growthwtf

I haven't finished this but I like the metaphor of Isekai at the start of the article. Nice parallel to a potential model of llm gestalt.

Sure, in production envs I have seen humans being used in 3 places: 1. the pose data calibration 2. cleaning up covariances (reducing blobbiness) 3. adding metadata for app usage But, to your point, hard to say which of these or any apply without more info. I would be very very impressed if there were no humans and it's 'just' a training time issue though!

Without knowing the specifics of their pipeline I would also hesitate to comment further.

You could be correct, but it would be a real indictment of their rendering farm I think.

Rendering takes a few hours means humans are building it at least partially.

You all need to stop being so pessimistic. This is a great idea.

Want PBS to stick around? Make it so anybody who's sticking on chat GPT gets great answers from PBS and every time ChatGPT scrapes it, PBS gets money.

Is it extremely difficult? Obviously. Will it work? Probably not, very few things do. Is it a great thing that some folks are doing it and trying to make it work so that we can have a functional media ecosystem in a post-social-media age? Absolutely.

Lattner's comment aside (which I'm fanboying a little bit at), I do tend to agree with your pessimism/realism for what it's worth. It's gonna be a long long time before that whole mess you're describing is sorted out, but I'm confident that over the next decade we will do it. There's just too much money to be made by fixing it at this point.

I don't think it's gonna happen instantly, but it will happen, and Mojo/Modular are really the only language platform I see taking a coherent approach to it right now.

A fun project for somebody who has more time than myself would be to see if they can get it working with the new Mojo stuff from yesterday for Apple. I don't know if the functionality would be fully baked out enough yet to actually do the port successfully, but it would be an interesting try.

Nah. There's huge alpha here, as one might say. I feel like this comment could age even more poorly than the infamous dropbox comment.

Even with Jax, PyTorch, HF Transformers, whatever you want to throw at it--the dx for cross-platform gpu programming that are compatible with large language models requirements specifically is extremely bad.

I think this may end up be the most important thing that Lattner has worked on in his life (And yes, I am aware of his other projects!)

I'm not the original commentator, that makes a lot of sense! I had assumed there was a huge overlap, personally.

Good thing they didn't nuke the data centers after all!

It's life changing for most people in the SF bay area too!

I think it's a perfectly valid take coming from some intersection of an engineering mindset and FOSS culture. And, the comparison you bring up is a bit of a category error.

We know how James Webb works and it's developed by an international consortium of researchers. One of our most trusted international institutions, and very verifiable.

We do not know how Genie works, it is unverifiable to non-Google researchers, and there are not enough technical details to move much external teams forward. Worst case, this page could be a total fabrication intended to derail competition by lying about what Google is _actually_ spending their time on.

We really don't know.

I don't say this to defend the other comment and say you're wrong, because I empathize with both points. But I do think that treating Google with total credulity would be a mistake, and the James Webb comparison is a disservice to the JW team.

I think it's an interesting correspondence—some general design principles about creating good auditory user interface somewhere in here. I would be interested if someone smarter than me can tell me what that principle is.

Me either. I have heard stories of it happening, but never personally seen one live. It's really a tooling issue. I think the causal story is super important and will only become more so in the future, but it would be basically impossible to implement and maintain longer-term with today's software.

Kind of like flow-based programming. I don't think there are any fundamental reason why it can't work, it just hasn't yet.