This repo contains the code to run a training pipeline on multiple GPUs.

The train.py contains all the machine learning components.

In order to automate the workflow, I use dstack.

In order to run the workflow, all you need to do is execute the following command in your terminal:

dstack run train-mnist-multi-gpu