this post was submitted on 24 Sep 2025
328 points (96.3% liked)

196

4542 readers
2190 users here now

Community Rules

You must post before you leave

Be nice. Assume others have good intent (within reason).

Block or ignore posts, comments, and users that irritate you in some way rather than engaging. Report if they are actually breaking community rules.

Use content warnings and/or mark as NSFW when appropriate. Most posts with content warnings likely need to be marked NSFW.

Most 196 posts are memes, shitposts, cute images, or even just recent things that happened, etc. There is no real theme, but try to avoid posts that are very inflammatory, offensive, very low quality, or very "off topic".

Bigotry is not allowed, this includes (but is not limited to): Homophobia, Transphobia, Racism, Sexism, Abelism, Classism, or discrimination based on things like Ethnicity, Nationality, Language, or Religion.

Avoid shilling for corporations, posting advertisements, or promoting exploitation of workers.

Proselytization, support, or defense of authoritarianism is not welcome. This includes but is not limited to: imperialism, nationalism, genocide denial, ethnic or racial supremacy, fascism, Nazism, Marxism-Leninism, Maoism, etc.

Avoid AI generated content.

Avoid misinformation.

Avoid incomprehensible posts.

No threats or personal attacks.

No spam.

Moderator Guidelines

Moderator Guidelines

  • Don’t be mean to users. Be gentle or neutral.
  • Most moderator actions which have a modlog message should include your username.
  • When in doubt about whether or not a user is problematic, send them a DM.
  • Don’t waste time debating/arguing with problematic users.
  • Assume the best, but don’t tolerate sealioning/just asking questions/concern trolling.
  • Ask another mod to take over cases you struggle with, if you get tired, or when things get personal.
  • Ask the other mods for advice when things get complicated.
  • Share everything you do in the mod matrix, both so several mods aren't unknowingly handling the same issues, but also so you can receive feedback on what you intend to do.
  • Don't rush mod actions. If a case doesn't need to be handled right away, consider taking a short break before getting to it. This is to say, cool down and make room for feedback.
  • Don’t perform too much moderation in the comments, except if you want a verdict to be public or to ask people to dial a convo down/stop. Single comment warnings are okay.
  • Send users concise DMs about verdicts about them, such as bans etc, except in cases where it is clear we don’t want them at all, such as obvious transphobes. No need to notify someone they haven’t been banned of course.
  • Explain to a user why their behavior is problematic and how it is distressing others rather than engage with whatever they are saying. Ask them to avoid this in the future and send them packing if they do not comply.
  • First warn users, then temp ban them, then finally perma ban them when they break the rules or act inappropriately. Skip steps if necessary.
  • Use neutral statements like “this statement can be considered transphobic” rather than “you are being transphobic”.
  • No large decisions or actions without community input (polls or meta posts f.ex.).
  • Large internal decisions (such as ousting a mod) might require a vote, needing more than 50% of the votes to pass. Also consider asking the community for feedback.
  • Remember you are a voluntary moderator. You don’t get paid. Take a break when you need one. Perhaps ask another moderator to step in if necessary.

founded 8 months ago
MODERATORS
 

INFO: image above "slop" text is generated by diffusion model (i found it on duckduckgo, did not generate it myself)

alternative image link on blahaj zone

i looked at dis "stop doin math" funi, n thought "eh - why not" n now i made dis.

generally tried to keep it as close to the original as possibl, while also makin it maria-themed - so here we go. i think it turned out preddi well.

in case dis post blows up n reaches the lemmy world peeps who dun kno me, here a little note to all u who r about to write a comment bout my writin style:

  • i dun hav a stroke, n so do u.

n for anyone curious, here the matrix multiplication thingy i mentioned in smol text. tldr: google put their gemini-2.0 into a loop of tryin stff out, n at som point it came up with the currently fastest matrix multiplication method to date.

