Can Replit be used to spin up and run a GPU cluster for executing ML training using e.g. Pytorch?

Am new here.
Trying to get my head around what I can hope to do on this platform. I train somewhat heavy/compute intensive ML models for financial applications. Can I use Replit for distributed GPU -based ML training and inference?


Hi @YorkvilleDS thanks for your great questions and welcome to the community!

I’ve reached out to Replit Engineering team with your query. I’ll update as soon as I can.

Any updates on this? Thanks!

Hi @YorkvilleDS and @doctorjuice1
There is something in the pipeline regarding this which should be available later in 2023. No firm release date yet so check back in the summer!


We have an updated data science template here:

And an older pytorch starter here:

1 Like