undefined | Better HN

0 pointskurtoid3y ago0 comments

Silly question: how does OpenAI host/serve it?

0 comments

I think on professional hardware you can get 80G of memory per GPU and they can likely do memory pooling.

j / k navigate · click thread line to collapse

0 pointskurtoid3y ago0 comments

Silly question: how does OpenAI host/serve it?

I think on professional hardware you can get 80G of memory per GPU and they can likely do memory pooling.

j / k navigate · click thread line to collapse