Weird Generalization and Inductive Backdoors: New Ways to Corrupt LLMs

Add code
Dec 10, 2025

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: