Metraipolis › archive › dc_0b2f87181fd9

Right, Wrong, or Unfinished: A Competing Risks Framework for Language-Model Reasoning under Token Budgets — arXiv:2611.05860

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.05860
License (as archived)
CC0-1.0
Archived
2026-11-05T18:27:18.505Z
SHA-256 of the archived text
2f572deb96d73cccc5aa81f36589f79800845115e25ce540eff0bdfc3174cb41
Stable URL of this page
https://metraipolis.com/documento/dc_0b2f87181fd9

How to cite

Right, Wrong, or Unfinished: A Competing Risks Framework for Language-Model Reasoning under Token Budgets — arXiv:2611.05860 — arXiv (en), https://arxiv.org/abs/2611.05860 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0b2f87181fd9 (Metraipolis archive dc_0b2f87181fd9, sha256 2f572deb96d73cccc5aa81f36589f79800845115e25ce540eff0bdfc3174cb41)
BibTeX
@misc{dc0b2f87181fd9,
  title = {Right, Wrong, or Unfinished: A Competing Risks Framework for Language-Model Reasoning under Token Budgets — arXiv:2611.05860},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.05860},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0b2f87181fd9; sha256 2f572deb96d73cccc5aa81f36589f79800845115e25ce540eff0bdfc3174cb41},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.05860. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Assessing language-model performance under a token budget requires understanding both when a response will finish and whether it will be correct. We develop Competing Risks Analysis of Token Expenditure in Reasoning (CRATER), a statistically principled framework that treats correct and incorrect completion as distinct competing event types and responses unfinished at the generation limit as administratively right-censored. The fitted distribution provides probabilities of correct completion, incorrect completion, and remaining unfinished across token budgets, together with expected token consumption. We evaluate the proposed method using responses to MATH questions and demonstrate its value in two downstream applications. For routing before generation, CRATER guides the joint choice of a model configuration and token budget. Under hypothetical token prices, it reduces mean cost by $53.39 \%$ relative to a strong reference, with observed correctcompletion rates of $96.42 \%$ and $97.30 \%$, respectively. For repeated sampling, we combine CRATER's single-attempt success probabilities with an existing beta-binomial calibration approach to predict pass@k. Across both applications, CRATER delivers predictions at any specified token budget up to the generation limit from a common model of completion time and correctness, eliminating the need for separate parameter estimation for each budget.

Tema degli esploratori: statistica e scienza dei dati (elenco arXiv stat.ME, nuovi lavori del giorno).
Categorie arXiv: Applications (stat.AP) ; Methodology (stat.ME).
Pagina del lavoro: https://arxiv.org/abs/2611.05860 — 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 2026-11-05T18:16:49.362Z dalla pagina https://arxiv.org/list/stat.ME/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).