Created on May 21, 2025
2025
New preprint about extracting hidden knowledge from LLMs with mechanistic interpretability is out!