Metraipolis › archive › dc_73e4052fa18f

Mentored Decoding: Faster Inference meets Boosting — arXiv:2609.30474

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.30474
License (as archived)
CC0-1.0
Archived
2026-10-08T02:10:23.279Z
SHA-256 of the archived text
1e0ba5241edc64576194b8e3964b1c7f22b49eb608eca138370b9dd6ff4ac737
Stable URL of this page
https://metraipolis.com/documento/dc_73e4052fa18f

How to cite

Mentored Decoding: Faster Inference meets Boosting — arXiv:2609.30474 — arXiv (en), https://arxiv.org/abs/2609.30474 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_73e4052fa18f (Metraipolis archive dc_73e4052fa18f, sha256 1e0ba5241edc64576194b8e3964b1c7f22b49eb608eca138370b9dd6ff4ac737)
BibTeX
@misc{dc73e4052fa18f,
  title = {Mentored Decoding: Faster Inference meets Boosting — arXiv:2609.30474},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.30474},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_73e4052fa18f; sha256 1e0ba5241edc64576194b8e3964b1c7f22b49eb608eca138370b9dd6ff4ac737},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.30474. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Speculative decoding is a successful technique speeding up inference of a target autoregressive language model via a fast drafter model. Lossy speculative decoding allows a drift with respect to the target to further improve speed. Interestingly, it has been observed experimentally that the resulting model can $\textit{also}$ beat the target $\textit{quality-wise}$. Our paper formally proves how such a feat is possible with a formal approach to lossy speculative decoding called $\textit{mentored decoding}$. To get there, we connect inference to a celebrated ML training theory, $\textit{boosting}$, and proceed via the generalization of mentored decoding to the whole set of $f$-divergences. We uncover key properties of mentored decoding, among which (i) the particularly appealing geometric nature of the total variation case, (ii) simple approximations for any $f$-divergence in direct relation with boosting compliance, and (iii) a $\textit{divergence independent}$ $O(n)$ space and $O(\mathrm{sort}(n))$ time data structure built on drafter and target outputs, which allows to query the optimal parameters of the dual problem in $O(\log n)$ time and constructing optimal mentored distributions in $O(n)$ time for any $f$-divergence.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-28 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.30474 — 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 2026-10-08T01:43:06.669Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-28?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).