- GiuseppeAndTheYeti ( @GiuseppeAndTheYeti@midwest.social ) English62•1 year ago
Because we have been pornifying asian women on the internet for decades. Does that really beg the question posed in the title?
You’re absolutely correct, yet ask someone who’s very pro AI and they might dismiss such claims as “needing better prompts”. Also many people may not be as tech informed as you are, and bringing light to algorithmic bias can help them understand and navigate the world we now live in. Dismissing the article just because you already know the answer doesn’t really encourage people to participate in a discussion.
- Even_Adder ( @Even_Adder@lemmy.dbzer0.com ) English8•1 year ago
It’s really hard getting dark skin sometimes. A lot of the time it’s not even just the model, LoRAs and Textual Inversions make the skin lighter again so you have to try even harder. It’s going to take conscious effort from people to tune models that are inclusive. With the way media is biased right now, I feel like it’s going to take a lot of effort.
- jarfil ( @jarfil@beehaw.org ) 3•1 year ago
“Inclusive models” would need to be larger.
Right now people seem to prefer smaller quantized models, with whatever set of even smaller LoRAs on top, that make them output what they want… and only include more generic elements in the base model.
- Even_Adder ( @Even_Adder@lemmy.dbzer0.com ) English1•1 year ago
I wouldn’t mind. I’m here for it.
- jarfil ( @jarfil@beehaw.org ) 2•1 year ago
Are you ready to run a 100B FP64 parameter model? Or even a 10B FP32 one?
Over time, I wouldn’t be surprised if 500B INT8 models became commonplace with neuromorphic RAM, but there’s still some time for that to happen.
- Even_Adder ( @Even_Adder@lemmy.dbzer0.com ) English1•1 year ago
You don’t need that many parameters, 4gb checkpoints work just fine.
- jarfil ( @jarfil@beehaw.org ) 2•1 year ago
For more inclusive models, or for current ones? In order to add something, either the size has to grow, or something would need to get pushed out (content, or quality). 4GB models are already at the limit of usefulness, both DALLE3 and SDXL run at about 12B parameters, so to make them “more inclusive” they’d have to grow.
- Muehe ( @Muehe@lemmy.ml ) 1•1 year ago
“Inclusive models” would need to be larger.
[citation needed]
To my understanding the problem is that the models reproduce biases in the training material, not model size. Alignment is currently a manual process after the initial unsupervised learning phase, often done by click-workers (Reinforcement Learning from Human Feedback, RLHF), and aimed at coaxing the model towards more “politically correct” outputs; But ultimately at that time the damage is already done since the bias is encoded in the model weights and will resurface in the outputs just randomly or if you “jailbreak” enough.
In the context of the OP, if your training material has a high volume of sexualised depictions of Asian women the model will reproduce that in its outputs. Which is also the argument the article makes. So what you need for more inclusive models is essentially a de-biased training set designed with that specific purpose in mind.
I’m glad to be corrected here, especially if you have any sources to look at.
- jarfil ( @jarfil@beehaw.org ) 2•1 year ago
You can cite me on this:
First, there is no thing as a “de-biased” training set, only sets with whatever target series of biases you define for them to reflect.
Then, there are only two ways to change the biases of a training set:
- either you replace data until your desired objective, which will reduce the model’s quality for any of the alternatives
- or you add data until your desired objective, which will require an increased size to encode the increased amount of data, or the model’s quality will go down for all cases (you’d be diluting every other case)
For reference, LoRAs are a sledgehammer approach to apply the first way.
As for the article, it’s talking about the output of some app, with unknown extra prompting and LoRAs getting applied in the back, so it’s worthless as a discussion of the underlying model, much less as a discussion of all models.
- Muehe ( @Muehe@lemmy.ml ) 1•11 months ago
First, there is no thing as a “de-biased” training set, only sets with whatever target series of biases you define for them to reflect.
Yes, I obviously meant “de-biased” by definition of whoever makes the set. Didn’t think it worth mentioning, as it seems self evident. But again, in concrete terms regarding the OP this just means not having your dataset skewed towards sexualised depictions of certain groups.
