Metraipolis › archive › dc_a88aa6fed7c7

Discrete Beckmann Transport Models for One-Step Language Modeling and Reasoning — arXiv:2609.15903

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.15903
License (as archived)
CC0-1.0
Archived
2026-10-16T00:12:09.167Z
SHA-256 of the archived text
b88cb7f453f11ee8d0972158edea209a725a9ea317c38eba307b399b62376294
Stable URL of this page
https://metraipolis.com/documento/dc_a88aa6fed7c7

How to cite

Discrete Beckmann Transport Models for One-Step Language Modeling and Reasoning — arXiv:2609.15903 — arXiv (en), https://arxiv.org/abs/2609.15903 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a88aa6fed7c7 (Metraipolis archive dc_a88aa6fed7c7, sha256 b88cb7f453f11ee8d0972158edea209a725a9ea317c38eba307b399b62376294)
BibTeX
@misc{dca88aa6fed7c7,
  title = {Discrete Beckmann Transport Models for One-Step Language Modeling and Reasoning — arXiv:2609.15903},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.15903},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a88aa6fed7c7; sha256 b88cb7f453f11ee8d0972158edea209a725a9ea317c38eba307b399b62376294},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.15903. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Discrete diffusion and flow models are a promising alternative to autoregressive language models, but compressing many-step sampling into fewer steps typically requires distilling a pretrained teacher model. This caps the student at the teacher's quality and requires a costly two-stage training pipeline. We introduce Discrete Beckmann Transport Models (DBTM), built on a time-independent flow whose autonomous transport map provably carries any point in the ambient space to a fixed point on the vertices of the simplex in a single step. We show that this fixed-point property is characterized by a conservation equation whose residual can be minimized directly from data, removing the requirement for a teacher flow and time conditioning. Under this construction, a partially trained map corresponds to the flow truncated at finite time, so generation reduces to iterating one map until it reaches a fixed point. We further extend the map to a partial-context interpolant where additional function evaluations act as refinement steps rather than ODE integration steps. On language modeling and reasoning tasks, DBTM enables one- and few-step generation that improves quality and accuracy over discrete diffusion and continuous flow baselines.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-15 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.15903 — 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 2026-10-15T23:54:14.800Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-15?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).