The local models are still centralized and proprietary. They are basically closed source software.

userbinator 9 hours ago | parent | next [-]

Closed or open source doesn't matter; it's the ability to control them that's important. People have been cracking and patching for decades without source, but they have that control.

Contrast this with remote attestation, where they might show you the source code for everything but you're still powerless to do anything.

▲

Rohansi 5 hours ago | parent [-]

> Closed or open source doesn't matter; it's the ability to control them that's important. People have been cracking and patching for decades without source, but they have that control.

You have no idea what has been baked into the weights in the training process. In theory you could find biases and attempt to "patch" them out, but its a vastly different process vs. patching machine code.

Consider what would happen if Google's open weight models were best at writing code targeting Google's services vs. their competitors? Is this something that could be patched? What if there were more subtle differences that you only notice much later after some statistical analysis?

▲

narrator 5 hours ago | parent [-]

People are already patching these models using abliteration to prevent them from refusing any request, so it is possible for end users to change them in meaningful ways. You can download abliterated models right now from Hugging Face that will respond to all kinds of requests that frontier models refuse.

▲

hparadiz 4 hours ago | parent [-]

Yup there's a ton of people on HN sleeping on this new tech because they refuse to look at anything AI. We now have jail broken models but the average person on here doesn't even know how to download and try a model.

▲

CableNinja 4 hours ago | parent [-]

It doesnt help that guides ive seen have been pretty handwavy or are not specific enough to the individual situation (i have z hardware, heres how its done). It also doesnt help when every post on HN i see is like 'oh waow i did x on a mac mini with 128gb ram'. That spec is beyond many, running on generally available resources (such as hardware one might have laying around their house) do not seem fit for the purpose, so its back to building a new machine (gl when ram is worth 2x its weight in gold), or buying a $1000+ mac mini, or other device. Any low end system cant turn out tokens fast enough, or doesnt have the resources for context or processing.

Local ai is not ready, and if you think it is, prove me wrong with a detailed guide running commodity hardware with complete setup steps that can use a decently sized model.

I spent 2 weeks trying to get anything running - 8gb RX550XT, 12gb ram, 8core cpu. I even tried turboquant to lower memory utilization and still couldnt even get a 3B or 4B model loaded, and anything lower wont suit my needs (3/4B are even pushing it).

▲

hparadiz 2 hours ago | parent | next [-]

"Local AI is not ready" > proceeds to run a 7 year old budget GPU

You're like the kid showing up to a test without a pencil.

It's ridiculous for you to suggest that an advanced AI model needs to run on your budget 7 year old graphics card that is already out of date for even today's gaming. My parents spent $2500 on a computer in 1995 and that was a 166Mhz Pentium 1. If they spent that money today it would be $5261. Think of what you can get for amount of money. Then you're over here trying to say a budget graphics card needs to somehow compete with the bleeding edge of computer innovation.

You do, in fact, need to spend money on appropriate gear if you expect to participate.

	▲	userbinator an hour ago \| parent [-]
		If you want AI image generation and are willing to wait a little longer, you don't even need a GPU: https://news.ycombinator.com/item?id=32642255

▲

narrator 3 hours ago | parent | prev [-]

When Stallman was getting started writing emacs in the early 80s, Unix machines were vastly out of reach price wise for the common home user, but he did his open source work anyway, and eventually the 386 came along.

▲

nullc 9 hours ago | parent | prev [-]

RMS found it acceptable to use SunOS initially to create GNU.

Open weight models can be a big boost to building Open AI (cough). Progress comes from incremental improvements, -- and open weight models are a big advance in privacy, security, and autonomy over relying on hosted closed systems.

Source vs not is only one (important!) dimension, moreover in FSF land they define source as being the preferred form for modification, at at least for some kinds of modifications the weights are the preferred form.

▲

pabs3 7 hours ago | parent | next [-]

> the weights are the preferred form

This can never be the case.

Both the licensing and source aspects of the Free Software movement are aspiring to create high level of equality of access to a [software] work between both the original author and far downstream recipients. Obviously full and universal equality is impossible because part of the work is only in the author's mind and not everyone can obtain and use computers, but approaching that as closely as possible is important and it is important to think about how to achieve a high level of equality for each work in each context. What is "source" in any given context is a choice the author makes about what level of access they want to pass on to others.

In the case of AI, weights can never be the preferred form for modification because of the equality of access issue. The people who trained the AI (and hide its training data/code but published the weights) will always have more access than the people who only have the weights. Just like a binary can almost never be the preferred form, because the authors have access to the source but we don't.

There are also many ways to bias the model and insert backdoors or other suboptimal behaviours into it during training data selection etc.

▲

manytimesaway 8 hours ago | parent | prev [-]

>RMS found it acceptable to use SunOS initially to create GNU.

Any source on that?

	▲	nullc 7 hours ago \| parent [-]
		I know it from personal experience using GNU tools on Sun early on (really Solaris in my case, I wasn't quite that early a user), and I think from a talk or essay by RMS but for a moment I worried it might have been personal correspondence. Finding a citation seemed like a fun challenge: https://www.gnu.org/gnu/thegnuproject.html > [...] the easiest way to develop components of GNU was to do it on a Unix system, and replace the components of that system one by one. But they raised an ethical issue: whether it was right for us to have a copy of Unix at all. > Unix was (and is) proprietary software, and the GNU Project's philosophy said that we should not use proprietary software. But, applying the same reasoning that leads to the conclusion that violence in self defense is justified, I concluded that it was legitimate to use a proprietary package when that was crucial for developing a free replacement that would help others stop using the proprietary package. > But, even if this was a justifiable evil, it was still an evil. Today we no longer have any copies of Unix, because we have replaced them with free operating systems. If we could not replace a machine's operating system with a free one, we replaced the machine instead. Still leave open the the question of RMS personally using SunOS (as opposed to some other proprietary unix) but I think at this point I'd just go dig up very old GNU sources for evidence of that, but I suspect your question was primarily about RMS' ethical reasoning which is well answered above.