一作ACL 2026Poster· Natural Language Processing
LLM Beliefs Are in Their Heads
Alessandro Corona Mendozza · Anders Søgaard
摘要
We investigate belief-like representations in decoder-only autoregressive LLMs using linear controlled probes on residual stream activations and single attention heads. Following Herrmann and Levinstein’s (2025) criteria (Accuracy, Use, Coherence, and Uniformity) we find that large models exhibit...