A community for everything relating to the GNU/Linux operating system (except the memes!)
Also, check out:
Original icon base courtesy of lewing@isc.tamu.edu and The GIMP
https://discussion.fedoraproject.org/t/fedora-ai-developer-desktop-objective/184941
Seems to me like its mostly just an out of box spin with CUDA and other libs preinstalled so you can spin up pytorch/tensorflow dependent stuff without manually installing all the necessary packages from different sources.
The only unusual thing is the LTS kernel. Thst should probably be out of scope or as a separate preoject not under Fedora's resources.
I think we will go through the same process as we did with electric cars. First "terrible, the work if satan.", "this is no good", "it is worse than the old" .. resist resist. A few will adopt, the tech gets better, more will adopt, then finally almost everyone agrees it is actually quite good, we want this.
Smart use of local LLM is good :) Producing funny pics, using big llm as search engine, or worse as an encyclopaedia, etc is bad.
Two problems:
The technology cannot actually improve. Increasing model scale increases power costs and compute time in return for higher accuracy. The returns from increasing model scale at optimal compute times diminish at a rate which will stop producing better results when about 94% accuracy is reached with literally infinite training data and compute time. There is no evidence that this approach will ever be improved upon in a way that approximates human output such as an AGI.
Even if it were free it creates liability and it's still also free to simply just not to use it and since it isn't necessary for literally any task: the costs will never be justified under any circumstance.
in response to your first point: 24B models from this year are far better than 24B models from 3 years ago. Same model size, similar energy consumption, far better results.
30 years ago there was a lot of doubt that Moore's law could continue the pace for so long. There was no evidence that PCs would continue improving, and yet they did.
So it's really anybody's guess whether or not AI will continue improving. But with the amount of money being poured into it I'm willing to bet it will.
Okay, first of all, Moore's law continuing was caused by a different approach to memory technology which allowed vertical stacking of silicon. Secondly, you literally just gave an example of diminishing returns making steady improvement physically impossible in spite of public expectations, unless a new and different technology is developed, so if anything that's a great argument against AI not for it. Thirdly, Moore's Law is still fucked because right now we're etching silicon with gamma rays and a near-perfect lense and mirror which means we pretty much hit the physical constraints of etching smaller circuits.
Moore's law continuing was due to a ton of different advancements and innovations, not just one. And yes it's slowing down but it still went for 30+ years. If AI continues to improve at this rate for 30 years, hard to imagine how good it could get.
There's been a ton of innovation in the space right now. Like MoE, which was only introduced like 2 years ago and now it's everywhere. It's hard to say what can happen when you have millions of engineers working on something.
Spending fucktons more power to solve problems that don't exist isn't innovation, it's waste.
My argument was that the models would get more powerful over time, even when controlling for size and energy usage. I wasn't talking about waste
My argument is that the shitbots will always get more than 1/20 tokens wrong, have no contextual understanding, and have no morality to speak of such that it will never be improved upon past its current limited and useless form.
Unless a new so-far unheard of technology changes it, AI is currently worthless.
Well if we use vague metrics like those, then anybody can claim anything. "more than 1/20 tokens wrong" what does wrong mean? One view is that a computer program is never wrong, it does exactly what the code says. Another view is that if the AI ends up at a verifiably incorrect answer (for example if you prompted it with a math question), then all the tokens it gave out were wrong. But then humans can be wrong too. Are humans 1/20 neurons wrong on average?
For comparison, Moore's law uses well defined metrics like computations per second. That's what made it a useful concept.
Remember several comments ago when I said it cannot get above 94% with approaching infinite power, compute time, and data? 5% is 1 in 20. SMH.
That number is based on the OpenAI and DeepMind papers on AI Scaling Laws which predicted the performance of every model in the last 6 years.
Those papers are based on dense model architecture. The MoE architecture that I mentioned a few comments ago, does not follow the same laws. Architectural changes could push us past the wall.
