Building a pipeline for training on multiple GPUs with dstack
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