# TorchToolkit
[![PyPI version fury.io](https://badge.fury.io/py/torchtoolkit.svg)](https://pypi.python.org/pypi/torchtoolkit/)
[![GitHub workflow](https://img.shields.io/github/workflow/status/Natooz/MidiTok/Testing)](https://github.com/Natooz/TorchToolkit/actions)
[![GitHub license](https://img.shields.io/github/license/Natooz/MidiTok.svg)](https://github.com/Natooz/TorchToolkit/blob/main/LICENSE)
Hi 👋, this is a small Python package containing useful functions to use with PyTorch.
It includes [utilities](torchtoolkit/utils.py), [metrics](torchtoolkit/metrics.py) and [sampling](torchtoolkit/sampling.py) methods to use during and after training a model.
Feel free to use it, take the code for your projects, and raise an issue if you have question or a pull request if you want to contribute.
```shell
pip install torchtoolkit
```
It requires Python 3.8 or above.
Simplest example:
```python
from torchtoolkit.metrics import Accuracy
from torch import randint, randn
from pathlib import Path
acc = Accuracy(mode='top_k', top_kp=5)
for _ in range(10):
res = randn((16, 32))
expected = randint(0, 32, (16, ))
acc(res, expected) # saving results
acc.save(Path('path', 'to', 'save', 'file.csv'))
acc.analyze()
```
I built it for my own usage, so you won't find documentation besides the docstring.
Raw data
{
"_id": null,
"home_page": "https://github.com/Natooz/TorchToolkit",
"name": "torchtoolkit",
"maintainer": "",
"docs_url": null,
"requires_python": "",
"maintainer_email": "",
"keywords": "artificial intelligence,deep learning,transformer,nlp",
"author": "Nathan Fradet",
"author_email": "",
"download_url": "https://files.pythonhosted.org/packages/d8/15/d58824cb66c1c34840b83e464db40c960ae58aa16f25f2a45ecf7a98de4e/torchtoolkit-0.0.4.tar.gz",
"platform": null,
"description": "# TorchToolkit\n\n[![PyPI version fury.io](https://badge.fury.io/py/torchtoolkit.svg)](https://pypi.python.org/pypi/torchtoolkit/)\n[![GitHub workflow](https://img.shields.io/github/workflow/status/Natooz/MidiTok/Testing)](https://github.com/Natooz/TorchToolkit/actions)\n[![GitHub license](https://img.shields.io/github/license/Natooz/MidiTok.svg)](https://github.com/Natooz/TorchToolkit/blob/main/LICENSE)\n\nHi \ud83d\udc4b, this is a small Python package containing useful functions to use with PyTorch.\nIt includes [utilities](torchtoolkit/utils.py), [metrics](torchtoolkit/metrics.py) and [sampling](torchtoolkit/sampling.py) methods to use during and after training a model.\n\nFeel free to use it, take the code for your projects, and raise an issue if you have question or a pull request if you want to contribute.\n\n```shell\npip install torchtoolkit\n```\nIt requires Python 3.8 or above.\n\nSimplest example:\n\n```python\nfrom torchtoolkit.metrics import Accuracy\nfrom torch import randint, randn\nfrom pathlib import Path\n\nacc = Accuracy(mode='top_k', top_kp=5)\nfor _ in range(10):\n res = randn((16, 32))\n expected = randint(0, 32, (16, ))\n acc(res, expected) # saving results\nacc.save(Path('path', 'to', 'save', 'file.csv'))\nacc.analyze()\n```\n\nI built it for my own usage, so you won't find documentation besides the docstring.\n",
"bugtrack_url": null,
"license": "MIT",
"summary": "Useful functions to use with PyTorch",
"version": "0.0.4",
"split_keywords": [
"artificial intelligence",
"deep learning",
"transformer",
"nlp"
],
"urls": [
{
"comment_text": "",
"digests": {
"md5": "2decac37404654c0f6aefbe111691c98",
"sha256": "680586004127abd99de48c0525d36349ccf35fc3fb7fa90dd338640ad31650a2"
},
"downloads": -1,
"filename": "torchtoolkit-0.0.4-py3-none-any.whl",
"has_sig": false,
"md5_digest": "2decac37404654c0f6aefbe111691c98",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": null,
"size": 13863,
"upload_time": "2022-12-10T10:33:53",
"upload_time_iso_8601": "2022-12-10T10:33:53.769802Z",
"url": "https://files.pythonhosted.org/packages/94/30/af7e2f7953289a6877e600809f863ddf4649b6e753d3674178f2f7ede504/torchtoolkit-0.0.4-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"md5": "d4ed62dcbee4336c8dac09b79f172d5b",
"sha256": "1aa7351afe41a834e724bbae162eb023f76f8f1cd3569e01a33bb6b5ed73ad24"
},
"downloads": -1,
"filename": "torchtoolkit-0.0.4.tar.gz",
"has_sig": false,
"md5_digest": "d4ed62dcbee4336c8dac09b79f172d5b",
"packagetype": "sdist",
"python_version": "source",
"requires_python": null,
"size": 14911,
"upload_time": "2022-12-10T10:33:55",
"upload_time_iso_8601": "2022-12-10T10:33:55.566871Z",
"url": "https://files.pythonhosted.org/packages/d8/15/d58824cb66c1c34840b83e464db40c960ae58aa16f25f2a45ecf7a98de4e/torchtoolkit-0.0.4.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2022-12-10 10:33:55",
"github": true,
"gitlab": false,
"bitbucket": false,
"github_user": "Natooz",
"github_project": "TorchToolkit",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"requirements": [
{
"name": "torch",
"specs": [
[
">=",
"1.10.0"
]
]
},
{
"name": "numpy",
"specs": [
[
">=",
"1.19.0"
]
]
},
{
"name": "tqdm",
"specs": []
}
],
"lcname": "torchtoolkit"
}