# core-pytorch-utils

**Repository Path**: mjhapp/core-pytorch-utils

## Basic Information

- **Project Name**: core-pytorch-utils
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: MIT
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2024-11-02
- **Last Updated**: 2024-11-02

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

<img src="docs/_static/logo.png" alt="drawing" width="200"/>

[![Docs](https://readthedocs.org/projects/core-pytorch-utils/badge/?version=latest)](https://core-pytorch-utils.readthedocs.io/en/latest/?badge=latest)
[![GithubAction](https://github.com/serend1p1ty/core-pytorch-utils/actions/workflows/ci.yml/badge.svg)](https://github.com/serend1p1ty/core-pytorch-utils/actions)
[![Codecov](https://codecov.io/gh/serend1p1ty/core-pytorch-utils/branch/main/graph/badge.svg)](https://codecov.io/gh/serend1p1ty/core-pytorch-utils)
[![License](https://img.shields.io/github/license/serend1p1ty/core-pytorch-utils.svg)](https://github.com/serend1p1ty/core-pytorch-utils/blob/main/LICENSE)

# Core PyTorch Utils (CPU) *[Completed :tada:]*

This package is a light-weight core library that provides the most common and essential functionalities shared in various deep learning tasks:

- `Trainer`: does tedious training logic for you.
- `LRWarmupScheduler`: wraps all standard PyTorch LR scheduler to support warmup.
- `ConfigArgumentParser`: provides an argument parser that supports loading a YAML configuration file.
- ......

You can find a brief Chinese introduction at [zhihu](https://zhuanlan.zhihu.com/p/449181811).

## Updates

[2023/10/29]: Now CPU supports iteration-based training, checkpointing and evaluation!

## Installation

From PyPI.

```
pip install core-pytorch-utils
```

Or from source.

```
git clone https://github.com/serend1p1ty/core-pytorch-utils.git
cd core-pytorch-utils
pip install -r requirements.txt
pip install -v -e .
```

## Getting Started

In [examples/](https://github.com/serend1p1ty/core-pytorch-utils/tree/main/examples) folder, we show how to use our Trainer to train a CNN on MINIST.

It is **strongly** recommended that you run this code before using the CPU library.

## Advanced

Learn more from our [documentaion](https://core-pytorch-utils.readthedocs.io/en/latest/).

## Contributing

Pull request is welcomed! Before submitting a PR, **DO NOT** forget to run `./dev/linter.sh` that provides syntax checking and code style optimation.

## License

CPU is released under the [MIT License](LICENSE).

## Acknowledgments

We refered [mmcv](https://github.com/open-mmlab/mmcv.git), [detectron2](https://github.com/facebookresearch/detectron2.git) and [pytorch-image-models](https://github.com/rwightman/pytorch-image-models) when develping CPU.