Metraipolis › archive › dc_30ba53b0979f
RAPTOR: Role-Aware Private Training for Mixture-of-Experts — arXiv:2609.05770
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.05770
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T12:49:50.200Z
- SHA-256 of the archived text
42ad2fd90194917e731f655b21ae3701eb13a73bcaabc24d7424fcf8c341be1e- Stable URL of this page
https://metraipolis.com/documento/dc_30ba53b0979f
How to cite
RAPTOR: Role-Aware Private Training for Mixture-of-Experts — arXiv:2609.05770 — arXiv (en), https://arxiv.org/abs/2609.05770 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_30ba53b0979f (Metraipolis archive dc_30ba53b0979f, sha256 42ad2fd90194917e731f655b21ae3701eb13a73bcaabc24d7424fcf8c341be1e)
BibTeX
@misc{dc30ba53b0979f,
title = {RAPTOR: Role-Aware Private Training for Mixture-of-Experts — arXiv:2609.05770},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.05770},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_30ba53b0979f; sha256 42ad2fd90194917e731f655b21ae3701eb13a73bcaabc24d7424fcf8c341be1e},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.05770. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Differentially private (DP) fine-tuning methods treat sparse Mixture-of-Experts (MoE) models as a single dense block, ignoring that shared layers see all data while experts only see routed records. We identify and formally characterize three resulting failure modes: global clipping suppresses expert gradients, batch-level normalization dilutes sparse expert updates, and fixed privacy noise degrades signal-to-noise ratio on low-load experts. We introduce RAPTOR - a Role-Aware Private Training framework, which alternates shared and expert optimization and targets each failure directly, using expert-specific clipping and noise together with a public expected-owner denominator and a count-independent update schedule that avoids conditioning on private, realized expert counts. We prove the resulting mechanism satisfies $(\varepsilon,\delta)$-DP: because each record is assigned to exactly one owner expert, per-expert mechanisms within a layer compose in parallel, so updating all $E$ experts costs no more, in privacy terms, than updating one, with shared and expert streams composing sequentially across training. We further derive a bias-variance decomposition of the public-denominator estimator showing its bias grows predictably with routing imbalance, yielding a privacy-free rule for selecting which layer to protect from routing entropy measured on a small public corpus. Experiments on Switch Transformer and OLMoE fine-tuning across GLUE tasks, and on DeepSeek-VL2-Tiny, show consistent gains over standard DP baselines across several privacy levels ($\varepsilon$), with the largest margins typically at the tightest budgets. Code and models are publicly available: this https URL Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-09 in cs.CL). Categorie arXiv: Machine Learning (cs.LG) ; Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2609.05770 — 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 2026-10-16T12:30:25.238Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-09?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).