Metraipolis › archive › dc_eec74a02e7bd

Completed Pairs Hide Capped Failures: A ReVerPi Case Study of Selective Context Projection — arXiv:2609.31381

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.31381
License (as archived): CC0-1.0
Archived: 2026-10-08T01:38:44.100Z
SHA-256 of the archived text: c94153c073fa06eb3404ec17f272cd7e220a55126f3f0137a1dcc916a9d645b5
Stable URL of this page: https://metraipolis.com/documento/dc_eec74a02e7bd

How to cite

Completed Pairs Hide Capped Failures: A ReVerPi Case Study of Selective Context Projection — arXiv:2609.31381 — arXiv (en), https://arxiv.org/abs/2609.31381 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_eec74a02e7bd (Metraipolis archive dc_eec74a02e7bd, sha256 c94153c073fa06eb3404ec17f272cd7e220a55126f3f0137a1dcc916a9d645b5)

BibTeX

@misc{dceec74a02e7bd,
  title = {Completed Pairs Hide Capped Failures: A ReVerPi Case Study of Selective Context Projection — arXiv:2609.31381},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.31381},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_eec74a02e7bd; sha256 c94153c073fa06eb3404ec17f272cd7e220a55126f3f0137a1dcc916a9d645b5},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.31381. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Context projection replaces older tool observations with compact, addressable excerpts, reducing repeated input while potentially adding evidence-retrieval turns. We study this trade-off in ReVerPi, a Pi extension with archived observations and matched full/projected continuations. In an 86-run source-reading campaign with 641 model requests, the 15 completed pairs show identical success: 12/15 per arm. Twelve further boundary runs stop, with the runner suppressing the companion whenever the first arm fails to complete. Restoring all 27 boundary runs bounds projected-minus-full success between $-$9 and +1 tasks. One omitted, selector-chosen projected continuation successfully retrieves archive text yet exhausts twelve requests; its full counterpart answers in three. The eleven jointly correct pairs form a fully observed success stratum within this recorded frame: projection reduces aggregate logical tokens by 25%, while increasing the median pair's tokens by 29% and total suffix requests from 35 to 55. Separating fitting from evaluation changes the selector's apparent tie: outside its four fitting pairs, it incurs one extra failure and 8.6% more logical tokens over thirteen comparable runs. This methodological case study connects stopping rules, known bounded failures, unexecuted companions, and resource aggregation. Its findings concern the recorded campaign, rather than population noninferiority or superiority over unrestricted Pi. Evaluations should retain every intervention boundary, execute both allocated arms independently of the first arm's completion, and report completion alongside interaction and token expenditure.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-28 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.31381 — 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 2026-10-08T01:12:14.377Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-28?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).