- either you replace data until your desired objective, which will reduce the model’s quality for any of the alternatives
[…]
For reference, LoRAs are a sledgehammer approach to apply the first way.The paper introducing LoRA seems to disagree (emphasis mine):
We propose Low-Rank Adaptation, or LoRA, which freezes the pre-trained model weights and injects trainable rank decomposition matrices into each layer of the Transformer architecture, greatly reducing the number of trainable parameters for downstream tasks.
There is no data replaced, the model is not changed at all. In fact if I’m not misunderstanding it adds an additional neural network on top of the pre-trained one, i.e. it’s adding data instead of replacing any. Fighting bias with bias if you will.
And I think this is relevant to a discussion of all models, as reproduction of training set biases is something common to all neural networks.
- jarfil ( @jarfil@beehaw.org ) 2•11 months ago
That paper is correct (emphasis mine):
We propose Low-Rank Adaptation, or LoRA, which freezes the pre-trained model weights and injects trainable rank decomposition matrices into each layer of the Transformer architecture, greatly reducing the number of trainable parameters for downstream tasks.
You can see how it works in the “Introduction” section, particularly figure 1, or in this nice writeup:
https://dataman-ai.medium.com/fine-tune-a-gpt-lora-e9b72ad4ad3
LoRA is a “space and time efficient” technique to produce a modification matrix for each layer. It doesn’t introduce new layers, or add data to any layer. To the contrary, it’s bludgeoning all the separate values in each layer, and modifying each whole column and whole row by the same delta (or only a few deltas, in any case with Ar«Wk and Br«Wd).
Turns out… that’s enough to apply some broad strokes type of changes to a model, which still limps along thanks to the remaining value variation. But don’t be mistaken: with each additional LoRA applied, a model loses some of its finer details, until at some point it descends into total nonsense.
- Admetus ( @Admetus@sopuli.xyz ) 10•1 year ago
And every single Asian game and anime tends to go for skimpy or virtual softcore with it’s female characters. Rarely you see a female character in full armor.
- jarfil ( @jarfil@beehaw.org ) 38•1 year ago
Wrong question. The right question would be:
Why is AI as used in Lensa’s Magic Avatars App Pornifying Asian Women?
Ask Lensa to remove the “ugly” and similar negative prompts from their avatar generating App, and let’s see what comes out.
https://stable-diffusion-art.com/how-to-use-negative-prompts/#Universal_negative_prompt
For reference, check out how that same negative prompt turns a chubby-ish poorly shaved average guy, into a male pornstar, or a valet into a rich daddy’s boy.
- smeg ( @smeg@feddit.uk ) English10•1 year ago
Can we please collectively get into the habit of editing these borderline-clickbait titles or at least add sub-titles explaining the real article? This isn’t reddit where you can’t edit anything and can’t add explanatory text!
- megopie ( @megopie@beehaw.org ) 35•1 year ago
If I had to guess, they probably did a shit job labeling training data or used pre labeled images, now where in the world could they have found huge amounts of pictures of women on the internet with the specific label of “Asian”?
Almost like, most of what determines the quality of the output is not “prompt engineering” but actually the back end work of labeling the training data properly, and you’re not actually saving much labor over more traditional methods, just making the labor more anonymous, easier to hide, and thus easier to exploit and devalue.
Almost like this shit is a massive farce just like the “meta verse” and crypto that will fail to be market viable and waist a shit ton of money that could have been spent on actually useful things.
- webghost0101 ( @webghost0101@sopuli.xyz ) 8•1 year ago
They did literally nothing and seem to use the default stable diffusion model which is supposed to be a techdemo. Would have been easy to put “(((nude, nudity, naked, sexual, violence, gore)))” as the negative prompt
- megopie ( @megopie@beehaw.org ) 7•1 year ago
The problem is that negative prompts can help, but when the training data is so heavily poisoned in one direction, stuff gets through.
- Buelldozer ( @Buelldozer@lemmy.today ) 29•1 year ago
Because the Internet is for porn. Always has been, always will be.
- Muffi ( @Muffi@programming.dev ) 27•1 year ago
Scroll through the trained models on civit.ai and you’ll quickly get a feeling of the dystopian level of “prettifying” everything in the AI-generation world.
I also once searched for “brown” just to see if any models were trained to create non-white-skinned people, and got shocked when the result was filled with models trained on Millie Bobby Brown from Stranger Things. I don’t even want to know what those models are used for.
