Metraipolis › archive › dc_25bee4f0716d
Learning When to Commit: Schedule Distillation for Fast Diffusion Language Models — arXiv:2610.35804
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.35804
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-27T08:03:27.746Z
- SHA-256 of the archived text
81aa689c29662847a57d3e7bcac4ad380306a750a9514399b4350ac96dd7921f- Stable URL of this page
https://metraipolis.com/documento/dc_25bee4f0716d
How to cite
Learning When to Commit: Schedule Distillation for Fast Diffusion Language Models — arXiv:2610.35804 — arXiv (en), https://arxiv.org/abs/2610.35804 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_25bee4f0716d (Metraipolis archive dc_25bee4f0716d, sha256 81aa689c29662847a57d3e7bcac4ad380306a750a9514399b4350ac96dd7921f)
BibTeX
@misc{dc25bee4f0716d,
title = {Learning When to Commit: Schedule Distillation for Fast Diffusion Language Models — arXiv:2610.35804},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.35804},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_25bee4f0716d; sha256 81aa689c29662847a57d3e7bcac4ad380306a750a9514399b4350ac96dd7921f},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.35804. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Diffusion large language models (dLLMs) can generate multiple tokens in parallel, but often require many decoding rounds to complete a response. Efficient generation depends not only on which tokens are easy to predict, but also on how committing to them helps predict the remaining positions. A highly predictable token may provide little useful context, while revealing a less predictable token can help several other positions become predictable sooner. We propose dSpark, a new schedule distillation-based framework for learning when to commit tokens for faster diffusion language generation. To achieve this, we construct unmasking schedules that prioritize tokens whose reveal makes other positions easier to predict, reducing uncertainty earlier in decoding. We then distill these schedules into the model to enable faster parallel sampling, with a KL-divergence regularizer to help preserve generation quality. We extensively evaluate our approach across mathematical reasoning and code generation tasks. On GSM8K, our method achieves remarkable 10.12$\times$ and 5.12$\times$ speedups over the various baselines on LLaDA-8B-Instruct and Dream-7B-Instruct, respectively, while maintaining superior task performance. Our code is available at this https URL . 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); Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2610.35804 — 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 2026-10-27T07:55:31.018Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).