Our gibberish tokens have varying degrees of robustness in combinations with contexts. E.g. if xx produces birds, ‘xx flying’ is an easy prompt ‘xx on a table’ is a neutral prompt, and ‘xx in space’ is a hard prompt. (8/N)
2022-06-03 15:09:35Our hidden vocabulary seems robust in easy and sometimes neutral prompts but not in hard ones. These tokens may produce low confidence in the generator and small perturbations move it in random directions. "vicootes" means vegetables in some contexts and not in others. (9/N) pic.twitter.com/r2s5I6Bnp0
2022-06-03 15:09:37We want to emphasize that this is an adversarial attack and hence does not need to work all the time. If a system behaves in an unpredictable way, even if that happens 1/10 times, that is still a massive security and interpretability issue, worth understanding. (10/N, N=10).
2022-06-03 15:09:37@benjamin_hilton, @realmeatyhuman, @BarneyFlames, @mattgroh, @rctatman, @Plinz, @Thomas_Woodside hopefully some of your concerns are addressed! Let us know what you think. We will update the pre-print with this discussion: arxiv.org/abs/2206.00169
2022-06-03 15:09:38論文の著者の1人
My student Giannis discovered that DALLE2 has a secret language. This can be used to crate absurd prompts that generate images. E.g. ''Apoploe vesrreaitais eating Contarra ccetnxniams luryca tanniounons'' generates Birds eating Bugs! We wrote a short paper on our experiments. twitter.com/giannis_daras/…
2022-06-01 02:47:43DALLE-2 has a secret language. "Apoploe vesrreaitais" means birds. "Contarra ccetnxniams luryca tanniounons" means bugs or pests. The prompt: "Apoploe vesrreaitais eating Contarra ccetnxniams luryca tanniounons" gives images of birds eating bugs. A thread (1/n)🧵 pic.twitter.com/VzWfsCFnZo
2022-06-01 02:44:25OpenAIの代表による反応
Turns out DALL-E can read the seemingly gibberish writing it produces. Built its own mini-language that is consistent between its text input space and image output space: twitter.com/giannis_daras/…
2022-06-01 09:07:50A pretty good hypothesis on how this arises (as an artifact of how the text input space is tokenized): twitter.com/BarneyFlames/s…
2022-06-01 09:11:49I took a look at the BPE encoding of the name DALL-E uses for birds. Its "apo, plo, e</w>, ,ve, sr, re, ait, ais</w>". Apo-didae & Plo-ceidae are families of birds, each with 100+ species. Apo-diformes is the biggest order of birds with 400+ species of birds. twitter.com/giannis_daras/…
2022-06-01 05:37:44その他外部サイト
ai-scholar
ITmedia
gigazine
反論のまとめもあり
GClue