Corey Knowles

There's nano, super, and ultra.

And I liked how you explained the thought process behind which one goes where.

I guess we kind of already know

like, what the North Star was that governed those splits was, like, the actual GPU that you would run it on.

But what would you think is, like, what job should each tier own so teams don't just default to the biggest one?

Like...

If you're going in the weeds.

Maybe it's different for everyone.

So they can run like as many as possible, right?

I have a question kind of related to that, which is like when you work with like these companies, are they hiring ML engineers to do this stuff for them or are they trying to learn it on their own?

I think that's kind of interesting.

Right.

So where do you feel like the Nemotron family really shines?

And then I guess the flip side of that is where does it have room to grow in your opinion?

I like that.

Yeah, it's very fitting for speech.

I do too.

Yeah, and what's great about Parakeet and Canary is I think both of those, there's versions that you can run locally too, right?

That's right.

Yeah, that's so cool.