redcalcium

I'm not familiar with yunohost, but if your Lemmy instance is running under docker, then 127.0.0.1 loopback IP address will point to the docker container instead of your host computer.

I'll have to try this later. Thanks for the tip!

I really doubt that, at least for the next few years. "AI Assistant" usually means LLMs, and even M2 struggles to run them mostly due to large compute and RAM requirements. If Microsoft could somehow release a truly local AI assistant feature that can run on average windows users' hardware, that would be shake the whole ML industry.

Microsoft has recently announced Windows Copilot, an AI-powered assistant for Windows 11. Windows Copilot sits at the side of Windows 11, and can summarize content you’re viewing in apps, rewrite it, or even explain it. Microsoft is currently testing this internally and promised to release it to testers in June before rolling it out more broadly to Windows 11 users.

Assuming this will use OpenAI API like other Microsoft's AI products, this is going to be expensive to operate. Subsidizing it indefinitely is surely not an option. How would Microsoft monetize it? By charging subscription like GitHub Copilot, or monetizing it somehow using users data they collected? I assume it would be the latter.

There goes my dream building my own computer if I survived a nuclear apocalypse.

Looks like there are 16 wires for each core in this module?

All this stuff is really cool! Looks like you can use one ring to store multiple bits by using multiple sense wire. So that's why there are a lot of wires in the images. Seems like there are 16 bits for each core?

I wonder how feasible it is for someone to built their own retro computers by soldering a bunch of nand gates and weaving their own ram and rom. The only problem seems to be getting the ferrite cores in huge quantities (are those still being sold these days?).

After digging some more, looks like core rope memory is different than the magnetic core memory in your link above (seems to describe magnetic core memory instead of rope memory). Core rope memory is used for ROM while magnetic core memory is used for RAM, and they have different working mechanism.

Another form of core memory called core rope memory provided read-only storage. In this case, the cores, which had more linear magnetic materials, were simply used as transformers; no information was actually stored magnetically within the individual cores. Each bit of the word had one core. Reading the contents of a given memory address generated a pulse of current in a wire corresponding to that address. Each address wire was threaded either through a core to signify a binary [1], or around the outside of that core, to signify a binary [0]. As expected, the cores were much larger physically than those of read-write core memory. This type of memory was exceptionally reliable. An example was the Apollo Guidance Computer used for the NASA Moon landings.

Basically, 1 if the sense wire going through the core, or 0 if the wire bypass the core. You basically woven 1 and 0 manually into the rope. And yeah, it doesn't matter if the cores touch each other as long as the wires are woven correctly. This article have some wiring diagram for the core rope memory: http://www.righto.com/2019/07/software-woven-into-wire-core-rope-and.html .

Is that... 32 bytes of memory? Wow, I didn't know they also came in rope form factor as well.

Some cores appear to touch other neighboring cores. Won't that cause issues to the core's magnetic properties?

I imagine we'll hit the ceiling soon enough until the next major breakthrough is found. There is also concern about gathering good training data in the future as the internet started to get littered with AI generated contents. I thin I read an article earlier about how a machine learning system perform worse if it's trained with secondary data generated by another ML system.

along with being able to consume other posts via ActivityPub

I read a new article that said it remains to be seen whether P92 will allow users to see posts from other site (they'll broadcast to ActivityPub but undecided about displaying contents from federated servers): https://tech.co/news/meta-decentralized-social-media

A source close to the project also told MoneyControl that “the plan as of now is that the MVP (minimum viable product) will definitely allow our users to broadcast posts to people on other servers”, but admitted the company is yet to decide whether to allow users “to follow and view the content of people on other servers.”

If they only broadcast, but not displaying contents from other servers or allow their users to follow people from other server, then what's the point of adding federated support if people from other servers can't interact with them?

The FOSS community is wary due to "embrace, extend, extinguish" approach by various tech giants in the past. When a tech giant suddenly want to embrace federation while offering no details whatsoever, people are right to be wary.