יום שלישי, 15 בספטמבר 2026 LIVE
AI־INFO

כתבה arXiv cs.AI ·

TRACE: Trajectory Correction from Cross-layer Evidence for Hallucination Reduction

תקציר מקורי באנגליתarXiv:2605.18163v2 Announce Type: replace Abstract: Hallucination correction is not a one-direction problem. We show that intermediate layers are neither uniformly more truthful than final layers nor uniformly less trustworthy. Yet hallucination reduction is usually instantiated through one fixed intervention form: contrast one layer against another, steer along a truthfulness direction, or defer to external evidence. This framing is structurally incomplete. Cross-layer factual evidence does not evolve uniformly: in some failures truthful support is present internally and later suppressed, whereas in others candidate competition remains genuinely multi-directional across depth, so no single signed scalar family is generally sufficient. We introduce Trajectory Correction from Cross-layer Ev
קרא במקור המקורי