"Best-of-Venom: Attacking RLHF by Injecting Poisoned Preference Data."

Tim Baumgärtner et al. (2024)

Details and statistics

DOI: 10.48550/ARXIV.2404.05530

access: open

type: Informal or Other Publication

metadata version: 2024-05-15

a service of  Schloss Dagstuhl - Leibniz Center for Informatics