Metraipolis › archive › dc_7a45077e7c2d

Recipes for Steering and Scaling LLMs via Sampling — arXiv:2608.26120

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.26120
License (as archived)
CC0-1.0
Archived
2026-10-17T15:18:11.326Z
SHA-256 of the archived text
421d13152934300a9fb0dc844a7655ec09929f9c4aa844631fdc6728d5ea54fc
Stable URL of this page
https://metraipolis.com/documento/dc_7a45077e7c2d

How to cite

Recipes for Steering and Scaling LLMs via Sampling — arXiv:2608.26120 — arXiv (en), https://arxiv.org/abs/2608.26120 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7a45077e7c2d (Metraipolis archive dc_7a45077e7c2d, sha256 421d13152934300a9fb0dc844a7655ec09929f9c4aa844631fdc6728d5ea54fc)
BibTeX
@misc{dc7a45077e7c2d,
  title = {Recipes for Steering and Scaling LLMs via Sampling — arXiv:2608.26120},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.26120},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7a45077e7c2d; sha256 421d13152934300a9fb0dc844a7655ec09929f9c4aa844631fdc6728d5ea54fc},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.26120. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large Language Models (LLMs) are probabilistic models, typically defined by an autoregressive factorization. While recent work has begun to study richer target distributions beyond the base model, the sampling strategies remain highly inefficient. In this paper, we present a flexible and theoretically grounded framework for steering and scaling autoregressive LLMs with sampling. Within this framework, we describe two algorithms -- one based on Sequential Monte Carlo (SMC) and one based on Replica Exchange (RE) -- that steer generation toward powering, product or tilting of the base model distribution. We illustrate this framework through scaling the generation quality of LLMs without external supervision or reward models. Experimental results demonstrate our methods scale more favorably than Best-of-N and standard MCMC baselines. Overall, this paper offers a systematic recipe for probabilistic inference with LLMs via sampling.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-28 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2608.26120 — 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 2026-10-17T15:15:18.011Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-28?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).