I think if enough people never gave them Internet access, the manufacturers would start adding in cellular modems to ensure they get the data flowing (that is, data on your viewing habits and sending you ads).

Yea that’s on the list for some point. I have a small k3s cluster running on some Pis and experimenting with tailscale.

I maxed out the free tier in my first month somehow lol… $20/yr isn’t a bad deal for essentially pihole everywhere.

Hey, that comment’s a bit off the mark. Transformers don’t just memorize chunks of text, they’re way more sophisticated than that. They use attention mechanisms to figure out what parts of the text are important and how they relate to each other. It’s not about memorizing, it’s about understanding patterns and relationships. The paper you linked doesn’t say anything about these models just regurgitating information.

Exactly. When does that ever happen? Covid 19 contact tracing was the pretense to get this tech through the door. Literally government trackers in everyone’s pockets.

Looks like I need to start saving things as I stumble upon them because searching for them later is fruitless. I’ll delete my comment as I can’t dig up a source.

The affiliate links are done by (almost?) every search engine so it’s not fair to single out Brave for it. Note I’m not defending them, if you’re truly up in arms about it talk about all of them doing it.

I’ve tried Kagi several times and the results for me are not good. I’ve pretty much gone back to Google unfortunately.

I used to subscribe to Reddit for the ad-free experience when I was a mobile web user. They kept making mobile web worse and worse and didn’t listen to user feedback after a point and made it so unusable I unsubscribed then found Apollo after refusing apps for years. Only been on Apollo maybe a year and now they’re destroying that. I’ve tried their app and it is a battery hog (spyware is my guess), works like crap and has too few features that I want .

There’s a few communities that I will miss over there but other than that I’m very excited for the fediverse and hope meta and bots don’t kill this platform before it gets going.

Last time I used 11 it had a few things which are deal breakers for me.

1. All windows of same application are grouped. In 10, you can have it group them never or only when task bar is full. On 11 you have no option. It takes an extra click every time you want to cycle to a specific window.
2. Seconds on the clock are gone. I use that on 10 occasionally and want it back.
3. Context menu in explorer was changed and is way more frustrating to use. My biggest gripe is the keyboard shortcuts were changed for no reason.
4. The ribbon menu is going away for nested menu system. Ribbon is objectively better for repeating multiple commands.
5. Gaming performance is worse, specifically VR.
6. UI/UX changes, some are good, some are bad. It’s not a polished OS with inconsistencies everywhere. In 10 at least they tried to make each level of the UI consistent with itself. 11 is just a hodgepodge of useless shit.

I would rather a lemmy instance only supported a single community. That would force the horizontal scaling better

Wikipedia says it was initially based on LaMDA then PaLM.

I’m not sure how you’re being downvoted, you make some great criticisms of the fediverse. Maybe staying small is better for the communities than reaching a critical mass as that only attracts corporate sponsorship (VC/ads/etc.). Working out the security should be the initial goals for any interested programmers who venture over.

From the documentation:

A PeerTube instance can mirror other PeerTube videos to improve bandwidth use.

The instance administrator can choose between multiple redundancy strategies (cache trending videos or recently uploaded videos etc.), set their maximum size and the minimum duplication lifetime. Then, they choose the instances they want to cache in Manage follows -> Following admin table.

Videos are kept in the cache for at least min_lifetime, and then evicted when the cache is full.

I haven’t yet looked at the Lemmy architecture of the server, but it should be fairly straightforward to have a redundant region on AWS and use Route53 for the failover.