Metraipolis › archive › dc_4018eda69805

EntroPrefill: Renyi-Guided Context Pruning with Conditional Stability Guarantees for Retrieval-Augmented Generation — arXiv:2610.09757

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.09757
License (as archived)
CC0-1.0
Archived
2026-10-08T11:59:10.626Z
SHA-256 of the archived text
da9a1b678becadbe2a48eb9e770bccbb5f0a16a7af734ccf2da89d079a64a175
Stable URL of this page
https://metraipolis.com/documento/dc_4018eda69805

How to cite

EntroPrefill: Renyi-Guided Context Pruning with Conditional Stability Guarantees for Retrieval-Augmented Generation — arXiv:2610.09757 — arXiv (en), https://arxiv.org/abs/2610.09757 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_4018eda69805 (Metraipolis archive dc_4018eda69805, sha256 da9a1b678becadbe2a48eb9e770bccbb5f0a16a7af734ccf2da89d079a64a175)
BibTeX
@misc{dc4018eda69805,
  title = {EntroPrefill: Renyi-Guided Context Pruning with Conditional Stability Guarantees for Retrieval-Augmented Generation — arXiv:2610.09757},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.09757},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_4018eda69805; sha256 da9a1b678becadbe2a48eb9e770bccbb5f0a16a7af734ccf2da89d079a64a175},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.09757. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Mid-prefill pruning can reduce the sequence processed by deeper transformer layers, but attention concentration alone does not certify that discarded context is dispensable. We formulate EntroPrefill as a Renyi-guided proposal mechanism coupled to explicit constraints on discarded attention mass. Sink-isolated, regularized head pooling respects grouped-query attention while exposing a quantitative trade-off between specialization and worst-head coverage. We derive a mixture-to-head deletion envelope, a computable upper bound on feasible token removal, and a finite-sample observer guarantee that remains valid when the pruning layer is selected adaptively. We then establish a conditional transformer perturbation bound with explicit sufficient Lipschitz constants and a first-token decision-margin corollary. A counterexample shows why shallow observations alone cannot imply an unconditional future-output guarantee. The systems analysis distinguishes query-head unions, physical page allocation, and KV-transfer payload, and gives an arithmetic break-even condition for pruning. This manuscript is theoretical in scope: it defines the procedure, its assumptions, and its formal limits, but does not report measured acceleration or task-accuracy preservation. Experiments are reserved for subsequent validation of the assumptions, approximation tightness, and end-to-end resource trade-offs.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Machine Learning (stat.ML).
Pagina del lavoro: https://arxiv.org/abs/2610.09757 — 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 2026-10-08T10:50:37.711Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).