Flawless1202/FlawlessProjectTemplate

A custom deep learning project template based on PyTorch and PyTorch-Lightning.

FlawlessProjectTemplate

A custom deep learning project template based on PyTorch and PyTorch-Lightning.

Requirements

• torch >= 1.8.0
• pytorch-lightning

Usage

1. Rename the my_package directory and replace the package name my_package in main.py correspondingly.
2. Inherit BaseDataset in my_package/datasets/base.py and implement your Dataset class.
3. Inherit BaseMetrics in my_package/metrics/base_metrics.py and implement your custom Metrics class.
4. Implement your model and put all the components in my_package/models
5. Use registry to organize all the above parts.
6. Follow the example configure file configs/example_cfg.py to make your custom configure file.
7. Train your model by the following command:

   python main.py ./configs/example_cfg.py --train

8. Test your model by the following command:

   python main.py ./configs/example_cfg.py --test

Note

1. Unit tests are really important.
2. Always make your code clean and readable-friendly.