3 accepted papers
Detecting AI risks becomes more challenging as stronger models emerge and find novel methods such as Alignment Faking to circumvent these detection attempts. Inspired by how risky behaviors in humans (i.e., illegal activities that may hurt others) are sometimes guided by strongly-held values, we bel…
Large Language Models (LLMs) are often used as automated judges to evaluate text, but their effectiveness can be hindered by various unintentional biases. We propose using linear classifying probes, trained by leveraging differences between contrasting pairs of prompts, to directly access LLMs’ late…
The deployment of language models brings challenges in generating reliable text, especially when these models are fine-tuned with human preferences. To extract the encoded knowledge in these models without (potentially) biased human labels, unsupervised probing techniques like Contrast-Consistent Se…