knowingly-wrong

Ariana M. Villegas-Suarez, Guido Freire, Sohan Venkatesh, Bartosz Cywiński

Adapters and datasets developed during the LASR Labs fellowship to reproduce experiments and figures from Knowingly Wrong: Training LLMs Against Their Own Beliefs can Cause Misalignment.

Structure

wrongness/

persona-vectors/

realistic-scenarios/