Metraipolis › archive › dc_b2bfdd838595

Think Shallow, Solve Deep: Controlling Recurrent Dynamics for Reliable Test-Time Depth — arXiv:2608.18222

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.18222
License (as archived)
CC0-1.0
Archived
2026-10-18T13:53:09.229Z
SHA-256 of the archived text
199576f2b9e1417cf19ea759dc96b6ec7a8ecb63e15e8ccfa0c227238be351c9
Stable URL of this page
https://metraipolis.com/documento/dc_b2bfdd838595

How to cite

Think Shallow, Solve Deep: Controlling Recurrent Dynamics for Reliable Test-Time Depth — arXiv:2608.18222 — arXiv (en), https://arxiv.org/abs/2608.18222 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_b2bfdd838595 (Metraipolis archive dc_b2bfdd838595, sha256 199576f2b9e1417cf19ea759dc96b6ec7a8ecb63e15e8ccfa0c227238be351c9)
BibTeX
@misc{dcb2bfdd838595,
  title = {Think Shallow, Solve Deep: Controlling Recurrent Dynamics for Reliable Test-Time Depth — arXiv:2608.18222},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.18222},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_b2bfdd838595; sha256 199576f2b9e1417cf19ea759dc96b6ec7a8ecb63e15e8ccfa0c227238be351c9},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.18222. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Recurrent-depth reasoners aim to solve harder problems by iterating their update longer at test time, but additional iterations can improve, preserve, or degrade an answer. We show that a measurable property of the trained operator, its finite-time dynamical regime (estimated as settling, marginal, or drifting), indicates which of these occurs. We give a sufficient condition for depth-safety: once an operator's per-step displacement is small relative to the decoder margin, the decoded answer cannot change under further iterations. Empirically, on algorithmic tasks trained from $800$ unaugmented examples per difficulty tier, settling operators do not degrade with added depth, and on some tasks convert it into higher accuracy on harder unseen instances (Sudoku, $0.19$ to $0.34$ past the training horizon). A single terminal fixed-point objective moves the regime and the depth behavior together: removing it induces drift and removes the gains, and adding it to a generic recurrence yields depth-safe extrapolation on carry propagation. We give four operational criteria for useful test-time depth, use them to catalogue failure modes, and, as a consistency check, apply the same measurements to Huginn-3.5B, which falls in the non-settling family.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-20 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG) ; Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2608.18222 — 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 2026-10-18T13:50:58.891Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-08-20?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).