Metraipolis › archive › dc_1cdaf0cd9911

Trainable Boundary-Keeping Where No Formal Answer Exists: A Controlled Knock-In Study of Reward Training in Evidence-Sufficiency Judgment — arXiv:2610.35226

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.35226
License (as archived)
CC0-1.0
Archived
2026-10-27T07:58:30.485Z
SHA-256 of the archived text
26d940f7a6ce79bc1c367ba8af0b63d4e80b8497c1d188104014681397160261
Stable URL of this page
https://metraipolis.com/documento/dc_1cdaf0cd9911

How to cite

Trainable Boundary-Keeping Where No Formal Answer Exists: A Controlled Knock-In Study of Reward Training in Evidence-Sufficiency Judgment — arXiv:2610.35226 — arXiv (en), https://arxiv.org/abs/2610.35226 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_1cdaf0cd9911 (Metraipolis archive dc_1cdaf0cd9911, sha256 26d940f7a6ce79bc1c367ba8af0b63d4e80b8497c1d188104014681397160261)
BibTeX
@misc{dc1cdaf0cd9911,
  title = {Trainable Boundary-Keeping Where No Formal Answer Exists: A Controlled Knock-In Study of Reward Training in Evidence-Sufficiency Judgment — arXiv:2610.35226},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.35226},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_1cdaf0cd9911; sha256 26d940f7a6ce79bc1c367ba8af0b63d4e80b8497c1d188104014681397160261},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.35226. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Reasoning-tuned and reward-trained models are often assumed to verify more reliably because they reason longer or receive correctness signals. We test this in evidence-sufficiency judgment: deciding whether available information actually supports a causal claim, a domain with no decidable correctness reward. Using a pre-registered, same-base, controlled knock-in design, we compare a boundary-keeping recipe (SFT then GRPO) with its own baseline, a scrambled-pairing placebo, and a general-purpose reasoning model under an action-pressure protocol with no evaluation invitation. All load-bearing outcomes are rated by two separate model-rating sessions masked to arm labels; automatic judges only trigger. The trained arm raises spontaneous correct refusal from 25% to 100% (40 per arm; Wilson 95% CI lower bound 91.2%, non-overlapping with the baseline interval [14.2%, 40.2%]), while the scrambled placebo collapses (5% spontaneous / 7.5% invited correct; 91.1% false refusal). The effect is boundary-specific rather than blanket refusal: automatic false refusal on 281 well-formed claims is 1.1% (Wilson upper bound 3.09%), and a targeted 18-item arm-(c) recheck (3 flagged plus 15 accepted) yields 0/18 and 1/18 false refusals in two sessions. The dissociation is the main result: final boundary-keeping is trainable without an inference-time verifier, while observed active numeric/calculation checking in the H3 assay remains near zero; SFT saturation makes the GRPO increment not separately identifiable. A pre-registered mechanism adjudication favors reward internalization of final-action status (H2, best-supported, not proven). The scope is a controlled 14B same-family study: we do not claim proof, cross-family generality, or independent human-expert validation; the rating sessions are model sessions, not human reviewers.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.35226 — 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 2026-10-27T07:55:30.742Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).