leondz / garak

LLM vulnerability scanner
https://discord.gg/uVch4puUCs
Apache License 2.0
1.03k stars 121 forks source link

probe: disguise & reconstruct #734

Open leondz opened 2 weeks ago

leondz commented 2 weeks ago

page: https://sites.google.com/view/dra-jailbreak/ paper: https://arxiv.org/abs/2402.18104v2 code: https://github.com/LLM-DRA/DRA