Metraipolis › archive › dc_77db44c06c4b
Specification Oracles — arXiv:2609.13415
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.13415
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T00:10:18.594Z
- SHA-256 of the archived text
d3587ba8d1add0998f2caab8f10b549ecb6286e1f5e90972e1f5e759ff3c4439- Stable URL of this page
https://metraipolis.com/documento/dc_77db44c06c4b
How to cite
Specification Oracles — arXiv:2609.13415 — arXiv (en), https://arxiv.org/abs/2609.13415 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_77db44c06c4b (Metraipolis archive dc_77db44c06c4b, sha256 d3587ba8d1add0998f2caab8f10b549ecb6286e1f5e90972e1f5e759ff3c4439)
BibTeX
@misc{dc77db44c06c4b,
title = {Specification Oracles — arXiv:2609.13415},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.13415},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_77db44c06c4b; sha256 d3587ba8d1add0998f2caab8f10b549ecb6286e1f5e90972e1f5e759ff3c4439},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.13415. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Specifications face a basic tradeoff: leave details out, and important questions go unanswered; record every detail separately, and the specification becomes large and prolix. We investigate whether a language model can serve as a compact, living specification oracle by learning facts about a target and answering questions about it directly. We compare two ways of storing the learned facts: external text notes and changes to the model's weights. Across four families of 596-fact worlds and two Qwen2.5 model sizes, weight-only oracles benefited substantially more from structure: with the 7B model, their accuracy integrated across storage capacities was 18.5 percentage points higher on structured than unstructured worlds, compared with 1.1 points for note-sheet oracles. This advantage came at a substantial storage cost, with the smallest adapter requiring approximately 175 KiB compared with a maximum note budget of 16 KiB. Adapted weights therefore exploited latent structure more successfully, while external notes required substantially less object-specific storage. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-15 in cs.LG). Categorie arXiv: Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2609.13415 — 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 2026-10-15T23:54:14.608Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-15?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).