

Yes. But compared to the number of people who currently use LLMs, not that many have hardware capable of doing that.


Yes. But compared to the number of people who currently use LLMs, not that many have hardware capable of doing that.


What I’m saying is that not all local LLMs are more power efficient, and that we are currently massively over-using the technology for things it isn’t all that useful for.


I suppose, but that’s a very different class of model. I think a more important question might be whether people actually need LLMs at all.


No? If everyone who uses LLMs globally switched over to a local LLM (after buying the necessary hardware), that’d still be a crazy amount of energy usage, just less centralised.


That might be the case. And you might in fact be perfectly well intentioned (though, as you might have found this community isn’t, on average, very into closed source software).
On the project site, this post, your replies, I have read nothing that sounded like genuinely you. It all reads like marketing. Or, more precisely, as if you have a LLM write/rewrite your responses. This, to me, makes them feel incredibly disingenuous. You might just sound like that naturally. In that case, I’m sorry.
If you want to win over this community, a good avenue would probably be open sourcing your application and arranging for some form of donation.
Edit: Also, you didn’t state that it’s closed source in your first sentence?? Well, you did in the first sentence of the reply to the comment that called out there not being source code on the repo, but the cat was out of the bag at that time.


An unkind observer might suspect the goal to be having the aesthetic of an open source or at least source available project without actually being one.


I don’t think this fully applies to Ferrari in the way it might apply to mass market cars.


It’s not really renaming. It’s just that the big distros have different naming conventions, and documentation often uses the names used either by Fedora/Redhat or Debian/Ubuntu.


I use tumbleweed. Mostly just works. Sometimes, you have to go look for packages that are named differently than in other distros.


I would argue that those sanctions, if general, shouldn’t include necessities like food or medicine. Also, I feel that sanctions against specific figures in the government might be relatively effective (many countries are doing this already). Make it difficult to enjoy the fortune they stole from their people.
I also don’t like the idea of sanctions that aren’t based on specific/ongoing actions, but the existence of a certain government/regime, though I see where you are coming from.
I would also still argue that the Russian people chose that government, back when they did choose, because they didn’t see many other options, and keeping them in perpetual destitution long term isn’t likely to change that.


Demonising a whole people like this is in itself wrong.
What is your supposed solution, then? How does one deal with tens of millions of people who are supposedly evil and wrong?


Where they, though? People drafted and sent to the front generally have few options. I get that people dying is what happens in war but casting all individual soldiers of one side as “bad”, of the other as “good” just serves to dehumanise people and excuse atrocities.
This was a legitimate military target, but I don’t think anyone should cheer about a hundred dead people.


Huh, Authentik was what I used before Kanidm. Wasn’t anything wrong with it per se, but there where a lot of moving parts and complexity rhat didn’t really serve a purpose for me.
I thought about kubernetes or proxmox, but I don’t really see any reason to. All my containers are controlled via podman quadlets, and either run on a single machine locally, or on a VPS.


If you can autofill passwords without authenticating in some way, they are probably either stored in plaintext, or encrypted with a key that is stored in plaintext. Cause, like, how is it supposed to magically encrypt it.


On how you want to slice up the hardware - I feel like there isn’t one right answer, and I’d do whatever feels most comfortable to admin for you. I feel like for homelab workloads, any half-reasonable setup should work fine, just make sure you have good backups.
On SSO - I have never tried Authelia, but am personally very enamoured with Kanidm. It’s very lightweight, and has pretty good default settings.
On reverse proxy - I personally use Caddy, but Traefik is good too, and can do more stuff out of the box. I just mount the certs I need readonly in the container of the service that needs them. Clunky, but works well enough for me.


I think you might have misread my comment. I was saying the feature only just gets to the same performance, while using it means having to deal with Windows.
If you mean the incompatibility of some anticheat titles - I think kernel level anticheat is an unmeritted level of system access, so I wouldn’t use them either way.


limitations of linux
I’d look at it the other way around - why go to a platform that has marginally worse performance, while also making you deal with the limitations of Windows.


No TQ Christians?
Yes and no. You could, in theory, put it behind a reverse proxy, but you shouldn’t without further security measures.
You could put your server in a tailnet (tailscale, there are FOSS alternatives, but those are more difficult to set up), which is extremely, easy to configure, and allows every device in it (smartphones, tablets, computers…) to access your media server.
I have two live hdds resting on high density foam on top of my server. Will be three soon. I really should get a bigger case…