Metraipolis › archive › dc_7100bf7e7318

Value-Conditioned Proposal Acquisition for Test-Time Reasoning — arXiv:2611.02852

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.02852
License (as archived)
CC0-1.0
Archived
2026-11-03T10:44:08.361Z
SHA-256 of the archived text
a4d3addafbf2c87bfd360d0831e021db907ac2f5531b32e962b2c13bfa655c1f
Stable URL of this page
https://metraipolis.com/documento/dc_7100bf7e7318

How to cite

Value-Conditioned Proposal Acquisition for Test-Time Reasoning — arXiv:2611.02852 — arXiv (en), https://arxiv.org/abs/2611.02852 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7100bf7e7318 (Metraipolis archive dc_7100bf7e7318, sha256 a4d3addafbf2c87bfd360d0831e021db907ac2f5531b32e962b2c13bfa655c1f)
BibTeX
@misc{dc7100bf7e7318,
  title = {Value-Conditioned Proposal Acquisition for Test-Time Reasoning — arXiv:2611.02852},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.02852},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7100bf7e7318; sha256 a4d3addafbf2c87bfd360d0831e021db907ac2f5531b32e962b2c13bfa655c1f},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.02852. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Sampling and selection spend test-time compute inside a candidate pool, but they cannot recover a correct answer that never enters it. We introduce ValueFormation, a test-time policy that decides after each primary-model draw whether to stop, continue sampling, or acquire one proposal from a heterogeneous reasoning specialist. The three actions are compared on one cost-regularized value scale learned from vote-state features; the calibration is frozen across tasks and primary models, and acquired proposals are accepted only through a frozen posterior comparison. On held-out MATH-500, one Llama-fitted calibration improves over the strongest of eight identically evaluated baselines by more than fifteen points for each of three 8B primaries and transfers unchanged to Qwen3-14B. Removing acquisition lowers Llama MATH accuracy by 20.43 points, while same-trace and same-access replays attribute the gain to state-conditioned acquisition rather than specialist access alone. Where primary coverage is high, as on GSM8K, acquisition is rare and gains mainly come from cheaper allocation; on MBPP, a fixed code route attains the top accuracy at the lowest interaction cost. Varying the token price monotonically reduces calls without retraining.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2611.02852 — 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 2026-11-03T10:39:50.319Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).