Using LLMs to Uncover Memorization in Instruction-Tuned Models

A study introducing a black-box prompt optimization approach to uncover higher levels of memorization in instruction-tuned LLMs.

October 11, 2024 · 2 min · Chengyu Zhang