Metraipolis › archive › dc_6eba658fda10
Efficient Reasoning via Constrained Optimization in Latent Space — arXiv:2609.34181
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.34181
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T20:57:31.346Z
- SHA-256 of the archived text
927ec77bc4888ea2b0a7961fe1161cb5e83657faaacc855a3e87fd85ada0406b- Stable URL of this page
https://metraipolis.com/documento/dc_6eba658fda10
How to cite
Efficient Reasoning via Constrained Optimization in Latent Space — arXiv:2609.34181 — arXiv (en), https://arxiv.org/abs/2609.34181 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_6eba658fda10 (Metraipolis archive dc_6eba658fda10, sha256 927ec77bc4888ea2b0a7961fe1161cb5e83657faaacc855a3e87fd85ada0406b)
BibTeX
@misc{dc6eba658fda10,
title = {Efficient Reasoning via Constrained Optimization in Latent Space — arXiv:2609.34181},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.34181},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_6eba658fda10; sha256 927ec77bc4888ea2b0a7961fe1161cb5e83657faaacc855a3e87fd85ada0406b},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.34181. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Large Reasoning Models (LRMs) have shown remarkable reasoning capabilities, yet they still suffer from overthinking, generating redundant reasoning steps which incur substantial token consumption. Existing methods, such as suppressing reflective keywords or forcing shorter reasoning lengths, attempt to mitigate this issue but inevitably truncate necessary steps and induce underthinking, thereby compromising performance. To address this dilemma, we investigate the latent representations and observe that efficient reasoning steps naturally cluster into a concentrated region in latent space, while those deviating from this region tend to produce verbose sequences. To leverage this, we keep reasoning focused within this region via a quadratic program which projects deviating hidden states back into the region. Then we propose a novel training-free framework to achieve efficient reasoning that reduces token generation costs without sacrificing performance. Extensive experiments conducted on four models ranging from 1.5B to 14B, and across six benchmarks in math reasoning, coding, and scientific QA, validate the effectiveness of our method, up to a 12.1\% improvement in accuracy while reducing generated tokens by 11.8\% to 52.8\%. Codes are available at \href{ this https URL }{ this https URL }.
Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Optimization and Control (math.OC).
Pagina del lavoro: https://arxiv.org/abs/2609.34181 — 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 2026-10-07T20:29:01.470Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).