#thought experiment: llama chat playground

1 messages · Page 1 of 1 (latest)

fair anchor
#

🧵

#

this 👆is hosted on replicate.com . but it’s open source. Unclear how valuable the hosted version is - is it mostly convenience? Access to more powerful GPU than your local machine? 🤷🏻‍♂️

#

Do we want announcements like this to link to a Dagger environment in the future?

jovial girder
#

I guess if our goal is to be platform agnostic, then if we do a good enough job (we're not fully there yet, though on the way) it should be possible to run the exact same environment locally and in some hosted infrastructure. You should only be limited by physical reality (i.e. we can't magically download a GPU for you).

That being said, while I have no idea, I wouldn't be surprised if there are "gpus-in-cloud" available over an API that could be connected to as part of an environment's implementation. That's probably as close to magic as we could get.

#

Either way, +100 to the general idea of announcements like this linking to a dagger environment. Specifically, a one liner like "dagger shell ..." or "dagger artifact --export", etc. whenever possible. What were doing is the absolute perfect fit for something like that and could be viral in terms of ecosystem growth