Deep Learning Project Template for Lightning and Hydra

Setup

If you already have a python environment and have poetry installed, you can skip ahead to the Installing Dependencies section. Otherwise, it is recommended to first go through the Virtual Environment section to setup an environment with all the required tools.

Virtual Environment

If you don't operate inside a virtual environment, or only have access to an incompatible python version (<3.8), it is recommended you create a virtual environment using conda:

conda env create -f environment.yml
conda activate deep-learning-template

Installing Dependencies

Once you have a python interpreter and poetry setup, simply install the project's dependencies:

poetry install

Principal Command

python run.py task=TASK model=MODEL datamodule=DATAMODULE

• task str: Task configuration to use for the run.
• model str: Model configuration to use for the run.
• datamodule str: Datamodule configuration to use for the run.

Once you've provided the required parameters described above, you can see detailed help on how to customize any other aspect of the configuration by running:

python run.py task=TASK model=MODEL datamodule=DATAMODULE --help

Basic Example Command

python run.py task=pixel_wise_autoencoding model=simple_ae datamodule=mnist

How to Contribute

If you want to contribute to the project, then you have to install the pre-commit hooks, on top of the basic setup for using the project, detailed above. The pre-commit hooks are there to ensure that any code committed to the repository meets the project's format and quality standards.

pre-commit install

Be aware that for some pre-commit hooks to run correctly, poetry MUST be installed globally on the machine you'll be committing from. If you don't already have it installed globally, the instructions to do so are available here.