D0miH / PyTorch-Lightning-Template

A template to get started with PyTorch Lightning

PyTorch-Lightning-Template

A PyTorch-Lightning template that supports yapf autoformatting and includes a docker container to run the code.

To see examples on how to use the configs, have a look at the config readme

Setup

Setup Using Docker

Build the Docker Image

To build the Docker image run:

./docker_build.sh -n <image-name> -w <wandb-key>

Start Docker Container

To start the Docker container run the following command:

./docker_run.sh -n <container_name> -i <image_name> -d <comma_separated_device_ids>
# e.g. ./docker_run.sh -n pytorch_lightning_template -i pytorch_lightning_template -d "0,1"

or

docker run --rm --name <container-name> --gpus '"device=<device-ids>"' -v $(pwd):/workspace  -it <image-name> bash

That's it!🎊️
Now you are setup and ready to develop using Docker!

Local Setup

If you don't want to use Docker and instead want to set up the project locally to use yapf and pre-commit you have to first install PyTorch:

conda install pytorch torchvision torchaudio pytorch-cuda=11.7 -c pytorch -c nvidia

Then you have to install all other dependencies:

pip install -r requirements.txt

Finally, you have to set up 'pre-commit` to run yapf each time before committing:

pre-commit install

That's it!🎊️ Now you are setup and ready to go!