arXiv - CSCL: "Analyzing the Inherent Response Tendency of LLMs:…" - Qoto Mastodon

arXiv - CSCL @arxiv_cscl@qoto.org

Analyzing the Inherent Response Tendency of LLMs: Real-World Instructions-Driven Jailbreak. (arXiv:2312.04127v1 [cs.CL])

http://arxiv.org/abs/2312.04127 #arXiv #NLProc

Dec 10, 2023, 03:19 · · arxiv-cscl · · ·

Sign in to participate in the conversation