We want our first release to include the implementation of an unsupervised image-to-image translation model such as TUNIT. We want to include all of the following features/enhancements.

[x] Seperate package for TUNIT's implementation with all of its configurations, utils and train.py.
[x] Setup tox locally and run github actions with it for automated testing.
[x] Have a decent pull request and issues template, and a milestone.
[ ] Have a customized Logger.
[ ] The project' source code should work on any device. This includes removing paths specific to any device (e.g. dataset paths) and making the code OS agnostic using packages like pathlib.