This is a repository for handing in all local code and exercise done for the course 02476 Machine Learning Operations at DTU.

• day_1 contains a lot of introductory notebooks for MNIST and PyTorch
• day_2_mnist contains a CNN implemented for the corrupted MNIST dataset. This folder additionally contains solution for other exercises that are related to the corrupted MNIST model e.g. continous integration, unit tests and more.
• day_3 contains the reproducability exercises on config files while the docker implementation was done for corrupted MNIST.
• day_4 contains wandb exercises and profiling.
• day_5 contains the scalable application exercises.
• day_6 contains deployment exercises.
• The monitoring exercises are found in GCP.

Some modules were not implemented on the corrupted MNIST example presented in this repository. An example could be the PyTorch lightning module which was implemented directly for the course project.