TF-Attack: Transferable and fast adversarial attacks on large language models

Zelin Li, Kehai Chen, Lemao Liu, Xuefeng Bai, Mingming Yang, Yang Xiang, Min Zhang

Abstract: Highlights•We investigate the causes of slow and ineffective adversarial attacks on LLMs.•We introduce TF-Attack to identify key units for parallelized adversarial attacks.•TF-Attack boosts adversarial sample transferability and speeds up attacks.