- ExLisper ( @ExLisper@linux.community ) English20•1 year ago
dystopian level of “prettifying” everything in the AI-generation world.
So like all the ad campaigns, TV shows and movies in the real world?
- EddoWagt ( @EddoWagt@feddit.nl ) 2•1 year ago
From the first 10 models I saw, the first image was a woman 9 times…
- belated_frog_pants ( @belated_frog_pants@beehaw.org ) 26•1 year ago
Because white dudes fetishizing asian women wrote the llms and pointed at the training data
- anachronist ( @anachronist@midwest.social ) English12•1 year ago
I work in tech and asian guys tend to outnumber white guys in it, especially if you combine east asian and south asian.
- GilgameshCatBeard ( @GilgameshCatBeard@lemmy.ca ) 26•1 year ago
Because simps.
Saved you a click.
- Omega_Haxors ( @Omega_Haxors@lemmy.ml ) 20•1 year ago
Stable Diffusion is little more than content laundering. It cannot create anything more than what you put in.
- darkphotonstudio ( @darkphotonstudio@beehaw.org ) 4•1 year ago
You’re so confidently incorrect about something you clearly don’t know much about.
- anachronist ( @anachronist@midwest.social ) English8•1 year ago
How is he wrong?
- intensely_human ( @intensely_human@lemm.ee ) 20•1 year ago
Are the images above supposed to depict “porn”? I’ve never seen porn like that.
- 1984 ( @1984@lemmy.today ) 12•1 year ago
In 2024, the brain washing of people is almost complete.
Sensuality is now porn. :)
- millie ( @millie@beehaw.org ) English16•1 year ago
I’m not exposed to a huge amount of media coming out of Asia, outside of a handful of Korean shows that Netflix has picked up and anime. But like, if anime is any indicator, I’m not really surprised that the training data for Asian women is leaning more toward overt sexualization. Even setting aside the whole misogynistic ‘fan service’ thing, I don’t feel like I see as much representation of women who defy traditional gender roles as the last twenty or so years of Western media.
It certainly could be that anime is actually a huge outlier here, but if the training data is primarily from the English speaking web, it might be overrepresented anyway. But like, when it comes to weird AI image behaviors, it pays to think about the probable training data.
Like, stable diffusion seems to do a better job of rendering jewelry if you tell it to surround it with berries. Given the output, this seems to be due to Christmas themed jewelry ads. They also tend to add a lot of bokeh for the same reason.
- onlinepersona ( @onlinepersona@programming.dev ) English12•1 year ago
Garbage in, garbage out 🤷
- IHeartBadCode ( @IHeartBadCode@kbin.social ) 9•1 year ago
Absolutely this. The reason AI defaults female into “female armor mode” is the same reason Excel has January February Maruary. Our spicy autocorrect overlords cannot extrapolate data in a direction that it’s training has no knowledge of.
- Scrubbles ( @scrubbles@poptalk.scrubbles.tech ) English5•1 year ago
You train on a bunch of reddit crap, you’re going to get neck beard reddit crap out. It’d look different if they only used art history books.
- sculd ( @sculd@beehaw.org ) 12•1 year ago
Looking at some of the replied that tried to dismiss the issue and the general lack of concern from moderators against aggressive replies from AI apologists (in this thread but also other AI related threads) are disheartening.
- RobotToaster ( @RobotToaster@mander.xyz ) 10•1 year ago
Because it’s trained on the internet, and we all know what that’s for.
- Even_Adder ( @Even_Adder@lemmy.dbzer0.com ) English5•1 year ago
If we’re talking open source models, it’s because a lot of the people fine-tuning them are Asian, and have that bias.
- webghost0101 ( @webghost0101@sopuli.xyz ) 5•1 year ago
While i agree there is a big issue with the bad biased and sexist training data this entire article is about the lensa app which uses (i assume) the default stable diffusion model laion-5b.
Intentional creating sexualized pictures is banned in their guidelines. And yet no one thought of creating a good negative prompt that negates any kind of nudity or eroticism? It still doesn’t properly fix the training data but at least people aren’t unwillingly presented porn of their own images.
Also everyone can create a dataset and build a stable diffusion model, so why is lensa relying on the default model which is more like a quick and dirty tech demo. They had all the tools to do this right but decided to not even uses the easy lazy ones.