That is a little like saying every photograph is a copy of the thing. That is just factually incorrect. I have many three layer networks that are not the thing they were trained on. As a compression method they can be very lossy and in fact that is often the point.
Removed by mod
Sure, if that is what the network has been trained to do, just like a librarian will if that is how they have been trained.
Removed by mod
Interesting article. It seems to be about a bug, not a designed behavior. It also says it exposes random excerpts from books and other training data.
Removed by mod
That is a little like saying every photograph is a copy of the thing. That is just factually incorrect. I have many three layer networks that are not the thing they were trained on. As a compression method they can be very lossy and in fact that is often the point.