Metraipolis › archive › dc_10398233eccf

RangeSWA: Training-Free Length Extrapolation for Continuous Diffusion Language Models — arXiv:2610.42084

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.42084
License (as archived)
CC0-1.0
Archived
2026-11-02T07:03:54.784Z
SHA-256 of the archived text
a6262880395bb15ae238359ffe0e29e58e268d89bc409da0d1ac75255ab0827a
Stable URL of this page
https://metraipolis.com/documento/dc_10398233eccf

How to cite

RangeSWA: Training-Free Length Extrapolation for Continuous Diffusion Language Models — arXiv:2610.42084 — arXiv (en), https://arxiv.org/abs/2610.42084 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_10398233eccf (Metraipolis archive dc_10398233eccf, sha256 a6262880395bb15ae238359ffe0e29e58e268d89bc409da0d1ac75255ab0827a)
BibTeX
@misc{dc10398233eccf,
  title = {RangeSWA: Training-Free Length Extrapolation for Continuous Diffusion Language Models — arXiv:2610.42084},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.42084},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_10398233eccf; sha256 a6262880395bb15ae238359ffe0e29e58e268d89bc409da0d1ac75255ab0827a},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.42084. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Continuous diffusion language models show promising potential in text generation tasks, but are prone to problems such as text repetition, semantic drift, and generation quality degradation when the generation length exceeds the training length. To address these challenges, we propose RangeSWA, a training-free length extrapolation method. The method combines local sliding-window attention, remote block summaries, and dynamic content retrieval to preserve local details and support long-range context access. In addition, generation degradation caused by length extrapolation is mitigated through positional remapping, conditioning attention correction, and denoising-stage-adaptive window scheduling. We conducted experiments on multiple continuous diffusion language models, and the results show that RangeSWA can effectively reduce generative perplexity and text repetition rates in long-text generation.

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/2610.42084 — 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 2026-11-02T06:57:13.235Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).