Metraipolis › archive › dc_dd38ce101373

Evaluating Exact Output and Checkpoint-State Prediction in Real Programs — arXiv:2610.11889

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.11889
License (as archived)
CC0-1.0
Archived
2026-10-09T06:03:18.613Z
SHA-256 of the archived text
92dfcce640b1f33a3b872fc622822d32b761e8ed6ecc3fa734a148ef05edd28e
Stable URL of this page
https://metraipolis.com/documento/dc_dd38ce101373

How to cite

Evaluating Exact Output and Checkpoint-State Prediction in Real Programs — arXiv:2610.11889 — arXiv (en), https://arxiv.org/abs/2610.11889 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_dd38ce101373 (Metraipolis archive dc_dd38ce101373, sha256 92dfcce640b1f33a3b872fc622822d32b761e8ed6ecc3fa734a148ef05edd28e)
BibTeX
@misc{dcdd38ce101373,
  title = {Evaluating Exact Output and Checkpoint-State Prediction in Real Programs — arXiv:2610.11889},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.11889},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_dd38ce101373; sha256 92dfcce640b1f33a3b872fc622822d32b761e8ed6ecc3fa734a148ef05edd28e},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.11889. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

We present a benchmark for predicting final output and checkpoint state from source and input alone. It extends CRUXEval-style output prediction with paired shorter- and longer-trace inputs and checkpoints inside and after a loop. The benchmark contains 400 cases from 371 Python and C++ programs, evaluated under seven settings from four model families without tools or code execution. Of 11,200 planned predictions, 11,151 produced gradable responses. Reasoning-enabled settings outperform their off counterparts by 33.1 to 55.2 percentage points on completed responses. The strongest setting scores 93.0% on shorter-trace final output, 77.0% on longer-trace final output, and 65.5% and 63.5% on the two state tasks; these scores also hold when missing responses count as wrong. Across 2,397 matched Python comparisons with identical source, changing to the longer-trace input yields 528 correct-to-wrong changes and 147 reversals. Source-clustered analyses preserve this accuracy gap, while adjusted Python models give no evidence of a positive incremental association between cumulative state load and error. Changed inputs and checkpoint tasks alter several factors together, so the gaps do not isolate trace length or an internal state-tracking mechanism. The benchmark exposes errors hidden by short-output scores alone.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Software Engineering (cs.SE) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.11889 — 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 2026-10-09T05:47:01.910Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).