Metraipolis › archive › dc_7f7ce6af92eb

Architecture-Aware Mixed Precision for Replica-Exchange Training of Geometry-Aware Neural Operators with Uncertainty Quantification — arXiv:2611.02427

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.02427
License (as archived)
CC0-1.0
Archived
2026-11-03T18:37:20.998Z
SHA-256 of the archived text
88903c4ac8cf5297c462b803e2c9785f59884ffb53bd2d63ab40c4ce22c53ad4
Stable URL of this page
https://metraipolis.com/documento/dc_7f7ce6af92eb

How to cite

Architecture-Aware Mixed Precision for Replica-Exchange Training of Geometry-Aware Neural Operators with Uncertainty Quantification — arXiv:2611.02427 — arXiv (en), https://arxiv.org/abs/2611.02427 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7f7ce6af92eb (Metraipolis archive dc_7f7ce6af92eb, sha256 88903c4ac8cf5297c462b803e2c9785f59884ffb53bd2d63ab40c4ce22c53ad4)
BibTeX
@misc{dc7f7ce6af92eb,
  title = {Architecture-Aware Mixed Precision for Replica-Exchange Training of Geometry-Aware Neural Operators with Uncertainty Quantification — arXiv:2611.02427},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.02427},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7f7ce6af92eb; sha256 88903c4ac8cf5297c462b803e2c9785f59884ffb53bd2d63ab40c4ce22c53ad4},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.02427. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Geometry-aware neural operators face substantial computational and memory demands when learning across complex physical domains. We introduce Architecture-Aware Mixed Precision (AA-MP), a replica-exchange training strategy that allocates numerical precision according to replica roles and architecture-specific operations. By retaining FP32 for refinement while selectively employing BF16 for exploration, AA-MP preserves predictive accuracy while improving training efficiency. Across fluid dynamics and three-dimensional structural mechanics benchmarks, AA-MP achieves comparable accuracy to full-precision training over an ensemble of runs, while reducing hot-replica peak GPU memory by 46.9% and training time by up to 25.8%. It also outperforms naive Automatic Mixed Precision (AMP) in memory efficiency and computational cost, with increasing memory advantages at larger model and geometry scales. Ablation studies show that retaining FP32 in the cold-replica is important for predictive accuracy, while additional FP32 operations in the exploratory replica provide limited benefits relative to their computational cost. Finally, conformal prediction for uncertainty calibration shows that AA-MP achieves near-nominal coverage with narrower prediction intervals than the alternatives. These results advance the scalability of geometry-aware scientific machine learning by reducing computational costs without compromising predictive accuracy or uncertainty quantification.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2611.02427 — 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 2026-11-03T18:12:54.569Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).