Ziggurat @ Ziggurat @sh.itjust.works

Posts

86

Comments

814

Joined

2 yr. ago

Subscribed, danke

In general I try to so. that said even if I fail to do so, why would you follow a community in a language you don't know.

By the way, beside ich_iel, which relax, laid back german community shall I follow to not forget the few German I know. All I do at the moment is listening once a month to DW Langsam gesprochene nachricten which isn't enough to stay fluent.

I installed gpt4all https://gpt4all.io/index.html it's a bit slow but runs on CPU. Then if you have a powerful GPU and like to play with technology I hard good stuff about that project https://github.com/oobabooga/text-generation-webui which wants to be the automatic111 of language model. But I haven't used it yet

Phone, SMS, Signal/whatsapp (which technically belongs to facebook)

Why would I use a social media to do so

Is it linked to the ongoing Drama on the french wikipedia ?

How does federation works with with "SEO" ? Wikipedia is always among the top result on search engine, how would peopel find about Ibis ?

How big is your instance ?

On a medium sized instance, The local feed might be a good start-up to find "popular users and hashtags" (Actually, I don't do it on Lemmy, but may-be I should)

What stuff do you find interesting ? You can follow writer like Greg Egan or Neil Gaiman, you can fallow European agencies like the one in charge of data protection, and then tons of bloggers, streamers, and other users.

Bonus question, was being a stock photo model a "full time job" or it was more an actor/model job that you do between more interesting projects ?

That said, with the ridiculous price of stock photo, and the remaining glitches in AI model, not sure whether it's really cheaper to generate an AI image (Need to get high resolution image, and to review them, you don't want a bad buzz because a necklace didn't close or whatever glitches are left) . It's basically depending whether some stock photo bank have implemented a natural language search.

Switching back to English if you don't mind.

I know that following the whole community works. But I wanted to try a lighter way by following only the V4uban to get only the daily thread on my mastodon feed (and then click on the main message to see the whole thread). I'm going to check the meta, but at least I know that it'not supposed to work that way

Mais tu es partout ma parole :) C'est justement l bot V4uban que j'ai essayé de suivre, mais sans résultat. Tant pis je repoueterais pas les discu interessante du FL sur Masto alors :)

I'm coming late, but here is it

To my understanding, the trick is that most AI aren't I'm drawing water, it's a reflecting material, so I need to build reflexion but more generate random image and denoise them until matching the description so if there was no symmetric noise fluctuation, it won't have reflexion.

May-be for next gen algorithm

For the rest, for one second I thought I was on the pics community and it was a photo

Do you understand how compound interest work ? If yes write it down on a paper, and the code will flow. I don't think we'll be helping you on the long term by doing your homework.

peertube uses bittorent to stream video.

Pro tip. NPC want to live too offers more alternative to combats, make the one happening more interesting.

The Borg games have a whole morale rules to let the NPC flee, but this can be managed based on what makes sense story wise

In the same category, I've seen instance using the photon interface e.g. https://p.jlai.lu/ and it's pretty neat.

But may-be there is a way to use all these interface in stand-alone without bothering our be-loved admin.

Just out of curiosity, what would have made an image with a clear theme

Regarding the theme is clear rule, I feel like it's badly named. At the time of the Monsters are back to school I came with some arbitrary rules to count point, to be neither the most upvoted nor OP's favourite, my original them is clear idea was to make sure that people stay in the week theme, but indeed that one hasn't been used much (in the sense everyone getting the bonus), so may-be worth removing it.

Another rule change which may be interesting is the weight of the "most upvoted", meaning that in the end it's always the most upvoted winning. May be it's interesting to try most upvoted 2, second most upvoted 1, or even Most upvoted 3, 2nd most upvoted 2, 3rd most upvoted 1, but these are just thought on point weighting

Pinging @Halcyon@discuss.tchncs.de as they'll be the one offering the next challenge. If you want to improve the rule, feel free, when I came with that point system 6 month ago I wasn't expected to be "holy rules"

Facebook and google killed XMPP that way, which is why people are very concerned regarding federation with threads.

However, I am not sure that this strategy worked that weight. Microsoft never killed competition from Linux and Mozilla