AI RESEARCH
From Flat Facts to Sharp Hallucinations: Detecting Stubborn Errors via Gradient Sensitivity
arXiv CS.LG
•
ArXi:2605.00939v1 Announce Type: new Traditional hallucination detection fails on "Stubborn Hallucinations" -- errors where LLMs are confidently wrong. We propose a geometric solution: Embedding-Perturbed Gradient Sensitivity (EPGS). We hypothesize that while robust facts reside in flat minima, stubborn hallucinations sit in sharp minima, ed by brittle memorization. EPGS detects this sharpness by perturbing input embeddings with Gaussian noise and measuring the resulting spike in gradient magnitude.