Metraipolis › archive › dc_f5264f76ecbe

Cord: Confidence-Anchored Order Residual Decoding for Masked Diffusion Language Models — arXiv:2610.39372

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.39372
License (as archived)
CC0-1.0
Archived
2026-10-31T06:13:12.546Z
SHA-256 of the archived text
5ed6e4e4266981cf1009e5b73bcf642e4284d5112d49b0f156d91274746a5709
Stable URL of this page
https://metraipolis.com/documento/dc_f5264f76ecbe

How to cite

Cord: Confidence-Anchored Order Residual Decoding for Masked Diffusion Language Models — arXiv:2610.39372 — arXiv (en), https://arxiv.org/abs/2610.39372 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f5264f76ecbe (Metraipolis archive dc_f5264f76ecbe, sha256 5ed6e4e4266981cf1009e5b73bcf642e4284d5112d49b0f156d91274746a5709)
BibTeX
@misc{dcf5264f76ecbe,
  title = {Cord: Confidence-Anchored Order Residual Decoding for Masked Diffusion Language Models — arXiv:2610.39372},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.39372},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f5264f76ecbe; sha256 5ed6e4e4266981cf1009e5b73bcf642e4284d5112d49b0f156d91274746a5709},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.39372. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

The decoding order of masked diffusion language models (dLLMs) can substantially affect generation quality, yet common decoders typically apply the same reveal rule throughout generation. Confidence provides a strong default, but it measures how decisively a position can be predicted now, not how useful revealing that position will be for subsequent decoding. We propose Cord, a lightweight prompt-conditioned policy that retains confidence decoding as a stable reference and learns when and how to depart from it through geometric residuals over generation direction, proximity to revealed context, and two-sided infilling. The residuals adapt to both the prompt and the evolving partial response. Rather than sampling a new policy action at every decoding step, Cord samples a single low-dimensional coefficient matrix per rollout, which induces an entire state-dependent reveal trajectory. This yields a contextual-bandit formulation trained directly from sequence-level rewards, without step-level credit assignment or backbone updates. Across three dLLMs and five reasoning, code, and constraint-satisfaction benchmarks, Cord improves average task success from 32.0% under confidence decoding to 48.2%, and further improves over the best per-domain fixed residual. Behavioral audits show that the learned residuals induce substantial, task-dependent changes in reveal locality and two-sided infilling. Cord adds only $0.53$M trainable parameters, requires no additional backbone forward passes, and incurs only 5.39% inference latency overhead.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.39372 — 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 2026-10-31T06:07:08.710Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).