News for hackers. Stuff that matters.
Re:Mistral NeMo by causal on Thursday July 18, 2024 @11:44AM (#40996768)
Yeah it will be interesting to see if we ever settle on standard sizes here. My preference would be:

- 3B for CPU inference or running on edge devices.

- 20-30B for maximizing single consumer GPU potential.

- 70B+ for those who can afford it.

7-9B never felt like an ideal size.

There are no comments on this comment yet.