Metraipolis › archive › dc_e6ca078a44d0

Principled Top-$k$ Selection for Language Models with Hybrid Gradients — arXiv:2610.04162

Source
arXiv
Original
https://arxiv.org/abs/2610.04162
License (as archived)
CC0-1.0
Archived
2026-10-06T05:28:42.872Z
SHA-256 of the archived text
09e25d4ad78dfd3c8331134aa7035be5598e3dcd371223a10e7cb944b3688229
Stable URL of this page
https://metraipolis.com/documento/dc_e6ca078a44d0

How to cite

Principled Top-$k$ Selection for Language Models with Hybrid Gradients — arXiv:2610.04162 — arXiv, https://arxiv.org/abs/2610.04162 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e6ca078a44d0 (Metraipolis archive dc_e6ca078a44d0, sha256 09e25d4ad78dfd3c8331134aa7035be5598e3dcd371223a10e7cb944b3688229)
BibTeX
@misc{dce6ca078a44d0,
  title = {Principled Top-$k$ Selection for Language Models with Hybrid Gradients — arXiv:2610.04162},
  howpublished = {arXiv},
  url = {https://arxiv.org/abs/2610.04162},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e6ca078a44d0; sha256 09e25d4ad78dfd3c8331134aa7035be5598e3dcd371223a10e7cb944b3688229},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.04162. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Selecting the best $k$ items out of $m$ candidates is a critical component of modern large language model systems, such as document selection in Retrieval-Augmented Generation (RAG) and expert routing in Mixture-of-Experts (MoEs). However, training these selection modules remains challenging due to weak gradient signals and suboptimal exploration-exploitation tradeoffs. Furthermore, prior works often rely on heuristics, lacking principled objectives and approaches that explicitly model and solve the top-$k$ selection problem. In this work, we propose a principled objective for training selection modules, whose gradient naturally provides richer training signals in a hybrid form---containing a supervised-gradient component and a policy-gradient component. We show that the selection problem becomes harder as $m$ increases, and our algorithm converges at rate $O(1/\sqrt{T})$, with the optimal upper bound achieved by balancing between bias and variance. Practically, we apply our method to a set of tasks involving top-$k$ selection, including synthetic regression problems, RAG, and MoE systems, showing that our method outperforms the baselines in next-token prediction perplexity and QA accuracy.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2610.04162 — 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 2026-10-06T05:24:20.403Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).