Metraipolis › archive › dc_f2b29015aee0
Replica Fragmentation and Glassy Dynamics in Parity Learning — arXiv:2610.08503
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.08503
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T16:07:44.898Z
- SHA-256 of the archived text
5765837dae7e17955b3bbe02e4d18658da048cf93b3efd4bca888dd2bbde102d- Stable URL of this page
https://metraipolis.com/documento/dc_f2b29015aee0
How to cite
Replica Fragmentation and Glassy Dynamics in Parity Learning — arXiv:2610.08503 — arXiv (en), https://arxiv.org/abs/2610.08503 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f2b29015aee0 (Metraipolis archive dc_f2b29015aee0, sha256 5765837dae7e17955b3bbe02e4d18658da048cf93b3efd4bca888dd2bbde102d)
BibTeX
@misc{dcf2b29015aee0,
title = {Replica Fragmentation and Glassy Dynamics in Parity Learning — arXiv:2610.08503},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.08503},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f2b29015aee0; sha256 5765837dae7e17955b3bbe02e4d18658da048cf93b3efd4bca888dd2bbde102d},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.08503. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
We study how independently trained Transformer neural networks reconstruct a binary string from its local domain walls. Runs sharing the data and training protocol can realize different functions. We treat them as replicas and measure truth alignment $m$, prediction confidence $q_{\mathrm{self}}$, and cross-replica agreement $q_{\mathrm{cross}}$. Confident disagreement defines the finite-size replica fragmentation that we call glass-like. With small training sets, replicas predict all training examples correctly but remain confident in incorrect predictions for unseen inputs, a regime we call memorization. With larger sets, runs can generalize and then retreat. Retreat occurs when outputs start to deviate from truth while confidence remains high. The frontier between learned and unlearned outputs recedes toward shorter strings. Many later recover as the frontier advances again. The self--cross gap $q_{\mathrm{self}}-q_{\mathrm{cross}}$ clearly distinguishes the three learning regimes of memorization, retreat, and recovery. With overall and position-dependent truth alignment subtracted, their residual correlations also differ: memorizing replicas have weakly and uniformly correlated residuals, while retreat has the largest fraction of replica pairs whose residuals are anti-correlated. That is, on the inputs where one replica of such a pair does better than its average, the other tends to do worse. These finite-size observations distinguish persistent memorization from ongoing retreat--recovery dynamics. The thermodynamic and long-time limits, and extensions to other learning tasks, remain open.
Tema degli esploratori: fisica e astronomia (elenco arXiv cond-mat.stat-mech, nuovi lavori del giorno).
Categorie arXiv: Statistical Mechanics (cond-mat.stat-mech) ; Disordered Systems and Neural Networks (cond-mat.dis-nn); Strongly Correlated Electrons (cond-mat.str-el).
Pagina del lavoro: https://arxiv.org/abs/2610.08503 — 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 2026-10-07T15:40:59.580Z dalla pagina https://arxiv.org/list/cond-mat.stat-mech/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).