MLOps.community  cover image

CPU vs GPU // Ronen Dar & Gijsbert Janssen van Doorn // MLOps Coffee Sessions #99

MLOps.community

00:00

Machine Learning Start Ups Coming Out of Israel

You never scale down to zero in that case, because it's always there. Another approach is to just deploy one model, or a small number of models, in one selver and then, right size, they're compute resourcessed. You know, use fractions of a gpu instead of a whole gPU. So still, you get, you have one gp u that you're paying for, it's up and running, and then you youl o many mode data ready to cive requests using that p that makes sense.

Transcript
Play full episode

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!
App store bannerPlay store banner
Get the app