Indeed, this seems like a big step forward, and here’s a link to the model https://github.com/ridgerchu/matmulfreellm

What metadata is collected by third parties is completely tangential to the topic of the submission. However, as I’ve repeatedly tried to explain to you, there is no practical difference between running on bare metal which nobody does nowadays, or running a VPS. At this point it’s quite clear that you’re just trolling, so I’m going to stop here. Bye.

They’re not almost the same thing at all, and your whole position is weird given that the context is social media which is fundamentally content people want to publish publicly.

This has nothing to do with the original topic of discussion or Hollo in particular. You’re now arguing about pros and cons of using a VPS service. I also have no idea why you keep making statements like “not having everything encrypted on a server you don’t own is a massive flaw”. You absolutely can have everything encrypted running a VPS. You don’t understand the subject you’re discussing.

I don’t know what to tell you, but this is how modern internet works. Also, nobody is forcing you to get a server in a jurisdiction where US has access to. Meanwhile, any traffic is encrypted via HTTPS, so the provider can’t actually log it. It sounds like you have a very superficial understanding of the subject you’re debating here.

It’s really up to you how you set up your server and the datastore. This has nothing to do with Hollo. Again, there’s no difference between this and running a Mastodon server that will also need infrastructure like a db to back it.

pretty much nobody runs servers on bare metal nowadays

Yes, it’s a virtual server that you can get from a provider like Digital Ocean. It’s not running on your machine locally, it’s the same thing that the admins of Mastodon instances have to do to run Mastodon servers.

No, it means you run your own VPS to host your personal blog.

This is about making your own personal instance of a microblog that’s ActivityPub enabled. It’s much lighter than running Mastodon that’s mean to be a hosting platform for a lot of users.

indeed

Yeah also true, you’re generally not gonna start doing anything if you know you’re getting interrupted anyways.

yup

I’d say it’s not so much that this tech doesn’t have value, but that it gets hyped up and used for things it really shouldn’t be used for. Specifically, the way models work currently, they’re not suitable for any scenario where you need an exact answer. So, it’s great for stuff like generative art or creative writing, but absolutely terrible for solving math problems or driving cars. Understanding the limitations of the tech is key for applying it in a sensible way.

it’s not actively developed, but it does work

not working due to hallucinations

It’s pretty clear that hallucinations are an issue only for specific use cases. This problem certainly doesn’t make ML useless. For example, I find it’s far faster to use a code oriented model to get an idea of how to solve a problem than going to stack overflow. The output of the model doesn’t need to be perfect, it just needs to get me moving in the right direction.

Furthermore, there is nothing to suggest that the problem of hallucinations is fundamental and can’t be addressed going forward. I’ve linked an example of a research team doing precisely that above.

wasteful in terms of resources

Sure, but so are plenty of other things. And as I’ve illustrated above, there are already drastic improvements happening in this area.

creates problematic behaviors in terms of privacy

Not really a unique problem either.

creates more inequality

Don’t see how that’s the case. In fact, I’d argue the opposite to be true, especially if the technology is open and available to everyone.

and other problems and is thus in most cases (say outside of e.g numerical optimization as already done at e.g DoE, so in the “traditional” sense of AI, not the LLM craze) better be entirely ignored.

There is a lot of hype around this tech, and some of it will die down eventually. However, it would be a mistake to throw the baby out with the bath water.

what I mean is that the argument of inevitability itself is dangerous, often abused.

The argument of inevitability stems from the fact that people have already found many commercial uses for this tech, and there is a ton of money being poured into it. This is unlikely to stop regardless of what your personal opinion on the tech is.

Again, I’m not arguing that open source automatically solves problems, just that since AI is obviously going to continue being developed, it’s better if it’s done in the open.

😄

Open source does actually pave the way towards addressing many of the problems. For example, Petals is a torrent style system for running models which allows regular people to share resources to run models.

Problems like hallucinations and energy consumption aren’t inherent either. These problems are actively being worked on, and people are finding ways to make models more efficient all the time. For example, by using the same techniques Google used to solve Go (MTCS and backprop), Llama8B gets 96.7% on math benchmark GSM8K. That’s better than GPT-4, Claude and Gemini, with 200x fewer parameters. https://arxiv.org/pdf/2406.07394

And here’s an approach being explored for making models more reliable https://www.wired.com/story/game-theory-can-make-ai-more-correct-and-efficient/

The reality is that we can’t put the toothpaste back in the tube now. This tech will be developed one way or the other, and it’s much better if it’s developed in the open.

Couple that with the deteriorating economic situation in US and rampant racism. People are finally starting to see burgerland for the shithole it really is.