I have several of these set up. Mostly for opening windows to reduce AC use. If the temp outside is dropping and is within two degrees of interior temp, then it’s time to open the windows. And the inverse during autumn to close them. I could have saved you some time probably.

I’m not either but I still do it.

That is exactly it, yes. And I am thinking about reaching out to them about it and their blocklist. But after thinking about it, anyone with $16 in hosting can start an AI hacking instance. It’s just going to get worse.

This one was 66.187.5.19. I tried to find a way to report their behavior on hostodo (the service they’re using) but they don’t seem to have a report option. I’m guessing because they don’t want to prevent this kind of thing.

Mom can we watch ‘GRUB Not Found’ again?

And that’s really why they’re hoarding them.

That is a solid enough reason and one I wasn’t aware of. Exactly the kind of thing I was hoping to find out, thank you!

I’ve been eyeing Gitea. Do you have any reason for using Forgejo that I might overlook Gitea in favor of it?

Same boat, the exact same refurb 10tb drives I’ve got in my system are 2.5x my last purchase. But I did get a lot of fast RAM before the prices went nuts.

Don’t forget to check out Youtarr. I think it hits all the targets. You can do channels, playlists or individual videos. It has an option for a proxy URL to use to download which I believe can be set up with a self-hosted VPN. And it formats and sends things directly to Jellyfin.

Audiobookshelf for audiobooks and Navidrome for music.

This is some Men in Black shit I think

We don’t even know if it’s currently being produced at greater numbers (as far as I’m aware). The only public info I’ve seen is that the data centers that haven’t been built will need all of the produced RAM based on handshake deals (not contracts). The RAM makers themselves are doubling down on the bubble by investing into the AI generators as well as increasing costs to insane levels in the process which surely reduces the amount of items sold.

China is largely focusing on the only good parts of AI. The scientific data that can be gained from specialized processing. They’re not going all in on chatbots. Even though they’ve got the green energy to to it. So in that respect, they’re already leading. The US is just racing to bottom.

Music Assistant most of the time. Sometimes I just use Symfonium directly but they both tap into my Navidrome server.

The article author doesn’t understand what Silicon Valley’s actual vision for the future is, but the students do.

It’s worked for their channel so far.

Just call it Condensation.

That’s the thing that’s been an issue. Companies give their LLMs access to everything so certain key people have access to these documents. But normally access is key coded, and without hacking in a way that’s usually very visible to sysadmins, you just cannot get access at all. With LLMs, it wants to give you what you want. There is not currently a way to keep it from being a pushover in some way. It is in part weakness of human language, and part weakness of programming it to work for whomever is doing the asking prompts. There is likely not a way to use language to make it keep secrets through all the possible ways to ask it to give you things. Nothing akin to the hardened ability of good old fashioned password protection at least. And that’s true with potential designs that we’ve not even seen yet. Currently, it can’t keep track of where data originated after a short time. It’s just all data to the model. So you might not easily get access to a file directly, but you can access what it knows about a file because again, it’s all just data and words at that stage.

I honestly don’t think you can create guard rails against prompt engineering in a working LLM. At some point, they’re going to fail or the LLM isn’t functioning. The only solution is to make sure they can’t read data you don’t want shared.