All of the AI uses I've listed have been around for almost a decade or more and are the only computational solutions to those problems. If you've ever used speech to text that wasn't a speak-n-spell you were using a very basic AI model. If you ever scanned a document and had the text be recognized, that's an AI model.
The catch here is I'm not talking about chatgpt or anything trying be very "general". These are all highly specialized ai models that serve a very specific function.
The firefox AI sidebar embeds an external open-webui. It doesn't roll its own ui for chat. Everything with AI is done in the quickest laziest way.
What exactly isn't very open about open-webui or ollama? Are there some binary blobs or weird copyright licensing? What alternatives are you suggesting?
Basically everything its used for that isn't being shoved in your face 24/7.
Lots of these existed before the AI hype to the point they're taken for granted, but they are as much AI an LLM or image generator. All the consumer level AI services range from annoying to dangerous.
If I can pick my own API (including local) and sampling parameters
You can do this now:
Hardest part is hosting open-webui because AFAIK it only ships as a docker image.
Edit: s/openai/open-webui
I think the overtightened heatsink theory is correct. I got it for an LGA1155 motherboard and it needed an adaptor for AM4 which has no safety springs so you can just keep tightening it with no indication of what's too tight.
There's also some sort of dirt on the center left edge that I cleaned out after taking this. I just used isopropyl and a softened (used) toothbrush. After doing that I got a channel of memory back. There's three sticks in the pic because I was trying to determine which channel was bad.
Update: Wow cleaning that bit of gunk really made a difference. I'm back to using the original Ryzen 5700. Really weird that it wasn't an issue in the old build.
Its like that but way too sensitive. I've deliberately been gentle and it will still fail especially if I set it upright with fans on the heatsink.
To build on this, it would help to install some sort of system monitoring to check temps, fanspeed, system usage and have those constantly going so OP can check for any red flags during a freeze.
Idk about that. In my case I believe my CPU was defective from the start and I lived with it because I always assumed it was my OS in some way.
If your CPU has seven years of not randomly freezing and its just doing that now then I wouldn't suspect the CPU.
However, unless you find some clues from journalctl -xeb1 or dmesg I would assume its faulty hardware somewhere.
Last time for me it was a bad CPU. Lived with it until I upgraded my CPU and recycled the old one into a new build. Then that one was having the same issue.
This is why Mamdani is so important. Average progressives are going to finally see what socialists have been saying for decades. If you want to move left on anything economical. If you just want to raise the minimum wage $1. The democratic establishment will treat you as the enemy. They will come at you harder than they do the GOP because they can and because you are a threat to them personally.
I don't recommend it for first time users. Going through the install process is basically the tutorial level for Arch. The arch install scripts are for shortcutting an install process you're already familiar with. If you want arch with a no pain installer use something like EndeavourOS.
For example, I've used Arch for years, and tried experiment with BTRFS using the archinstall script. I deeply regretted it because it didn't configure SWAP correctly so my laptop couldn't hibernate. I had to rip out zram and put in a swapfile.
You must have an idea of what its going to do before using it. In the case of you're bloated KDE install, it installed the full kde package group, which almost no one wants. You could use the archinstall script to install everything but the graphical environment and then do that yourself, post install.
Looks like a Jolly Bee got converted to a McDonalds.