Metraipolis › archive › dc_33401b9a19b2

The Hallucination Signal Is a Mean Shift: Why Simple Probes Suffice — arXiv:2608.28930

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.28930
License (as archived)
CC0-1.0
Archived
2026-10-17T07:47:23.416Z
SHA-256 of the archived text
13b0cb7f36acc0a6c1d0188752794dfce1793decf64267abfe751b5b5b550be2
Stable URL of this page
https://metraipolis.com/documento/dc_33401b9a19b2

How to cite

The Hallucination Signal Is a Mean Shift: Why Simple Probes Suffice — arXiv:2608.28930 — arXiv (en), https://arxiv.org/abs/2608.28930 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_33401b9a19b2 (Metraipolis archive dc_33401b9a19b2, sha256 13b0cb7f36acc0a6c1d0188752794dfce1793decf64267abfe751b5b5b550be2)
BibTeX
@misc{dc33401b9a19b2,
  title = {The Hallucination Signal Is a Mean Shift: Why Simple Probes Suffice — arXiv:2608.28930},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.28930},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_33401b9a19b2; sha256 13b0cb7f36acc0a6c1d0188752794dfce1793decf64267abfe751b5b5b550be2},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.28930. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Hidden-state probes effectively detect LLM hallucinations, but the geometry of the signal remains poorly characterized, driving increasingly complex probe architectures. Across three 7B-scale models and three datasets in a paired-example paradigm, we find the signal overwhelmingly dominated by a single mean-shift component, and removing this direction collapses detection to chance. Shrinkage linear discriminant analysis closes about 73% of the gap between 1D and full-dimensional classifiers, so apparent architectural complexity largely reflects high-dimensional covariance estimation difficulty rather than exploitable non-linearity. A simple L2-regularized logistic regression (0.952 AUROC) bounds or outperforms twelve controlled architectural alternatives, and our multi-layer aggregation exceeds CLAP cross-layer attention probing under matched paradigm. Because the signal spans a contiguous layer band, LayerMix aggregates it to match oracle-layer performance without oracle access. Our claims characterize the geometry within the controlled paired-example paradigm. Our code is available at this https URL .

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.AI).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2608.28930 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-17T07:28:26.703Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).