Metraipolis › archive › dc_0181fa4ae102

DLoop: Looped Speculative Decoding — arXiv:2610.07659

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.07659
License (as archived)
CC0-1.0
Archived
2026-10-07T11:23:08.887Z
SHA-256 of the archived text
fd725efa6b9778fdfbeb742f5dd9d0cc1f9bb562e4b2f1dcd09a1cab5dba2f88
Stable URL of this page
https://metraipolis.com/documento/dc_0181fa4ae102

How to cite

DLoop: Looped Speculative Decoding — arXiv:2610.07659 — arXiv (en), https://arxiv.org/abs/2610.07659 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0181fa4ae102 (Metraipolis archive dc_0181fa4ae102, sha256 fd725efa6b9778fdfbeb742f5dd9d0cc1f9bb562e4b2f1dcd09a1cab5dba2f88)
BibTeX
@misc{dc0181fa4ae102,
  title = {DLoop: Looped Speculative Decoding — arXiv:2610.07659},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.07659},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0181fa4ae102; sha256 fd725efa6b9778fdfbeb742f5dd9d0cc1f9bb562e4b2f1dcd09a1cab5dba2f88},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.07659. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Speculative decoding accelerates autoregressive generation in large language models. In each drafting stage, a lightweight draft model proposes tokens that the target model subsequently verifies. With increasingly capable draft models, we find that the target model frequently accepts all tokens produced in a drafting stage. A verification nevertheless follows each drafting stage, resulting in unnecessary target-model forward passes even when drafting could have continued. Adaptive draft length methods decide during decoding how many draft tokens precede a verification, but they raise the speedup only for autoregressive draft models. For a parallel draft model, drafting further requires target-model hidden states for draft tokens that have not been verified. We propose DLoop, a looped form of speculative decoding that adaptively performs multiple drafting stages before verification. DLoop continues drafting while the draft model remains confident and verifies all accumulated draft tokens together. Loop-aware training keeps the draft model reliable in the additional drafting stages by exposing it to its own hidden states for unverified draft tokens. By spending additional draft-model forward passes, DLoop reduces the number of target-model forward passes required for verification. Across diverse speculative decoding methods including EAGLE-3, DFlash, Domino, DSpark, and multi-token prediction modules, DLoop improves the wall-clock speedup by 5 to 41 percent while preserving lossless decoding. Code will be available at this https URL .

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/2610.07659 — 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 2026-10-07T11:15:39.425Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).