Getting Up and Running with PyTorch on Amazon Cloud

This is a succint tutorial aimed at helping you set up an AWS GPU instance so that you can train and test your PyTorch models in the cloud. If you don't own a GPU like me, this can be a great way of drastically reducing the training time of your models, so while your instance is furiously crunching numbers in some faraway Amazon server, you can peacefully experiment with and prototype new architectures from the comfort of a Starbucks couch. The cool part is that if you're a high school or college student, you can sign up for a Github Developer pack which will get you $150 worth of free AWS credits. That's around 167 hours or 7 days of compute time1, an amply sufficient amount for those fun weekend side projects and experiments. As usual, any code or script that appears on this page can be downloaded from my Blog Repository.