Comment

Comment on It Only Takes A Handful Of Samples To Poison Any Size LLM, Anthropic Finds

_cryptagion@anarchist.nexus ⁨2⁩ ⁨days⁩ ago

Ah, yes, the large limage model.

some random pixels have totally nonsensical / erratic colors,

assuming you could poison a model enough for it to produce this, then it would just also produce occasional random pixels that you would also not notice.

source

Sort:hotnew top

waterSticksToMyBalls@lemmy.world ⁨2⁩ ⁨days⁩ ago
That’s not how it works, you poison the image by tweaking some random pixels that are basically imperceivable to a human viewer. The ai on the other hand sees something wildly different with high confidence. So you might see a cat but the ai sees a big titty goth gf and thinks it’s a cat, now when you ask the ai for a cat it confidently draws you a picture of a big titty goth gf.

source
- Lost_My_Mind@lemmy.world ⁨2⁩ ⁨days⁩ ago
  …what if I WANT a big titty goth gf?
  
  source
  - phutatorius@lemmy.zip ⁨3⁩ ⁨hours⁩ ago
    You better stay away from mine, Romeo.
    
    source
  - TheBat@lemmy.world ⁨2⁩ ⁨days⁩ ago
    Get in line.
    
    source
  - waterSticksToMyBalls@lemmy.world ⁨2⁩ ⁨days⁩ ago
    Step 1: poison the ai
    
    source
- _cryptagion@anarchist.nexus ⁨2⁩ ⁨days⁩ ago
  Ok well I fail to see how that’s a problem.
  
  source
- Cherry@piefed.social ⁨2⁩ ⁨days⁩ ago
  Good use for my creativity. I might get on this over Christmas.
  
  source
PrivateNoob@sopuli.xyz ⁨2⁩ ⁨days⁩ ago
I have only learnt CNN models back in uni (transformers just came into popularity at the end of my last semesters), but CNN models learn more complex features from a pic, depending how many layers you add to it, and with each layer, the img size usually gets decreased by a multiplitude of 2 (usually it’s just 2) as far as I remember, and each pixel location will get some sort of feature data, which I completely forgot how it works tbf.

source