Metraipolis › archive › dc_1dfeebf704e1
Cross-Prompt Reuse of Entity-Replacement Responses — arXiv:2611.00352
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.00352
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-03T11:04:44.360Z
- SHA-256 of the archived text
73b0adc44b2194467e6f7074f3c22fa0b666f365e022ebf67127f81dc37dcd9a- Stable URL of this page
https://metraipolis.com/documento/dc_1dfeebf704e1
How to cite
Cross-Prompt Reuse of Entity-Replacement Responses — arXiv:2611.00352 — arXiv (en), https://arxiv.org/abs/2611.00352 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_1dfeebf704e1 (Metraipolis archive dc_1dfeebf704e1, sha256 73b0adc44b2194467e6f7074f3c22fa0b666f365e022ebf67127f81dc37dcd9a)
BibTeX
@misc{dc1dfeebf704e1,
title = {Cross-Prompt Reuse of Entity-Replacement Responses — arXiv:2611.00352},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.00352},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_1dfeebf704e1; sha256 73b0adc44b2194467e6f7074f3c22fa0b666f365e022ebf67127f81dc37dcd9a},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.00352. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
With model weights fixed, we measure how much of an entity-replacement response transfers between English factual questions, beyond whether answers change. We add the hidden-state change from two source questions to a target question about the same entity, comparing the predicted log-probability shift between two relations' published answers, given known answer prefixes, with the measured one. Hidden reuse injects the aligned change at 50% depth before the target's remaining layers; direct reuse adds source output-score changes. With other-relation sources, hidden reuse removes 45% and 49% of no-change error on 0.5-1B Qwen and Llama. Gemma-2-2B removes 22% and passes one of three comparisons in plain text without its default beginning-of-sequence (BOS) token. A later protocol restoring it yields 58-68%, passing every comparison on three cohorts. Larger checkpoints, chat format and new entities show replication and format- or cohort-dependent failures. On small plain-format checkpoints, other-relation hidden reuse mainly reproduces the shift both relations' questions share, without reliably lowering relation-selective error. With target-relation sources, hidden reuse's post hoc advantage over direct reuse is 0.2-0.9 nats outside chat format, unconfirmed and not replacement-specific. The same relation-neutral change at both questions' entity tokens lowers relation-selective error more than random or other-replacement changes, indicating receiving-question dependence. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2611.00352 — 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 2026-11-03T10:59:45.938Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).