Not to mention if we reach 90% accuracy (which was defined in those papers to be AGI level), there's no reason we will need to keep making new models and training them. AGI is good enough. After that we improve inference performance and bring inference cost down.
Hallucinating 1 in 10 fractions of a statement is not AGI by how anybody with half a brain defines it.
A statistical model hallucinating 1 in 1000 isn't even AGI.
AGI is the capability to solve a riddle without being trained on infinite copies of the same riddle, which these machines guessing the next word in a seque have never shown any capacity for.
I know humans that hallucinate more
The Fedora Council has paused the Community Initiatives process, stating that the current framework no longer effectively supports identifying, discussing, and advancing major strategic efforts.
Importantly, already approved Community Initiatives will continue.
Going forward, the Fedora Council says it wants to develop a new mechanism for setting strategic direction in a more open and transparent way, with stronger community involvement earlier in the process.
Context: Fedora’s AI Developer Desktop Was Approved, Then Blocked. Here’s Why. & Heavy Community Backlash Blocks Fedora's AI Developer Desktop Initiative
Seems like it is primarily a technical backlash about having to support another kernel branch for stability with proprietary nvidia drivers. A bit of pushback on AI branding. But also
Part of what made this blow up the way it did was a communications gap. Fabio Valentini of the FESCo noted that he only became aware the proposal was being voted on after stumbling across the council meeting on Matrix accidentally.
According to the Fedora Council, the AI Developer Desktop proposal exposed deeper problems with the current process. The Council said the Community Initiatives framework has failed to provide a good space where new ideas can surface, receive respectful feedback, and gain support when they fit Fedora’s present or future direction.
As a first step, the Fedora Council will immediately pause the Community Initiatives process. This pause affects only the process, not overall Fedora development.
Sure reads like "we don't care if the entire community is against it, we're doing this anyway and there won't be any new features until we finish"
the AI Developer Desktop proposal is closed under this process
Instead, the Fedora Council says that if the AI Desktop work matures, it can follow the existing path toward becoming an official Fedora offering. That would first require a Council ticket for trademark and branding approval, followed by a Fedora Change Proposal for technical review by FESCo. Until then, the proposal remains an independent exploration rather than an official Fedora offering.
I was thinking to test fedora for some time ... now i dont know I want to ...
They literally didn't want the AI spin
why? bevause they gave people an option that dorsnt affect anyone else?
why? bevause they gave people an option that dorsnt affect anyone else?
It always starts like this. I criticized the initial optional and ai features in Firefox. People said "oh its just this one time in the PDF" and so on. What happened? Mozilla added more and more Ai functionality, to a point where they had to integrate a kill switch because so many were bothered by. Also even if you disable all the features, it affects you in some way. Because now they will not integrate an alternative feature without the need of Ai. That means if you disable Ai, you miss out on features otherwise. And it takes lot of resources and development time from the team for stuff you don't care.
Therefore I do not accept the defense of "it is optional, it does not affect you" for these various reasons I just outlined.
Its a sign of whats to come sooner or later. Look at ubuntu. I am using arch right now, loved ubuntu in the past, but that has became a shitshow to say the least. Now I wanted a changed and was looking heavily at fedora, but Im afraid i will settle down and It will end up with something from that project being integrated in the main distro. Idk just my thoughts.
I mean you can try it. I've used it briefly in the past but moved on from it because it's a damn frustratingly opinionated distro with a community that gives the GNOME user/dev base a run for their money.
It works well enough. some things you really got to play with to get working on it. If you're on Nvidia I'd just save yourself the headache and skip it. If you want a distro you can install and just completely forget about and just use flatpaks or whatever, yeah it's good.
If you're on Nvidia I'd just save yourself the headache and skip it
Just to offer a counter balance, my gaming desktop with Nvidia is running Fedora. Other than an weird issue with the mouse cursor - which was easily fixed - it has been working flawlessly.
I'm on AMD, and yes right Noe I want a disteo with flatpacka I can just run and forget about haha :)
so glad I abandoned that shithole of a distro when IBM started making anticonsumer changes.