Learning to Collaborate: An Orchestrated-Decentralized Framework for Peer-to-Peer LLM Federation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Singh, Inderjeet, Vissol-Gaudin, Eleonore, Otung, Andikan, Sekiya, Motoyoshi
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911395770007552
author Singh, Inderjeet
Vissol-Gaudin, Eleonore
Otung, Andikan
Sekiya, Motoyoshi
author_facet Singh, Inderjeet
Vissol-Gaudin, Eleonore
Otung, Andikan
Sekiya, Motoyoshi
contents Fine-tuning Large Language Models (LLMs) for specialized domains is constrained by a fundamental challenge: the need for diverse, cross-organizational data conflicts with the principles of data privacy and sovereignty. While Federated Learning (FL) provides a framework for collaboration without raw data exchange, its classic centralized form introduces a single point of failure and remains vulnerable to model inversion attacks. Decentralized FL (DFL) mitigates this risk by removing the central aggregator but typically relies on inefficient, random peer-to-peer (P2P) pairings, forming a collaboration graph that is blind to agent heterogeneity and risks negative transfer. This paper introduces KNEXA-FL, a novel framework for orchestrated decentralization that resolves this trade-off. KNEXA-FL employs a non-aggregating Central Profiler/Matchmaker (CPM) that formulates P2P collaboration as a contextual bandit problem, using a LinUCB algorithm on abstract agent profiles to learn an optimal matchmaking policy. It orchestrates direct knowledge exchange between heterogeneous, PEFT-based LLM agents via secure distillation, without ever accessing the models themselves. Our comprehensive experiments on a challenging code generation task show that KNEXA-FL yields substantial gains, improving Pass@1 by approx. 50% relative to random P2P collaboration. Critically, our orchestrated approach demonstrates stable convergence, in stark contrast to a powerful centralized distillation baseline which suffers from catastrophic performance collapse. Our work establishes adaptive, learning-based orchestration as a foundational principle for building robust and effective decentralized AI ecosystems.
format Preprint
id arxiv_https___arxiv_org_abs_2601_17133
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Learning to Collaborate: An Orchestrated-Decentralized Framework for Peer-to-Peer LLM Federation
Singh, Inderjeet
Vissol-Gaudin, Eleonore
Otung, Andikan
Sekiya, Motoyoshi
Machine Learning
Artificial Intelligence
Cryptography and Security
Distributed, Parallel, and Cluster Computing
Multiagent Systems
68T05
I.2.11; C.2.4; K.4.1
Fine-tuning Large Language Models (LLMs) for specialized domains is constrained by a fundamental challenge: the need for diverse, cross-organizational data conflicts with the principles of data privacy and sovereignty. While Federated Learning (FL) provides a framework for collaboration without raw data exchange, its classic centralized form introduces a single point of failure and remains vulnerable to model inversion attacks. Decentralized FL (DFL) mitigates this risk by removing the central aggregator but typically relies on inefficient, random peer-to-peer (P2P) pairings, forming a collaboration graph that is blind to agent heterogeneity and risks negative transfer. This paper introduces KNEXA-FL, a novel framework for orchestrated decentralization that resolves this trade-off. KNEXA-FL employs a non-aggregating Central Profiler/Matchmaker (CPM) that formulates P2P collaboration as a contextual bandit problem, using a LinUCB algorithm on abstract agent profiles to learn an optimal matchmaking policy. It orchestrates direct knowledge exchange between heterogeneous, PEFT-based LLM agents via secure distillation, without ever accessing the models themselves. Our comprehensive experiments on a challenging code generation task show that KNEXA-FL yields substantial gains, improving Pass@1 by approx. 50% relative to random P2P collaboration. Critically, our orchestrated approach demonstrates stable convergence, in stark contrast to a powerful centralized distillation baseline which suffers from catastrophic performance collapse. Our work establishes adaptive, learning-based orchestration as a foundational principle for building robust and effective decentralized AI ecosystems.
title Learning to Collaborate: An Orchestrated-Decentralized Framework for Peer-to-Peer LLM Federation
topic Machine Learning
Artificial Intelligence
Cryptography and Security
Distributed, Parallel, and Cluster Computing
Multiagent Systems
68T05
I.2.11; C.2.4; K.4.1
url https://arxiv.org/abs/2601.17133