• 3 Posts
  • 84 Comments
Joined 2 years ago
cake
Cake day: December 18th, 2023

help-circle









  • I disagree that it should be allowed, but I think their proposal would be something like attaching an identifier to the model, the random seed, the “temperature,” and any other relevant parameters that allow exact reproduction of the image without having access to anything but the model. Then you can prove it came from the model.

    Here’s a thought experiment, though, what would prevent someone from taking a real image and a model, then working with them until they can reproduce a very close approximation of the real image from text and parameter input? These models aren’t like a hash function, they can be viewed in reverse to some extent. Backpropagation is how they are trained.





  • jkrtn@lemmy.mltoWorld News@lemmy.world*Permanently Deleted*
    link
    fedilink
    English
    arrow-up
    15
    ·
    1 year ago

    The person in the article who apparently viewed it multiple times over a long period of time absolutely did so deliberately.

    I think the point in sending you the image was to show that, in general, it is possible for images to be present on your computer without you actively attempting to access them. Not to say that the argument was valid in this particular case.



  • jkrtn@lemmy.mltoWorld News@lemmy.world*Permanently Deleted*
    link
    fedilink
    English
    arrow-up
    20
    ·
    1 year ago

    It is, probably.

    But it won’t be written over with zeros, so it’s all still there until something else actually writes over it. A mobile device is flash memory, so the controller wear leveling might not get back to that spot for a bit. It might decide that spot is a bad sector and never write over it.