To be fair with the above, even considering that he’s being disingenuous*, his [AFAIK incorrect] claim is not “anime is child porn”, it’s “that anime instance has child porn”.

*note how he’s trying to transform “is this CSAM?” into a subjective matter. That’s rather close to the moving goalposts fallacy.

Even in this thread there’s discussion of a show that blatantly tittilates the audience with underage characters that would absolutely qualify as csam in any other community except in the anime community, for some reason.

Emphasis mine. If what you are saying is indeed correct (is it? dunno), this is a sign that the acronym “CSAM” was completely derailed.

Originally the expression “child sexual abuse material” was coined to avoid implications of consent brought by the word “pornography”, and it boils down to “evidence of child sexual abuse”. Consent and sexual abuse are legal notions that only apply to real people, not to fictional characters.

In the meantime, at worst the instance in question depicts images of clearly fictional characters in suggestive poses and/or clothing. It does not classify even as pornography, let alone sexual abuse. (Note that not even hentai depicting clearly adult characters is allowed in that instance.)

I don’t care about what the maintainers’ view of the matter is, I make (and sometimes delete) my comments based on my own view of it.

Given that this is a touchy subject, I think that this matter is better handled neither by the maintainers’ views nor by our own views, but by 1) legal definitions of governments that might be relevant in the matter, and 2) explicit moral premises.

Yeah, but the admins, as the thread has shown, are mainly reining in violations of sitewide policy. Instance rules are mainly the job of mods.

So the admins are reining in violations of lemmy.ml-wide policy… while lemmy.ml rules are mainly the job of the mods??? Congratulations, that’s the dumbest thing that I’ve read today.

Couple the above with the backpedalling (from “This is what mods are for.” to “Instance rules are mainly the job of mods.”; emphasis on “mainly”) - a sleight of hand, while lying that I was the one using a sleight of hand - and I’m led to the conclusion that you have nothing meaningful to add to this discussion, and can be safely ignored as dead weight and noise.

Unlike the above, does anyone here have any decent counter-argument against “migrating this comm to that other instance would be sensible”?

Want some cuteness overload? Even big felines love cardboard boxes!

Removed by mod

Not just the mods. Admins can (and should) also moderate content in their instances, specially when it comes to the global rules. And it’s clear that lemmy.ml admins want to do so, otherwise this thread wouldn’t exist on first place.

Sorry for the double reply. Here’s a practical idea: what if the mods of this comm contacted lemmy.ml’s admins? Ideally doing two things:

1. Clarifying that the instance in question does not have child sexual abuse material, and requesting users to be allowed to link it.
2. Expressing desire to migrate !anime@lemmy.ml to the instance in question, and highlighting that this is convenient for both sides of the matter.

Among the admins I think that Nutomic would be the best to contact, given the github thread.

You’re talking about your thread about Mahou Shoujo ni Akogarete, right? It’s still in the modlog for me, even in private mode. I don’t think that they removed the entry.

Lunix sucks so much that it got stuck into the version 2 for years.

Removed by mod

In my opinion, the migration is sensible because:

1. You need to know a topic decently enough to be able to moderate a community about it. And yet it’s unreasonable to expect the admins of lemmy.ml - a community about free/open source and privacy - to know about anime.
2. There’s no inherent reason why this comm is the largest anime comm in Lemmy. It’s simply that this comm is three years old, from a time when “Lemmy” was mostly just lemmy.ml and lemmygrad.ml, there was no other place in Lemmy to discuss anime.
3. Lemmy as a whole benefits from redirecting traffic from larger instances to smaller ones. Specially instances unrelated to politics and tech. In fact, lemmy.ml’s admins asked users to use other instances not too long ago.

For me it says “admins”. I don’t know if it’s because I’m a mod of a few other lemmy.ml comms, or if it’s because of the different backend versions (SJW uses 0.18.5, lemmy.ml uses 0.19.1).

For context, I encourage people to check this discussion in the “join Lemmy” site github. Have in mind that both of the Lemmy developers in that discussion are also admins of the lemmy.ml instance, and they clearly disagree if the instance in question should be considered as “hosting CSAM” or not.

I think that this community should migrate, but for a different reason: topic.

The topic of lemmy.ml is privacy and free-as-freedom software. Most other content here is off-topic, including anime. It was fine when it was just lemmy.ml and lemmygrad.ml, as you’d have no other place to discuss anime in Lemmy, but now the situation is different.

And ideally, communities should be managed/moderated/administered by people who know well the topic of the community.

If I were to discipline my cats by picking them up, one of them would be exactly like in this pic. The other would be tearing my arms apart with the claws of her back legs.

Aaaaah. I really, really wanted to complain about the excessive amount of keys.

(My comment above is partially a joke - don’t take it too seriously. Even if a new key was added it would be a bit more clutter, but not that big of a deal.)

The source that I’ve linked mentions semantic embedding; so does further literature on the internet. However, the operations are still being performed with the vectors resulting from the tokens themselves, with said embedding playing a secondary role.

This is evident for example through excerpts like

The token embeddings map a token ID to a fixed-size vector with some semantic meaning of the tokens. These brings some interesting properties: similar tokens will have a similar embedding (in other words, calculating the cosine similarity between two embeddings will give us a good idea of how similar the tokens are).

Emphasis mine. A similar conclusion (that the LLM is still handling the tokens, not their meaning) can be reached by analysing the hallucinations that your typical LLM bot outputs, and asking why that hallu is there.

What I’m proposing is deeper than that. It’s to use the input tokens (i.e. morphemes) only to retrieve the sememes (units of meaning; further info here) that they’re conveying, then discard the tokens themselves, and perform the operations solely on the sememes. Then for the output you translate the sememes obtained by the transformer into morphemes=tokens again.

I believe that this would have two big benefits:

1. The amount of data necessary to “train” the LLM will decrease. Perhaps by orders of magnitude.
2. A major type of hallucination will go away: self-contradiction (for example: states that A exists, then that A doesn’t exist).

And it might be an additional layer, but the whole approach is considerably simpler than what’s being done currently - pretending that the tokens themselves have some intrinsic value, then playing whack-a-mole with situations where the token and the contextually assigned value (by the human using the LLM) differ.

[This could even go deeper, handling a pragmatic layer beyond the tokens/morphemes and the units of meaning/sememes. It would be closer to what @njordomir@lemmy.world understood from my other comment, as it would then deal with the intent of the utterance.]

Not quite. I’m focusing on chatbots like Bard, ChatGPT and the likes, and their technology (LLM, or large language model).

At the core those LLMs work like this: they pick words, split them into “tokens”, and then perform a few operations on those tokens, across multiple layers. But at the end of the day they still work with the words themselves, not with the meaning being encoded by those words.

What I want is an LLM that assigns multiple meanings for those words, and performs the operations above on the meaning itself. In other words the LLM would actually understand you, not just chain words.

Complexity does not mean sophistication when it comes to AI and never has and to treat it as such is just a forceful way to make your ideas come true without putting in the real effort.

It’s a bit off-topic, but what I really want is a language model that assigns semantic values to the tokens, and handles those values instead of directly working with the tokens themselves. That would be probably far less complex than current state-of-art LLMs, but way more sophisticated, and require far less data for “training”.

creating a label and checking the skip invoice box

That works great too, specially if you want to use less foolproof filters. Or even a mix of both strategies.