I recommend a used 3090, as that has 24 GB of VRAM and generally can be found for $800ish or less (at least when I last checked, in February). It’s much cheaper than a 4090 and while admittedly more expensive than the inexpensive 24GB Nvidia Tesla card (the P40?) it also has much better performance and CUDA support.
I have dual 3090s so my performance won’t translate directly to what a single GPU would get, but it’s pretty easy to find stats on 3090 performance.
The above post says it has support for Ollama, so I don’t think this is the case… but the instructions in the Readme do make it seem like it’s dependent on OpenAI.
16 GB of RAM, though? Is it even optimized for the Ryzen 9950X3D?
And a 4 TB SSD - not even necessarily NVME?
Doesn’t seem high powered to me.
Are you saying that NAT isn’t effectively a firewall or that a NAT firewall isn’t effectively a firewall?
Is there a way to use symlinks instead? I’d think it would be possible, even with Docker - it would just require the torrent directory to be mounted read-only in the same location in every Docker container that had symlinks to files on it.
Depending on setup this can be true with Jellyfin, too. I have a domain registered, use dynamic DNS, and have Traefik direct a subdomain to my Jellyfin server. My mobile clients are configured using that. My local clients use the local static IP.
If my internet goes down, my mobile clients can’t connect, even on the LAN.
I downgraded Firefox once last year, but after the next major version a couple weeks later, I was able to upgrade again. Never had to downgrade it before that, though.
I forget why, though. I think it was a pretty niche issue.
Do you mean like a FOSS version of https://soundiiz.com/transfer-playlist-and-favorites?
Or at a song/album level, a FOSS version of https://odesli.co/?
Ah, I assumed there were some areas where Firefox had been found lacking relative to Chromium browsers.
For me, the current version of any major browser or fork with consistently applied security updates and capability of running the full version of Ublock Origin is sufficiently secure for my threat model. Given that, and that they all offer the feature set I want, wouldn’t it be reasonable to avoid Chromium browsers because I don’t want to encourage the Chromium monopoly?
That’s only a small fraction of why I use Firefox, to be fair, but suppose for argument’s sake that I don’t care about MV3 extensions, Firefox Containers, etc… Would be it be so wrong for not wanting there to be a Chromium monopoly to be why I chose Firefox or one of its forks?
Use the better, more secure technology
More secure according to what?
Under notes, where you said my name, did you mean “Hedgedoc?”
Sure, but the license is limited to uses that “help you navigate, experience, and interact with online content as you indicate with your use of Firefox.”
Not sure how ads would help with that.
AI? Sure, if an AI solution did those things. But it wouldn’t be them training on your data. This would be them using your data in AI-powered services, whether that be search (especially relevant if Google is mandated to stop paying them to default to Google); automatic categorization of your web browsing to make Containers more streamlined and effective; or even just having a completely opt-in AI assistant chatbot that can access data entered elsewhere in Firefox once you activate it.
Worst case I suspect whatever they add will be things you can simply turn off in settings. Ideally it would be opt-in, of course, or at least prompted-opt-out and disabled until first use.
And there are plenty of things that aren’t ad or AI-related that this could apply to. Heck, this could be part of a step to consolidate licenses for other products - VPN, Pocket, email anonymizers, etc. - and to enable deeper integration of those into Firefox.
local docker hub proxy
Do you mean a Docker container registry? If so, here are a couple options:
Oh 100% agreed - in this instance, it’s clear that OBS has a well maintained package that should be prioritized. But they could keep their repo first and remove OBS (and other known-to-be-well-maintained apps) from it to accomplish that.
They put their repo first on the list.
Right. And are we talking about the list for OBS or of repos in general? I doubt Fedora sets the priority on a package level. And if they don’t, and if there are some other packages in Flathub that are problematic, then it makes sense to prioritize their own repo over them.
That said, if those problematic packages come from other repositories, or if not but there’s another alternative to putting their repo first that would have prevented unofficial builds from showing up first, but wouldn’t have deprioritized official, verified ones like OBS, then it’s a different story. I haven’t maintained a package on Flathub like the original commenter you replied to but I don’t get the impression that that’s the case.
Why did Fedora make their packages take priority? Is it because the priority is otherwise random and if you don’t have a priority set, that leads to the issue they mentioned? Because if so, that sounds like a reasonable action by Fedora and like the real culprit is Flathub.
Clearly they’re cosplaying as a Canonical engineer whose internal explanation and pleas for them to not take this approach fell upon deaf ears /j
If you’re a C developer who doesn’t know Rust, no.
I can’t use signal.
Why? Do you not have a phone number? Is it blocked in your country? Are you legally prohibited from using software with end to end encryption?
Look up “LLM quantization.” The idea is that each parameter is a number; by default they use 16 bits of precision, but if you scale them into smaller sizes, you use less space and have less precision, but you still have the same parameters. There’s not much quality loss going from 16 bits to 8, but it gets more noticeable as you get lower and lower. (That said, there’s are ternary bit models being trained from scratch that use 1.58 bits per parameter and are allegedly just as good as fp16 models of the same parameter count.)
If you’re using a 4-bit quantization, then you need about half that number in VRAM. Q4_K_M is better than Q4, but also a bit larger. Ollama generally defaults to Q4_K_M. If you can handle a higher quantization, Q6_K is generally best. If you can’t quite fit it, Q5_K_M is generally better than any other option, followed by Q5_K_S.
For example, Llama3.3 70B, which has 70.6 billion parameters, has the following sizes for some of its quantizations:
This is why I run a lot of Q4_K_M 70B models on two 3090s.
Generally speaking, there’s not a perceptible quality drop going to Q6_K from 8 bit quantization (though I have heard this is less true with MoE models). Below Q6, there’s a bit of a drop between it and 5 and then 4, but the model’s still decent. Below 4-bit quantizations you can generally get better results from a smaller parameter model at a higher quantization.
TheBloke on Huggingface has a lot of GGUF quantization repos, and most, if not all of them, have a blurb about the different quantization types and which are recommended. When Ollama.com doesn’t have a model I want, I’m generally able to find one there.