README.md

PytorchLightning Template

This repo includes the general template for deep learning training, using PyTorch Lightning library. Use this template to start new deep learning / ML projects.

• Built in requirements
• Examples with CIFAR10
• Badges
• Bibtex

File Organization

`Project name`
└── main.py # The entry to run the training and evaluation.
└── experiment.py # The Pytorch-Lightning training/evaluation module definition.
└── model # The folder to maintain model definition.
    └── __init__.py
    └── model1.py
    └── model2.py
    └── ...
└── utils # The folder to include some useful utility functions and evalutation metrics
    └── __init__.py
    └── utils.py
    └── metrics.py
└── data # The folder to maintain data module.
    └── __init__.py
    └── dataset1.py
    └── dataset2.py
    └── ...
└── conf # The folder to maintain the configuration files.
    └── setup.yaml
└── exp # The folder to maintain the logging files and model checkpointing.
    └── exp1
      └── log
      └── checkpoint

Lightning Philosophy

This template structures the deep learning project into 4 parts:

• Model specification
• Experiment settings
• Training settings
• Logger settings

These 4 parts are to be as independent as possible so that the code is more readable and flexible.

DELETE EVERYTHING ABOVE FOR YOUR PROJECT

---

Your Project Name

Description

What it does

How to run

First, install dependencies

# clone project   
git clone https://github.com/YourGithubName/pl-template

# install project   
cd pl-template
pip install -r requirements.txt

Next, navigate to any file and run it.

# module folder

# run module (example: mnist as your main contribution)   
python run.py --config config/config.yml

Config file

model_params:
  name: "<name of classification model>"
    .
    .
    .

exp_params:
  dataset: "<name of dataset>"
    .
    .
    .

trainer_params:
  gpus: 1
    .         # Other parameters required by the model
    .
    .

---

Results

...

Citation

@article{YourName,
  title={Your Title},
  author={Your team},
  journal={Location},
  year={Year}
}