Metraipolis › archive › dc_de080ff8eaf3

Contract Memory Compiler: Resolve, Then Traverse — arXiv:2609.32658

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.32658
License (as archived): CC0-1.0
Archived: 2026-10-07T20:55:51.241Z
SHA-256 of the archived text: 646458b94d2426805c965a35b91aac3119a4f3fa0298ddf8e8302e071758a812
Stable URL of this page: https://metraipolis.com/documento/dc_de080ff8eaf3

How to cite

Contract Memory Compiler: Resolve, Then Traverse — arXiv:2609.32658 — arXiv (en), https://arxiv.org/abs/2609.32658 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_de080ff8eaf3 (Metraipolis archive dc_de080ff8eaf3, sha256 646458b94d2426805c965a35b91aac3119a4f3fa0298ddf8e8302e071758a812)

BibTeX

@misc{dcde080ff8eaf3,
  title = {Contract Memory Compiler: Resolve, Then Traverse — arXiv:2609.32658},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.32658},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_de080ff8eaf3; sha256 646458b94d2426805c965a35b91aac3119a4f3fa0298ddf8e8302e071758a812},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32658. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

External memory lets language-model agents answer questions about histories too long for the answer model's context window. Updates create a harder problem than retrieving a recent fact: changing one relation can redirect a multi-hop question to records about an entity absent from the question. We study this update-dependent evidence selection problem and introduce the Contract Memory Compiler (CMC). Before seeing a question, CMC uses a language model to identify relations in the history and record where each one was stated. It applies later updates to determine the current relations, follows them from entities named in the question, and passes the corresponding original records to the answer model in one call. Thus the current state determines which evidence is read, rather than merely refreshing values in a previously selected context. To the best of our knowledge, CMC achieves state-of-the-art multi-hop accuracy on FactConsolidation, reaching 78.25% overall and 61.0% at 262K. With the extracted relations and answer model held fixed, selecting evidence before resolving updates reduces multi-hop accuracy to 21.50%. We also introduce MQuAKE-MemStream, a derived dataset of ordered memory streams built from MQuAKE-Remastered counterfactual cases.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.32658 — 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 2026-10-07T20:29:00.908Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).