> Sometimes they're so overfit that the compression isn't even lossy, and the data is encoded verbatim in the NN.

Here's an article from November 2023 that discusses this:

https://not-just-memorization.github.io/extracting-training-...