undefined | Better HN

0 pointsultra_nick4y ago0 comments

Everyone would need a next gen GPU to run it and someone would have to drop $500k on a gpu rack to train it.

A decade for general accessibility sounds right.

0 comments

1 comments · 1 top-level

No the ML inference can be done on the cloud or use tiny models and embed them in IDE code, possible today.

j / k navigate · click thread line to collapse