Metraipolis › archive › dc_6a88993497ec

Low-Margin Branch Points in Deterministic LLM Deliberatio — arXiv:2611.00546

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.00546
License (as archived)
CC0-1.0
Archived
2026-11-03T11:01:18.850Z
SHA-256 of the archived text
a60312b4f49433cc94191eb0b6930c27379dd01566cef780507a6b5bb684d950
Stable URL of this page
https://metraipolis.com/documento/dc_6a88993497ec

How to cite

Low-Margin Branch Points in Deterministic LLM Deliberatio — arXiv:2611.00546 — arXiv (en), https://arxiv.org/abs/2611.00546 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_6a88993497ec (Metraipolis archive dc_6a88993497ec, sha256 a60312b4f49433cc94191eb0b6930c27379dd01566cef780507a6b5bb684d950)
BibTeX
@misc{dc6a88993497ec,
  title = {Low-Margin Branch Points in Deterministic LLM Deliberatio — arXiv:2611.00546},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.00546},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_6a88993497ec; sha256 a60312b4f49433cc94191eb0b6930c27379dd01566cef780507a6b5bb684d950},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.00546. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Deterministic multi-agent LLM deliberation can exactly re- play a fixed transcript, yet exact replay alone does not test local resistance to controlled internal interventions. Because each response becomes context for later agents, a changed greedy token can redirect a transcript-mediated execution. We present a white-box audit of this local failure mode in a five-agent committee under greedy decoding. We prospectively select low- and high-margin token positions, require exact no-op re- play, inject norm-relative residual displacements, and measure first transcript divergence within ten generated tokens. Low- margin positions are more branch-sensitive than high-margin controls, while matched margin-reducing gradient directions branch substantially more often than margin-increasing con- trols. We then continue selected realized branches through the committee to distinguish textual persistence from preference and final-decision changes. The contribution is a controlled audit for local branch susceptibility in this specified protocol, not an estimate of deployment frequency, decision correct- ness, or a phenomenon unique to multi-agent systems.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2611.00546 — 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 2026-11-03T10:59:45.721Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).