Metraipolis › archive › dc_68fa09fbc469

What Limits Recursive Reasoning Models: Optimization, Architecture and Test-Time Scaling — arXiv:2609.39967

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.39967
License (as archived): CC0-1.0
Archived: 2026-10-07T12:45:59.768Z
SHA-256 of the archived text: 98c2892510d832429da702bd7b2d65667b79d6a40429a01935f0b6430302f1a8
Stable URL of this page: https://metraipolis.com/documento/dc_68fa09fbc469

How to cite

What Limits Recursive Reasoning Models: Optimization, Architecture and Test-Time Scaling — arXiv:2609.39967 — arXiv (en), https://arxiv.org/abs/2609.39967 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_68fa09fbc469 (Metraipolis archive dc_68fa09fbc469, sha256 98c2892510d832429da702bd7b2d65667b79d6a40429a01935f0b6430302f1a8)

BibTeX

@misc{dc68fa09fbc469,
  title = {What Limits Recursive Reasoning Models: Optimization, Architecture and Test-Time Scaling — arXiv:2609.39967},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.39967},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_68fa09fbc469; sha256 98c2892510d832429da702bd7b2d65667b79d6a40429a01935f0b6430302f1a8},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.39967. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Recursive reasoning models apply a small shared Transformer block many times to refine a latent state. This gives them large effective depth with few parameters and makes them strong on algorithmic tasks. Such compact solvers are natural candidates for tools that an LLM can call on narrow algorithmic subproblems. However, existing models such as HRM, TRM and URM differ in architecture, gradient propagation and training procedure simultaneously. This makes it hard to tell what drives their performance, and their optimization is still poorly understood and often unstable. In this work we address both of these gaps. First, we study these questions under a unified experimental pipeline spanning six algorithmic domains. Individual controlled ablations are performed on representative domains, while the resulting recipe is evaluated across the full suite. The study reveals a surprisingly simple recipe for stable and generalizable recursive reasoning: an intermediate gradient horizon, large physical batches and controlled updates of the recurrent state. An explicit hierarchical architecture is not needed. Second, we combine these findings into a stable 13.6M-parameter model that achieves the strongest overall performance among the evaluated recursive baselines, with particularly large gains on out-of-distribution generalization. It raises Arithmetic OOD accuracy to 71.2%, from 36.2% for the strongest baseline, while reaching 98.41% on Sudoku and 59.5% pass@2 on ARC-AGI-1. Our results show that, within the recursive architectures studied here, performance depends strongly on how recurrence is optimized and stabilized. More broadly, it shows how AI systems can be improved by optimizing their components one at a time.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.AI).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.39967 — 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 2026-10-07T12:14:14.189Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).