Metraipolis › archive › dc_7a223d6f2ef1
Replay-Gated Neural Execution: Decoupling Persistent Behavioral Specifications from Neural Realizations in Frozen Language Models — arXiv:2609.22243
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.22243
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-09T06:49:24.878Z
- SHA-256 of the archived text
4c89ec8aefe77dec05a3b81ce08c4f9a19b07b85e8acae51a4ce896504859b1c- Stable URL of this page
https://metraipolis.com/documento/dc_7a223d6f2ef1
How to cite
Replay-Gated Neural Execution: Decoupling Persistent Behavioral Specifications from Neural Realizations in Frozen Language Models — arXiv:2609.22243 — arXiv (en), https://arxiv.org/abs/2609.22243 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7a223d6f2ef1 (Metraipolis archive dc_7a223d6f2ef1, sha256 4c89ec8aefe77dec05a3b81ce08c4f9a19b07b85e8acae51a4ce896504859b1c)
BibTeX
@misc{dc7a223d6f2ef1,
title = {Replay-Gated Neural Execution: Decoupling Persistent Behavioral Specifications from Neural Realizations in Frozen Language Models — arXiv:2609.22243},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.22243},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7a223d6f2ef1; sha256 4c89ec8aefe77dec05a3b81ce08c4f9a19b07b85e8acae51a4ce896504859b1c},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.22243. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Input-conditioned neural interventions raise a runtime question: what persists when one behavioral specification admits multiple actions whose validity depends on execution state? We introduce replay-gated neural execution, separating five objects: a persistent behavioral predicate, its state-indexed certified realization set, a transient action witness, a budget-limited finder, and execution authorization. Candidates undergo isolated FP32/BF16 replay of the frozen model; commitment additionally requires a valid run audit. Experiments on Qwen3-0.6B and SmolLM2-360M-Instruct establish distinct failure modes for these objects. Independent initializations yield distinct certified actions in all 24 tested fixed-state cells. Unchanged SmolLM2 witnesses remain certified in all 128 native states but only 66 of 384 off-diagonal transfers. All 767 archived Qwen witnesses replay successfully, yet a budget-limited finder misses one known-realizable cell in all three prespecified runs. Of 1,141 replay-submitted candidates, 174 fail item certification. A frozen three-tier cascade uses these boundaries to reject uncertified proposals and escalate audit-valid search misses. On 256 previously sealed Qwen Fresh requests, 221 first certify at the lowest-cost tier and all 256 receive audited authorization, with no observed bypass. Relative to frozen full search, the median singleton search-and-certification cost ratio is 0.1055 and P95 is 1.3485, including failed tiers. Within the studied behavioral family on two small models, these results support state-indexed, set-valued execution semantics: specifications persist, search proposes witnesses, and replay certification plus run audit grants execution authority. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-22 in cs.CL). Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2609.22243 — 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 2026-10-09T06:48:44.912Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-22?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).