Metraipolis › archive › dc_fbd951ea3273

LeakScale: Estimating the Causal Effect of Benchmark Exposure — arXiv:2609.27176

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.27176
License (as archived)
CC0-1.0
Archived
2026-10-08T10:04:48.187Z
SHA-256 of the archived text
da20c826d1827a4bdf632dd453af3b58401d16c86c023c3ab29c22667c343961
Stable URL of this page
https://metraipolis.com/documento/dc_fbd951ea3273

How to cite

LeakScale: Estimating the Causal Effect of Benchmark Exposure — arXiv:2609.27176 — arXiv (en), https://arxiv.org/abs/2609.27176 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fbd951ea3273 (Metraipolis archive dc_fbd951ea3273, sha256 da20c826d1827a4bdf632dd453af3b58401d16c86c023c3ab29c22667c343961)
BibTeX
@misc{dcfbd951ea3273,
  title = {LeakScale: Estimating the Causal Effect of Benchmark Exposure — arXiv:2609.27176},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.27176},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fbd951ea3273; sha256 da20c826d1827a4bdf632dd453af3b58401d16c86c023c3ab29c22667c343961},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.27176. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Evidence that evaluation material entered training does not reveal how much it affected evaluation. This distinction leaves a contaminated benchmark score difficult to interpret: provenance can establish contact, but only a counterfactual can quantify the performance attributable to that contact. We present LeakScale, an interventional framework for estimating this missing quantity. LeakScale creates fresh executable tasks that require private, family-specific information absent from and non-derivable from the public task, controls access to that information, and estimates the resulting control-adjusted change in executable accuracy. Across 2,048 unique families, two model families, two executable domains, and 262,144 generations, exposure improves accuracy in every model-by-domain combination, with gains ranging from +7.17 to +27.31 percentage points. These findings separate two empirical questions that are often conflated: whether benchmark contact occurred and how strongly a reported score depends on it. LeakScale makes the latter directly measurable.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-24 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.27176 — 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 2026-10-08T09:11:49.597Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-24?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).