[-] joeldebruijn@lemmy.ml 2 points 7 months ago

To illustrate even more: I follow a Lemmy privacy "group" with an Akkoma account.

[-] joeldebruijn@lemmy.ml 2 points 8 months ago

Agree, no small feat. Two caveats tho:

  • These models prioritize plausibility above factual correctness. So verification often is needed.
  • Data from after the creation of trainingmaterial is absent of course.
[-] joeldebruijn@lemmy.ml 2 points 8 months ago

Train it online. Use it offline.

[-] joeldebruijn@lemmy.ml 2 points 8 months ago* (last edited 8 months ago)

Just found a good starting point, a repository of 64 templates, with each just index.html, some style.css and sometimes more but not much:

https://github.com/MichaelBarney/LinkFree/tree/master/Templates

[-] joeldebruijn@lemmy.ml 2 points 8 months ago

I hope to get there one day (being able to DIY) but so much to learn (in the middle of getting away from MS, Google and migrating to NextCloud, Linux, Yunohost etc.)

[-] joeldebruijn@lemmy.ml 2 points 8 months ago

Just to add to your list of steps and consequences: I also think academic studies about information retrieval, indexing and crawling became less popular. Aspirant students hearing the message: those studies / workfields will become obsolete once AI does all that.

[-] joeldebruijn@lemmy.ml 2 points 8 months ago

When I let fsearch index from root, it counted 1.9 million files, which baffled me a lot. Before knowing the things in this thread. A typical windows install can have 50k ~ 100k files, but .... 2 million I thought it was insane.

But in this context its something like if LibreOffice Calc had an API and upon start it registers a filesystem with a 'folder' for every worksheet and a 'file' named A1, A2, B1 .... for every cell. Not real I know but a novice way of understanding.

[-] joeldebruijn@lemmy.ml 2 points 8 months ago

Agree, wholeheartedly and reasons I want to avoid Discord et al. I do communicate my expectations rather cynically in case a community is starting and does have a choice in the beginning.

[-] joeldebruijn@lemmy.ml 2 points 8 months ago

For now ... ... endgame is integration in WP core and one needs forking to avoid it ....

[-] joeldebruijn@lemmy.ml 2 points 8 months ago* (last edited 8 months ago)

Now where is that article about storing data by etching it on glass ....

https://youtu.be/JUtWdASRjBc

[-] joeldebruijn@lemmy.ml 2 points 9 months ago

True! One of the main building blocks, sadly.

[-] joeldebruijn@lemmy.ml 2 points 1 year ago

Depends on your definition of tolerant.

If it implies adtech and trackers, then no. If it implies joining a random instance and spamming it, then no.

But any organisation is free to manage their own fediverse instance or get managed Mastodon hosting on their own domain.

So fans of their products can follow them and their webcare team can respond to questions.

view more: ‹ prev next ›

joeldebruijn

joined 1 year ago