There is likely a mass of devs that is probably, hmmm, 10x the number of Nix devs from over a year ago, who know exactly what Nix offers and why it is cool, but couldn't be bothered to master it until AI came along and basically "solved the problem".
So I'd expect AI to rapidly accelerate Nix adoption. Hopefully, because it is amazing.
Me too. I’ve known about Nix for a long time but resisted the idea of learning a new language just to manage my system. In retrospect, it would have been worth it, but now I don’t have to.
I use an agent to author my nix config - personally I don't have the determination to have picked Nix up entirely on my own so I find it to be a godsend. I can code review config changes and keep it all in git - also makes it easy to pick up the entire config and deploy it on a different machine.
LLMs can be deterministic too! People just don't bother because the applications of this aren't widely known yet. See https://lukechampine.com/repligraphs
I was just reading Michael Lynch's posts about Sia[0] and I came across this.
Its a very curious project, but don't you end up pinning repligraph usability on model weights? Since you take indeterminism out of the equation, a repligraph's notability is as significant as the producing model's weights, and since there is no dice rolls to be made, the eyeball problem:
> Our blind spots, while not perfectly correlated, have substantial overlap
is entirely replicated. Models that are diffused from one another can have the same blind spots, the same loose statistical reality that exists with humans. This is partially addressed in steering:
> A repligraph proves that a model generated some artifact. It does not prove that the model did a good job, or that the artifact is safe.
but I think the "Peer Review" solution is inadequate, and with some jailbreaking prompts' innocuous looks considered, "the attacker just needs to find one prompt" might be much easier than it appears.
Batching seems to be a huge economic turn off for proprietary model determinism, but I think its entirely viable for consumer models. Trustless evals are brilliant and should've been our reality. Nice project, good luck on your endeavor.
I find myself moving more and more non-build things into nix builds because I don't want to roll my own cache invalidation. Seems like this is yet another reason to do so.
I'd love to become a Nix user one day. I tried to write a rather complex flake one day; it ended up with Nix (nix command) deleting itself: I immediately lost trust with it.
Also Nix first tries to find a derivation in a remote cache and sets a very long timeout, so when you have a limited connection (e. g. a corporate limited network setup), it's painful.
Dude, this is amazing stuff. Farid really is putting out banger after Nix banger haha. This sounds like a mini/less capable Antithesis runtime, but running on your machine.
So I'd expect AI to rapidly accelerate Nix adoption. Hopefully, because it is amazing.
Its a very curious project, but don't you end up pinning repligraph usability on model weights? Since you take indeterminism out of the equation, a repligraph's notability is as significant as the producing model's weights, and since there is no dice rolls to be made, the eyeball problem:
> Our blind spots, while not perfectly correlated, have substantial overlap
is entirely replicated. Models that are diffused from one another can have the same blind spots, the same loose statistical reality that exists with humans. This is partially addressed in steering:
> A repligraph proves that a model generated some artifact. It does not prove that the model did a good job, or that the artifact is safe.
but I think the "Peer Review" solution is inadequate, and with some jailbreaking prompts' innocuous looks considered, "the attacker just needs to find one prompt" might be much easier than it appears.
Batching seems to be a huge economic turn off for proprietary model determinism, but I think its entirely viable for consumer models. Trustless evals are brilliant and should've been our reality. Nice project, good luck on your endeavor.
[0]: https://mtlynch.io/tags/sia/
Curious: how does Rewind deal with randomness?