Metraipolis › archive › dc_cbcd1ae71bb0

Does Uniform Discrete Diffusion Need Time? — arXiv:2609.30977

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.30977
License (as archived): CC0-1.0
Archived: 2026-10-08T01:39:37.609Z
SHA-256 of the archived text: 2f906858f6426705de011f531ae8e9a161a95da061facdec6bc45f7eefeacd10
Stable URL of this page: https://metraipolis.com/documento/dc_cbcd1ae71bb0

How to cite

Does Uniform Discrete Diffusion Need Time? — arXiv:2609.30977 — arXiv (en), https://arxiv.org/abs/2609.30977 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cbcd1ae71bb0 (Metraipolis archive dc_cbcd1ae71bb0, sha256 2f906858f6426705de011f531ae8e9a161a95da061facdec6bc45f7eefeacd10)

BibTeX

@misc{dccbcd1ae71bb0,
  title = {Does Uniform Discrete Diffusion Need Time? — arXiv:2609.30977},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.30977},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cbcd1ae71bb0; sha256 2f906858f6426705de011f531ae8e9a161a95da061facdec6bc45f7eefeacd10},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.30977. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Uniform discrete diffusion models (UDMs) commonly use explicit time conditioning, but we find that it can often be unnecessary in practice. In this paper, we first show that the population-optimal UDM predictor generally depends on time: time controls how much the model should trust the observed context. We then show that this dependence can become negligible in finite-data settings relevant to language. When a corrupted training sequence remains much closer to its original clean sequence than to competing training sequences, the empirical-optimal predictor is nearly insensitive to time over most of the diffusion trajectory, where the guarantee weakens toward the high-noise endpoint. Empirically, trained language UDMs exhibit limited time sensitivity over most of the trajectory, while time-agnostic predictors remain competitive with, and often outperform, time-conditioned models across datasets and training objectives. These results challenge the use of explicit time conditioning in UDMs: although the population optimum depends on time, explicitly conditioning on it may often be unnecessary in practice.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-28 in cs.AI).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI); Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.30977 — 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 2026-10-08T01:12:14.513Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-28?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).