Representational Curvature Modulates Behavioral Uncertainty in Large Language Models

ArXi:2604.23985v1 Announce Type: cross In autoregressive large language models (LLMs), temporal straightening offers an account of how the next-token prediction objective shapes representations. Models learn to progressively straighten the representational trajectory of input sequences across layers, potentially facilitating next-token prediction via linear extrapolation. However, a direct link between this trajectory and token-level behavior has been missing.