Metraipolis › archive › dc_fd24473dfb93

LoopCD: Loop-wise Contrastive Decoding for Improving Reasoning in Looped Language Models — arXiv:2609.24196

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.24196
License (as archived)
CC0-1.0
Archived
2026-10-09T06:50:05.344Z
SHA-256 of the archived text
595c1af44d8f373473843503773169c4dd78a654acac29b496a6bfdfe3ea4512
Stable URL of this page
https://metraipolis.com/documento/dc_fd24473dfb93

How to cite

LoopCD: Loop-wise Contrastive Decoding for Improving Reasoning in Looped Language Models — arXiv:2609.24196 — arXiv (en), https://arxiv.org/abs/2609.24196 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fd24473dfb93 (Metraipolis archive dc_fd24473dfb93, sha256 595c1af44d8f373473843503773169c4dd78a654acac29b496a6bfdfe3ea4512)
BibTeX
@misc{dcfd24473dfb93,
  title = {LoopCD: Loop-wise Contrastive Decoding for Improving Reasoning in Looped Language Models — arXiv:2609.24196},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.24196},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fd24473dfb93; sha256 595c1af44d8f373473843503773169c4dd78a654acac29b496a6bfdfe3ea4512},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.24196. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Looped Language Models (LoopLMs) perform "latent reasoning" by recursively refining internal latent representations with shared weights, offering a more effective alternative to explicit verbal reasoning. Despite their effectiveness, we find that LoopLMs remain prone to loop instability: unstable refinement across iterations can produce localized uncertain "hard" tokens associated with reasoning errors. To address this, we propose LoopCD, loop-wise contrastive decoding that enhances the reasoning performance of LoopLMs by intervening on these tokens at inference time. Specifically, we exploit the internal dynamics of LoopLMs and contrast the logits from earlier iterations with logits from the last refined iteration to form the final sampling distribution. We find that this strategy is highly efficient, introducing only negligible inference overhead and requiring no additional training, while effectively improving reasoning performance by naturally refining reasoning-critical hard tokens. Extensive experiments show that our method improves the performance of recent representative LoopLMs across various reasoning tasks.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-22 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.24196 — 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 2026-10-09T06:48:45.036Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-22?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).