Leak@$k$: Unlearning Does Not Make LLMs Forget Under Probabilistic Decoding

Add code
Nov 07, 2025

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: