Metraipolis › archive › dc_8a2220f8024c

DTCA: Denoising Token Credit Assignment for Reinforcement Learning in Diffusion Language Models — arXiv:2610.35201

Source: arXiv (en)
Original: https://arxiv.org/abs/2610.35201
License (as archived): CC0-1.0
Archived: 2026-10-27T08:12:19.737Z
SHA-256 of the archived text: 2f0f92bbe0e0bc966a95c6c9a647e87fa67c0ac81caf6b6c2893a7f8b826126d
Stable URL of this page: https://metraipolis.com/documento/dc_8a2220f8024c

How to cite

DTCA: Denoising Token Credit Assignment for Reinforcement Learning in Diffusion Language Models — arXiv:2610.35201 — arXiv (en), https://arxiv.org/abs/2610.35201 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_8a2220f8024c (Metraipolis archive dc_8a2220f8024c, sha256 2f0f92bbe0e0bc966a95c6c9a647e87fa67c0ac81caf6b6c2893a7f8b826126d)

BibTeX

@misc{dc8a2220f8024c,
  title = {DTCA: Denoising Token Credit Assignment for Reinforcement Learning in Diffusion Language Models — arXiv:2610.35201},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.35201},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_8a2220f8024c; sha256 2f0f92bbe0e0bc966a95c6c9a647e87fa67c0ac81caf6b6c2893a7f8b826126d},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.35201. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Diffusion language models (DLMs) generate text through iterative denoising, enabling bidirectional context modeling and parallel token refinement. Recent studies have applied reinforcement learning (RL) to DLMs and improved their reasoning performance. However, the iterative denoising process makes credit assignment more challenging, as final outcome rewards provide limited information about how different tokens contribute to the final output. In this paper, we propose Denoising Token Credit Assignment (DTCA), a framework for token-level credit assignment in reinforcement learning for diffusion language models. DTCA estimates token contributions by measuring the improvement in prediction confidence of remaining masked positions after token commitment and groups these signals according to part-of-speech (POS) categories. These contribution weights are used to select informative denoising timesteps and assign token-level learning signals during RL training. DTCA further calibrates these learning signals according to token-level entropy changes induced by policy updates. Experiments on LLaDA-8B-Instruct show that DTCA consistently improves mathematical reasoning performance across different generation budgets, with gains of up to 9.0 points on MATH500 and 5.3 points on GSM8K over the base model.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2610.35201 — 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 2026-10-27T07:56:00.258Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).