Metraipolis › archive › dc_2e465c0cf66d

When Does Longer Reasoning Help? Predicting Mathematical Reasoning Through Discovery and Execution — arXiv:2610.05322

Source
arXiv
Original
https://arxiv.org/abs/2610.05322
License (as archived)
CC0-1.0
Archived
2026-10-06T05:28:49.830Z
SHA-256 of the archived text
cf7040cdd8c26839675d7b52b32879ec7e05873214a18735ca691ce9cc517b27
Stable URL of this page
https://metraipolis.com/documento/dc_2e465c0cf66d

How to cite

When Does Longer Reasoning Help? Predicting Mathematical Reasoning Through Discovery and Execution — arXiv:2610.05322 — arXiv, https://arxiv.org/abs/2610.05322 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2e465c0cf66d (Metraipolis archive dc_2e465c0cf66d, sha256 cf7040cdd8c26839675d7b52b32879ec7e05873214a18735ca691ce9cc517b27)
BibTeX
@misc{dc2e465c0cf66d,
  title = {When Does Longer Reasoning Help? Predicting Mathematical Reasoning Through Discovery and Execution — arXiv:2610.05322},
  howpublished = {arXiv},
  url = {https://arxiv.org/abs/2610.05322},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2e465c0cf66d; sha256 cf7040cdd8c26839675d7b52b32879ec7e05873214a18735ca691ce9cc517b27},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.05322. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Test-time compute can improve mathematical reasoning, but can short-budget runs predict how mathematical reasoning scales with additional compute? We introduce a Discovery--Execution (DE) framework that predicts the aggregate held-out scaling curves through a convolution of strategy discovery and conditional execution. From independent short-budget attempts and oracle-sketch-conditioned runs, the framework estimates cumulative success along held-out reasoning trajectories under alternate compute allocations. We evaluate four models on 35 fresh Olympiad problems and non-geometry problems from IMO-ProofBench Advanced. Under the DE framework, near-saturated execution predicts geometric scaling, as observed for the GPT models. For Claude Opus 4.8, incorporating measured execution substantially improves held-out forecasts over geometric extrapolation across one- and two-arm allocations. As a secondary application, regularized DE (R-DE) decisions to continue or restart yield lower average regret than the best model-specific retrospective policy. Together, these results show that measuring conditional execution provides information about longer reasoning that short-budget success rates do not always capture.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2610.05322 — 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 2026-10-06T05:24:20.434Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).