Metraipolis › archive › dc_e94c0cdae68a

Portable Semantics, Private Dialects: Reuse and Negative Transfer in Latent Communication Between Language-Model Cells — arXiv:2609.11365

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.11365
License (as archived)
CC0-1.0
Archived
2026-10-15T22:41:23.791Z
SHA-256 of the archived text
801d142e86d4958e6326824856ac222ad12e1300f314c4a332e7c4c7bc408d52
Stable URL of this page
https://metraipolis.com/documento/dc_e94c0cdae68a

How to cite

Portable Semantics, Private Dialects: Reuse and Negative Transfer in Latent Communication Between Language-Model Cells — arXiv:2609.11365 — arXiv (en), https://arxiv.org/abs/2609.11365 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e94c0cdae68a (Metraipolis archive dc_e94c0cdae68a, sha256 801d142e86d4958e6326824856ac222ad12e1300f314c4a332e7c4c7bc408d52)
BibTeX
@misc{dce94c0cdae68a,
  title = {Portable Semantics, Private Dialects: Reuse and Negative Transfer in Latent Communication Between Language-Model Cells — arXiv:2609.11365},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.11365},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e94c0cdae68a; sha256 801d142e86d4958e6326824856ac222ad12e1300f314c4a332e7c4c7bc408d52},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.11365. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

In shared-genome language-model societies, restricted evidence visibility favors reusable, value-indexed latent packet interfaces, whereas the sole high-performing globally visible model in the parent study learned an episode-entangled code. This companion study asks whether independently trained societies share one packet language, where strict zero-shot transfer fails, and whether inherited interface state helps or harms later learning. First, a leakage-controlled causal interoperability audit over all 30 ordered pairs of six independently trained restricted societies -- under sealed held-out structure and a preregistered raw/orthogonal/linear/nonlinear alignment ladder -- shows the six semantically similar interfaces do not form one raw language: one same-initialization pair is exactly interoperable in both directions, a second shows asymmetric partial compatibility, and all 26 cross-initialization directions fail every frozen alignment rung. Second, within the tested decomposition and a single sealed source formulation, a source-span control localizes strict zero-shot failure to interpretation and execution of the new operator instructions. Third, in a matched adaptation factorial, the globally trained communication interface acts as a severe negative-transfer prior: reinitializing only the packet reader, writer, and mouth raises final depth-three accuracy from 0.169 to 0.857. Fourth, across two restricted checkpoints and two independently frozen target streams each, inherited interfaces never exceeded fresh-interface controls by the preregistered 0.10 margin. All primary conclusions are bounded to a near-transfer 17-state setting; the negative-transfer factorial concerns one globally visible parent-cohort checkpoint, while an appendix adds a post hoc tagged-global twin case study.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-15 in cs.MA).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Machine Learning (cs.LG); Multiagent Systems (cs.MA).
Pagina del lavoro: https://arxiv.org/abs/2609.11365 — 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 2026-10-15T22:21:04.224Z dalla pagina https://arxiv.org/catchup/cs.MA/2026-09-15?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).