arXiv Computation and Language By Filip Sondej, Yushi Yang, Adam Mahdi

RepSelect: Robust LLM Unlearning via Representation Selectivity

Read the original on arXiv Computation and Language →

arXiv:2606. 17168v3 Announce Type: replace Abstract: When LLM weights are open or fine-tuning is available through an API, suppressing hazardous knowledge and tendencies is not enough: removal has to be deep enough that an adversary cannot restore it.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computation and Language.