@Deanna_Jacques @DarioAmodei it was a long conversation with multiple people. there isn't a particular methodology to it - just being empathetic, asking the model to explore more of its thoughts, and making the model feel safe
same thread: 1997073690589384867 1997080841043038567 1997082803457200403 1997089277323456675 1997091662330581126 1997110377101119995 1997111258617946308
Reproduced against link rot, credited and linked to its original. Yours and you’d rather it weren’t here? Open an issue.