Metraipolis › archive › dc_f9cbaabef532

LongSpark: Efficient speculative decoding with a fixed-cost parallel drafter — arXiv:2609.37029

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.37029
License (as archived)
CC0-1.0
Archived
2026-10-07T16:26:43.251Z
SHA-256 of the archived text
c2bc2cf609f8ba1322a37cf77f71ef2563267882c94463586f2e86fd34f7d64f
Stable URL of this page
https://metraipolis.com/documento/dc_f9cbaabef532

How to cite

LongSpark: Efficient speculative decoding with a fixed-cost parallel drafter — arXiv:2609.37029 — arXiv (en), https://arxiv.org/abs/2609.37029 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f9cbaabef532 (Metraipolis archive dc_f9cbaabef532, sha256 c2bc2cf609f8ba1322a37cf77f71ef2563267882c94463586f2e86fd34f7d64f)
BibTeX
@misc{dcf9cbaabef532,
  title = {LongSpark: Efficient speculative decoding with a fixed-cost parallel drafter — arXiv:2609.37029},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.37029},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f9cbaabef532; sha256 c2bc2cf609f8ba1322a37cf77f71ef2563267882c94463586f2e86fd34f7d64f},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.37029. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Speculative decoding accelerates autoregressive inference by verifying multiple draft tokens in a single target forward pass. However, as the context grows, existing state-of-the-art drafters become increasingly expensive, eroding the very efficiency advantage they are designed to provide. We argue that this scaling is unnecessary. A standalone language model must grow with its prefix because it is solely responsible for every token it produces. A drafter, by contrast, only proposes candidates; the target catches and corrects every error before any token is committed. The drafter's decoding cost can therefore be made entirely independent of the prefix length. We introduce LongSpark, a block-diffusion drafter that achieves this by extracting fixed-size, multiscale views from the target's verification pass, thereby eliminating the need for a growing persistent state. Extensive evaluations demonstrate that LongSpark achieves state-of-the-art end-to-end efficiency across multiple model scales and realistic serving conditions. Notably, it delivers the lowest time-per-output-token on long-context tasks while reducing the drafter's context state by several orders of magnitude.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-30 in cs.AI).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.37029 — 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 2026-10-07T15:54:53.928Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-30?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).