it's not possible dawg
you cannot extract training data from a model. that's simply how they work

erm you don't get to engage in a conversation with someone and block them when they're right
and ftr these are all accounts i've had for a long time i just enjoy making accounts

the model doesn't have any code in it. the code tells the model what to do, but the code simply cannot make the model produce images because that's not what the model does

and the code does exactly nothing to a model that has already been trained
it's not like they're publishing a CSAM database. once the model is trained you simply cannot make it produce images. that's how ML works

you're the one that insists human moderators should be watching child porn lol

of course you can change the code. but you cannot extract the images from an already trained ML model, nor can you make an image detection model produce images
that's not what stalking is. i would love to have a real conversation about this

then provide some kind of high level overview of how you could make the detection model produce images? that's simply not how it works

straight cap unfortunately
you blocked me and everyone else that can form a coherent argument against you
again, please walk me through how you can reverse engineer an image detection model

xn--jw9h

this one is 🧼.bsky.social

i'm saving the punycode handle for if bluesky ever supports them

(fake)