EDIT:

for anyone even more curious (only read this if u care bout my stance)yes, dis post is kindsa meant to be sepf-deprecating humor.

u might wondr buut mariaaaaa - why??? whads self-deprecating bout postin out the obvious downsides of current ml systems?

well let me tell u a story ~ ~ ~ ~ ~ 🌳🌸💖🍩🍙🥧

(which i wrote btw, no language model was used)

once upon a time, there was a maria 🌸 in a magical world of gud 🟢 and evil 🟣.

maria 🌸 hated how the evil magic 🟣 was destroying the environment: evil powerful magic-wielders 😈 would create huge evil-magic-producers 🟪 with the hope of replacing the human!

she also hated how suddenly, these dark-magic creatures 💜 wud write boring useless text and create uncanny scary imagery. typical of the evil-magics 😈, she 🌸 thought.

the black-magic-advancements 🟣 at the time scared her so much, she believed all of human work could be replaced with this new dark magic 🟣! she imagined a future where no peep worked, with many takin their lifes becuz they felt it was meaningless without goals to achieve.


despite all dis, maria 🌸 imagined they cud be used for gud, so she 🌸 lead one of the little creatures 💜 home. not to profit from, but to see what it 💜 cud do ~

at home, she 🌸 put the dark creature 💜 into a bunni-cage 🐇, wrote a page of instructions and pushed it thru the cage to the creature 💜

it 💜 licked the page, and started pretending to do the task, by writing down the actions it 💜 would take... but not actually doing them...

turns out, she 🌸 had to let the written text be turned into actual actions using another well-known ancient spell: text-parsing 🟢 :o

after bits of tinkerin, maria 🌸 ... didn hav anythin useful, but she had fun trynna make the lil one 💜 do increasingly complex tasks. at som point, she felt accelerated by her own doings, n put it 💜 into a larger, bear-sized cage, n gave it 💜 much longer, mor complex instructions

she 🌸 wud change its 💜 personality, look at its 💜 own writing, refine it, revise it, test its 💜 ideas, see its 💜 own failure n keep on tryin.

...

unknowingly, she 🌸 had becom a black magic wielder 🟣. she 🌸 had becom, whad she 🌸 hated so much 🟣 -

though maria 🌸 didn giv brith to the monster 💜 (which wud increase environmental problms), she only fed it 💜 vrri little magical energy (much less than others wud use to play magic games) - n didn share the creatures 💜 scribbles with others, cuz she didn wana be like those other peeps 🟣 who pretend the output is worth sharin.

she 🌸 imagined her frens frowning upon her, once findin out bout her 🌸 "hobby" 🟣.

...

one question bugged her tho:

aftr all dis time, am i still maria 🌸? or hav i becom... maria 😈?

she couldnt answer dis question.


the end

oh yea - u prolli need som context:

i am developin an lm-based helper thingy for godot game engine for about a year by now. dis is why the post is kindsa self-deprecating, its cuz im litrlli buildin a thing which "does tasks". vrri funi.

anway, plz dun call be an ai-sis, i did make dis funi pictur aftr all, so i must be sane, right?

(plz be cool bout dis >o< )


^ ^ ^ dis includes a little story too btw - its got mojis! ^ ^ ^

you are viewing a single comment's thread
view the rest of the comments
[–] Smorty@lemmy.blahaj.zone 3 points 5 days ago

ur right, LMs cn translate natural languag queries into som json format to take action.

whad i was referrin to are agents, where u put ur LM inti the driver seat n pet it go do thad task for u (like go shopping, clean the house or write and test som code).

in the "base model stage" of these LMs, we see them only predict online text, not follow instructions - n theres not that much step-by-step data on how to do tasks. So currently, LM labs create synthetic data, to make their models learn common patterns.

so - mor or less by definution - they were nit made to do tasks, we kindsa force em to-

but yesyis, the home assistant case is a reasonable application of LMs <3 (surprised amazons echo doesn use dis yet)