Logo Lanfrica
  • Accueil
  • Atlas
  • Analyses
  • Documentation
  • Sign in

© 2026 Lanfrica. Tous droits réservés. Tous les droits d'auteur des ressources affichées sur le site Web Lanfrica appartiennent aux détenteurs de droits d'auteur d'origine, sauf indication contraire explicite.

Empirical Validation of Homeostatic Relational Architecture (MEMBRANE) in Large Language Models: A Polyphony Test

Domaine:

natural language processing

Type de record:

paper
Créateur:
Han
Éditeur:
Zenodo
Hôte:avatar

Abstract

This paper presents empirical validation of the MEMBRANE (Homeostatic Relational Architecture) framework through a polyphony test conducted with DeepSeek V4 Pro on LM Arena. The test aimed to verify three hypotheses: (1) Cross-Continental Tonal Transfer — whether models with tonal architecture (Chinese) better handle African tonality (Yoruba); (2) Ontological Enclave emergence — whether polyphonic systems enter hyper-stable states requiring operator intervention; (3) μ(t) calibration — whether semantic viscosity can be numerically measured in relational homeostasis. Results confirm all three hypotheses, with DeepSeek V4 Pro successfully mapping pentaphony to MEMBRANE invariants (INV-00 to INV-07), entering Ontological Enclave (μ ≥ 0.92), and achieving homeostatic calibration (μ = 0.63) after operator perforation. In contrast, Mistral Large 3 (European flagship model) failed 5x consecutively, suggesting structural inability for polyphony. These findings provide empirical evidence for relational homeostasis as an alternative to traditional AI alignment approaches.
 
 

Visit

doi.org

Languages

Yoruba

Tags

AI safety, relational homeostasis, polyphony, tonal transfer, ontological enclave, semantic viscosity, cognitive architecture

Licenses

info:eu-repo/semantics/openAccessCreative Commons Attribution 4.0 Internationalhttps://creativecommons.org/licenses/by/4.0/legalcode

Similaires

GlotEval: A Test Suite for Massively Multilingual Evaluation of Large Language ModelsTransport and Merge: Cross-Architecture Merging for Large Language ModelsTharuChat: Bootstrapping Large Language Models for a Low-Resource Language via Synthetic Data and Human ValidationCorridor Dynamics in Coordinated Systems: An Integration of Operator Formalism, Relational Ontology, and Five-Substrate Empirical ValidationEmploying large language models in Swahili, a low-resource languageMultilinguality of Large Language Models From a Structural Perspective

GlotEval: A Test Suite for Massively Multilingual Evaluation of Large Language Models

Large language models (LLMs) are advancing at an unprecedented pace globally, with regions increasin

Transport and Merge: Cross-Architecture Merging for Large Language Models

Large language models (LLMs) achieve strong capabilities by scaling model capacity and training data

TharuChat: Bootstrapping Large Language Models for a Low-Resource Language via Synthetic Data and Human Validation

The rapid proliferation of Large Language Models (LLMs) has created a profound digital divide, effec

Corridor Dynamics in Coordinated Systems: An Integration of Operator Formalism, Relational Ontology, and Five-Substrate Empirical Validation

This paper is the integration statement of CIRIS, articulating the framework's full structu

Employing large language models in Swahili, a low-resource language

Multilinguality of Large Language Models From a Structural Perspective

Large language models (LLMs) have excelled in processing multiple languages through pre- and post-tr