Metraipolis › archive › dc_c0fc9be9a45f

Multi-Draft On-Policy Distillation for Diffusion Drafters — arXiv:2610.37778

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.37778
License (as archived)
CC0-1.0
Archived
2026-10-31T06:15:27.751Z
SHA-256 of the archived text
8b05b7042fa8d51774f65464e42c7ade96efe498de5f7e5d7cf4233809d6fee4
Stable URL of this page
https://metraipolis.com/documento/dc_c0fc9be9a45f

How to cite

Multi-Draft On-Policy Distillation for Diffusion Drafters — arXiv:2610.37778 — arXiv (en), https://arxiv.org/abs/2610.37778 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_c0fc9be9a45f (Metraipolis archive dc_c0fc9be9a45f, sha256 8b05b7042fa8d51774f65464e42c7ade96efe498de5f7e5d7cf4233809d6fee4)
BibTeX
@misc{dcc0fc9be9a45f,
  title = {Multi-Draft On-Policy Distillation for Diffusion Drafters — arXiv:2610.37778},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.37778},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_c0fc9be9a45f; sha256 8b05b7042fa8d51774f65464e42c7ade96efe498de5f7e5d7cf4233809d6fee4},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.37778. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Speculative decoding is used to reduce the inference cost of language models. One-step diffusion drafters propose a block of multiple tokens in parallel, reducing the sequential cost of speculative decoding. On-policy distillation trains these drafters using verifier feedback on their own proposals. However, parallel drafting makes this feedback difficult to use: the drafter predicts each token without observing the preceding tokens sampled within the same block, while the verifier conditions its prediction on those tokens. Different sampled prefixes therefore give different verifier targets for the same drafter prediction. We show that distilling against these targets separately can suppress tokens that are plausible after some prefixes but unlikely after others, harming draft acceptance. We explain the effects of this mismatch and introduce Multi-Draft On-Policy Distillation (MDraft), a new parallel drafter distillation method which samples several draft blocks from the same context and aggregates the verifier's full-vocabulary conditionals across their prefixes. We use all the drafts to collect information from the verifier, allowing support from alternative prefixes to influence the training target. The resulting algorithm improves the acceptance length of a parallel drafter beyond the level of pre-training, and alleviates the effects of the distributional mismatch. MDraft changes only training, retaining the drafter's architecture and inference procedure, allowing the resulting drafter to be used as-is in place of the pretrained checkpoint. Experiments on a suite of mathematical, coding, and conversational tasks show longer accepted drafts than prior distillation methods.

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/2610.37778 — 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 2026-10-31T06:07:23.315Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).