BLOOM-WILT: Logit Tilting for Behaviour Elicitation in Automated LLM Auditing

Users of a deployed language model routinely encounter behaviours that testing almost never surfaces, since deployment puts the model through orders of magnitude more interactions than any evaluation can simulate.

Users of a deployed language model routinely encounter behaviours that testing almost never surfaces, since deployment puts the model through orders of magnitude more interactions than any evaluation can simulate. Automated auditors make testing cheap to scale and flexible…