Metraipolis › archive › dc_aff9974ea664

Representation Measurements Under Function-Preserving Reparameterizations — arXiv:2608.27020

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.27020
License (as archived)
CC0-1.0
Archived
2026-10-17T16:15:54.488Z
SHA-256 of the archived text
b30a922eae5dcd764b45b42e8dded6b3e6c430bddd0779172b04027ce0d137b5
Stable URL of this page
https://metraipolis.com/documento/dc_aff9974ea664

How to cite

Representation Measurements Under Function-Preserving Reparameterizations — arXiv:2608.27020 — arXiv (en), https://arxiv.org/abs/2608.27020 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_aff9974ea664 (Metraipolis archive dc_aff9974ea664, sha256 b30a922eae5dcd764b45b42e8dded6b3e6c430bddd0779172b04027ce0d137b5)
BibTeX
@misc{dcaff9974ea664,
  title = {Representation Measurements Under Function-Preserving Reparameterizations — arXiv:2608.27020},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.27020},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_aff9974ea664; sha256 b30a922eae5dcd764b45b42e8dded6b3e6c430bddd0779172b04027ce0d137b5},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.27020. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Hidden coordinates are not uniquely determined by a language model's input--output function, so representation-derived measurements should be invariant to function-preserving changes of basis. This study shows that column-permutation parallel analysis violates function-preserving reparameterization invariance because its reference distribution and selected component count can change while the model function and observed covariance spectrum remain fixed. More generally, a data-internal reference procedure cannot simultaneously preserve every coordinate marginal, remain orthogonally equivariant, and remove cross-coordinate covariance. Empirically, across five models, three retrieval domains, and 75 transformations, median component-count disagreement is 0.79 and median fixed-threshold decision disagreement is 0.26. A centering-only control isolates the reference-driven effect, with 1,141 of 1,200 component counts changing despite an unchanged observed spectrum, whereas independent parallel analysis seeds change none of the corresponding decisions. By contrast, orthogonally invariant comparator scores remain numerically stable with similar held-out discrimination. Together, these results show that parallel analysis-derived component counts and decisions can reflect hidden-coordinate choice rather than a well-defined property of the model.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-28 in cs.LG).
Categorie arXiv: Machine Learning (stat.ML) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2608.27020 — 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 2026-10-17T15:34:50.009Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-08-28?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).