Elvith Ma'for

Yeah, but if I understand that correctly, that’s just for the app itself the LLM is very likely still a proprietary one (ChatGPT, Grok,…)

non-commitment like “I can’t do that” or “that’s against policy” or “that’s not my dept”

Ok, I’m not a native English speaker but… I have the feeling that they don’t know what non-commitment means. Unless it’s commitment to fuck the customer, but then, why bother to offer a call center?

Time for a classic: The Case of the 500-Mile Email

Edit: The site seems to be overloaded, but it’s also on Archive.org: https://web.archive.org/web/20260220060645/https://www.ibiblio.org/harris/500milemail.html

Other than Friendica, Mastodon, Matrix, PeerTube and PieFed, what’s worth running

Maybe Nextcloud (not only for storage, but also calendar, video conferences, office, when combined with Collabora,…)? Also Immich (basically Google Photos) comes to mind. Your own instance of SearXNG.

Any kind of ToDo-list, Kanban board, …?

A ticketing system?

A Wiki to host your documentation? Note, that you may want to access it if the server fails, so…

Some stack of components around Grafana or such to visualize some data? Since you mentioned Hetzner, I’m guessing you’re from Germany. You could build a small container, ingest the gas prices that the gas station are required to publish and build a dashboard for the gas prices in your area? (Hint, here’s an API licensed under Creative Commons - https://creativecommons.tankerkoenig.de/ )

Delays - if you use the internet and request an answer from an LLM, you won’t notice if it’s 300-500ms slower than usual. But if you deploy and run a software stack, a delay of 5ms betweenntze app and the database can make the difference between a usable application and an inperfomant one.

See, when mommy fire and daddy fire are left alone long enough and it warm and cozy…

Real models are trained on copyright-, license- and ToS violations! Those cheap bastards found a way to only train with ToS violations (of a single source, mind ya!). That’s not real AI!

^/s

How do I do that on Mastodon, Lemmy, Piefed,…?

^/s

No, but I just had 14 verification cans burp urks

The ankles? Looks nervously in the rough direction of the middle east You mean her face?

Was it an official act, though?

lp0 on fire

My journey:

Had some form of Linux for a long time. Either in a VM (Oracle Virtual Box, then switched to S HyperV for compatibility reasons as I had Windows Pro anyways) or sometimes as dual boot.

Then came WSL which eased some things and complicated others. What this makes really easy is to start and play around with docker containers on your PC.

Then I experimented with Linux in a VM and put docker and other software there to practice.

Up until here, there were no costs involved (besides having Windows Pro, but depending on where you get your windows key, there’s not a real difference between pro and home anyways…).

After that I got my own VPS. As much as I don’t like AWS, Google Cloud (GCP), Azure and such, they usually offer a very small VPS for free and these can be a good point to start. If you want to really go and host things, it can be beneficial to look for a hoster that isn’t one of the big 3 cloud providers and pay for a VPS there.

For hosting at home: You could start with a raspberry pi, but looking at current prices, you usually get more flexibility and bang for the buck by buying a refurbished mini PC or repurposing an old notebook/PC. You can just put Yunohost or Proxmox on it and get going.

honks apologetically

If it fits, it sits

Looks to me like they’re essentially redirecting the request from the normal api to do age checks to their own api, and just saying “Sure, they’re an adult” to discord (since that is all the “proper” api tells them).

Wait… Those amateurs [at discord and the age check company] didn’t even think of signing the check in any way and then verifying the data they get send back? That’s not even hard to implement?!

Government? It’s just three billionaires in a trench coat!

Im using SchildiChat on Android