ASTPrompter: Preference-Aligned Automated Language Model Red-Teaming to Generate Low-Perplexity Unsafe Prompts.
Amelia Hardy, Houjun Liu, Allie Griffith, Bernard Lange, Duncan Eddy, Mykel J. Kochenderfer
Browse the full EMNLP paper archive.
Amelia Hardy, Houjun Liu, Allie Griffith, Bernard Lange, Duncan Eddy, Mykel J. Kochenderfer
Browse the full EMNLP paper archive.