://coveralls.io/repos/github/f-dangel/cockpit/badge.svg?branch=main&t=piyZHm)](https://coveralls.io/github/f-dangel/cockpit?branch=main)
[![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](https://github.com/f-dangel/cockpit/blob/master/LICENSE)
[![Code style: black](https://img.shields.io/badge/code%20style-black-000000.svg)](https://github.com/psf/black)
[![arXiv](https://img.shields.io/static/v1?logo=arxiv&logoColor=white&label=Preprint&message=2102.06604&color=B31B1B)](https://arxiv.org/abs/2102.06604)

---

```bash
pip install cockpit-for-pytorch
```

---

**Cockpit is a visual and statistical debugger specifically designed for deep learning.** Training a deep neural network is often a pain! Successfully training such a network usually requires either years of intuition or expensive parameter searches involving lots of trial and error. Traditional debuggers provide only limited help: They can find *syntactical errors* but not *training bugs* such as ill-chosen learning rates. **Cockpit** offers a closer, more meaningful look into the training process with multiple well-chosen *instruments*.

---

![CockpitAnimation](docs/source/_static/showcase.gif)

<!-- Installation -->
## Installation

To install **Cockpit** simply run

```bash
pip install cockpit-for-pytorch
```

<details>
<summary>Conda environment</summary>
For convenience, we also provide a conda environment, which can be installed via the conda yml file. It includes all the necessary requirements to build the docs, execute the tests and run the examples.
</details>

<!-- Documentation -->
## Documentation

The [documentation](https://cockpit.readthedocs.io/) provides a full tutorial on how to get started using **Cockpit** as well as a detailed documentation of its API.

<!-- Experiments -->
## Experiments

To showcase the capabilities of **Cockpit** we performed several experiments illustrating the usefulness of our debugging tool. The code for the experiments can be found in a [separate repository](https://github.com/fsschneider/cockpit-experiments). For a discussion of those experiments please refer to our [paper](https://arxiv.org/abs/2102.06604).

<!-- LICENSE -->
## License

Distributed under the MIT License. See [`LICENSE`](LICENSE.txt) for more information.

<!-- Citation -->
## Citation

If you use **Cockpit**, please consider citing:

> [Frank Schneider, Felix Dangel, Philipp Hennig<br/>
> **Cockpit: A Practical Debugging Tool for Training Deep Neural Networks**<br/>
> *arXiv 2102.06604*](http://arxiv.org/abs/2102.06604)

```bibtex
@misc{schneider2021cockpit,
   title={{Cockpit: A Practical Debugging Tool for Training Deep Neural Networks}},
   author={Frank Schneider and Felix Dangel and Philipp Hennig},
   year={2021},
   eprint={2102.06604},
   archivePrefix={arXiv},
   primaryClass={cs.LG}
}
```

# Changelog

All notable changes to this project will be documented in this file.

The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).

## [Unreleased]

## [1.0.2] - 2021-10-26

### Added

- Added references to a separate [experiment repository](https://github.com/fsschneider/cockpit-experiments) that publishes the code for all experiments shown in the paper.

### Fixed

- Protects the `batch_grad` field in the case where non-SGD is used together with other quantities that free `batch_grad` for memory performance. [[#5](https://github.com/f-dangel/cockpit/issues/5), [PR](https://github.com/f-dangel/cockpit/pull/18)]

## [1.0.1] - 2021-10-13

From this version on, `cockpit` will be available as `cockpit-for-pytorch` on
PyPI.

### Added
- Make library `pip`-installable as `cockpit-for-pytorch`
  [[PR](https://github.com/f-dangel/cockpit/pull/17)]
- Require BackPACK main release
  [[PR](https://github.com/f-dangel/cockpit/pull/12)]
- Added a `savename` argument to the `CockpitPlotter.plot()` function, which lets you define the name, and now the `savedir` should really only describe the **directory**. [[PR](https://github.com/f-dangel/cockpit/pull/16), Fixes #8]
- Added optional `savefig_kwargs` argument to the `CockpitPlotter.plot()` function that gets passed to the `matplotlib` function `fig.savefig()` to, e.g., specify DPI value or use a different file format (e.g. PDF). [[PR](https://github.com/f-dangel/cockpit/pull/16), Fixes #10]

### Internal
- Fix [#6](https://github.com/f-dangel/cockpit/issues/6): Don't execute
  extension hook on modules with children
  [[PR](https://github.com/f-dangel/cockpit/pull/7)]

## [1.0.0] - 2021-04-30

### Added

- First public release version of **Cockpit**.

[Unreleased]: https://github.com/f-dangel/cockpit/compare/v1.0.2...HEAD
[1.0.2]: https://github.com/f-dangel/cockpit/compare/1.0.1...1.0.2
[1.0.1]: https://github.com/f-dangel/cockpit/compare/1.0.0...1.0.1
[1.0.0]: https://github.com/f-dangel/cockpit/releases/tag/1.0.0

MIT License

Copyright (c) 2019 Frank Schneider, Felix Dangel & Philipp Hennig

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.


