Lennart Buerger

Aug 22, 2024 | lesswrong.com | Winnie Yang |Jojo Yang |Lennart Buerger

As models grow increasingly sophisticated, they will surpass human expertise. It is a fundamentally difficult challenge to make sure that those models are robustly aligned (Bowman et al., 2022). For example, we might hope to reliably know whether a model is being deceptive in order to achieve an instrumental goal. Importantly, deceptive alignment and robust alignment are behaviorally indistinguishable (Hubinger et al., 2024).

Jul 19, 2024 | lesswrong.com | Lennart Buerger

A short summary of the paper is presented below. TL;DR: We develop a robust method to detect when an LLM is lying based on the internal model activations, making the following contributions: (i) We demonstrate the existence of a two-dimensional subspace, along which the activation vectors of true and false statements can be separated. Notably, this finding is universal and holds for various LLMs, including Gemma-7B, LLaMA2-13B and LLaMA3-8B.