Metraipolis › archive › dc_f80d7692ab86
Thinking with Looped Flows — arXiv:2609.11801
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.11801
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T05:27:13.649Z
- SHA-256 of the archived text
781a6cc439ef921270472f0969fde4476d8adbccb767c0fa6d81c2a52057c5b5- Stable URL of this page
https://metraipolis.com/documento/dc_f80d7692ab86
How to cite
Thinking with Looped Flows — arXiv:2609.11801 — arXiv (en), https://arxiv.org/abs/2609.11801 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f80d7692ab86 (Metraipolis archive dc_f80d7692ab86, sha256 781a6cc439ef921270472f0969fde4476d8adbccb767c0fa6d81c2a52057c5b5)
BibTeX
@misc{dcf80d7692ab86,
title = {Thinking with Looped Flows — arXiv:2609.11801},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.11801},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f80d7692ab86; sha256 781a6cc439ef921270472f0969fde4476d8adbccb767c0fa6d81c2a52057c5b5},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.11801. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Humans and machines often solve harder problems by spending more time on computation. In deep learning, looped models implement this idea during inference by recurrently updating a hidden state. In practice, however, their training backpropagates through only one or a few updates, making it hard to train early updates to support future ones. We propose looped flows, an approach that sidesteps this issue by training the recurrence with local denoising objectives. By imposing temporal association across denoising objectives through progressively decreasing noise levels and shared noise, the model is incentivized to learn recurrent states that transfer useful computation over time, even when gradients cover only a few updates. We then formulate inference as integrating the velocity of a probability flow parameterized by the learned denoiser, coupled with recurrent states. This allows solving harder problems by spending more computation through a finer temporal grid and enables multiple valid predictions from different initial noise samples. Across six reasoning benchmarks including two multi-solution benchmarks, looped flows outperform prior state-of-the-art looped models overall, achieving 58.8% test accuracy on ARC-AGI-1 and 12.2% on ARC-AGI-2. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-11 in cs.AI). Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2609.11801 — 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 2026-10-16T05:17:52.198Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-11?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).