Metraipolis › archive › dc_79fa30724012

Schedule optimization for tau-leaping in masked discrete diffusion — arXiv:2609.21960

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.21960
License (as archived)
CC0-1.0
Archived
2026-10-10T01:47:39.179Z
SHA-256 of the archived text
7e1d21ff8dffd6d22d8a541cf6368b85925d8099009e64cdbc937d12cbf8bea2
Stable URL of this page
https://metraipolis.com/documento/dc_79fa30724012

How to cite

Schedule optimization for tau-leaping in masked discrete diffusion — arXiv:2609.21960 — arXiv (en), https://arxiv.org/abs/2609.21960 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_79fa30724012 (Metraipolis archive dc_79fa30724012, sha256 7e1d21ff8dffd6d22d8a541cf6368b85925d8099009e64cdbc937d12cbf8bea2)
BibTeX
@misc{dc79fa30724012,
  title = {Schedule optimization for tau-leaping in masked discrete diffusion — arXiv:2609.21960},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.21960},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_79fa30724012; sha256 7e1d21ff8dffd6d22d8a541cf6368b85925d8099009e64cdbc937d12cbf8bea2},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.21960. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Masked diffusions are popular generative models for discrete distributions. Unlike standard autoregressive sampling, they reveal several coordinates in parallel, approximating each block's joint conditional law by a product of one-coordinate conditionals. The resulting procedure, usually called tau-leaping, reduces computational cost but introduces a factorization error ($\varepsilon_\text{fact}$), even with perfectly learned predictors. We study the resulting tradeoff between generative accuracy and computational cost, focusing on how to choose a denoising schedule to minimize $\varepsilon_\text{fact}$ for a fixed sampling budget. To do so, we establish an exact integral representation of $\varepsilon_\text{fact}$ separating the schedule from the target's dependence structure, summarized by a dependence density $\rho$. This representation yields recursive stationarity equations for optimal schedules and allows us to quantify how estimation errors in $\rho$ affect schedule selection. As the dimension $N$ and sampling budget grow, we characterize the optimal schedule and quantify the cost of random block sizes relative to a deterministic planner. We highlight a fundamental dichotomy: if $\rho$ converges uniformly to a strictly positive continuous profile as $N\to\infty$, schedule optimization can only improve the leading constant of $\varepsilon_\text{fact}$, while if $\rho$ degenerates, schedule optimization can improve the asymptotic order. Examples based on stationary processes and exchangeable mixtures illustrate these regimes.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-21 in cs.LG).
Categorie arXiv: Statistics Theory (math.ST) ; Machine Learning (cs.LG); Machine Learning (stat.ML).
Pagina del lavoro: https://arxiv.org/abs/2609.21960 — 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 2026-10-10T01:14:16.724Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-21?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).