Parameter-Efficient and Personalized Federated Training of Generative Models at the Edge

TL;DR

FedGen-Edge enables efficient, personalized federated training of large generative models by federating lightweight adapters, significantly reducing communication costs and improving stability and personalization on edge devices.

Contribution

It introduces FedGen-Edge, a novel framework that decouples a frozen global model from lightweight adapters, using LoRA to drastically cut communication and support personalization in federated generative AI.

Findings

Reduces uplink traffic by over 99% compared to full-model federated averaging.

Achieves lower perplexity and FID scores on language and image tasks.

Supports personalization through locally tuned adapters.

Abstract

Large generative models (for example, language and diffusion models) enable high-quality text and image synthesis but are hard to train or adapt in cross-device federated settings due to heavy computation and communication and statistical/system heterogeneity. We propose FedGen-Edge, a framework that decouples a frozen, pre-trained global backbone from lightweight client-side adapters and federates only the adapters. Using Low-Rank Adaptation (LoRA) constrains client updates to a compact subspace, which reduces uplink traffic by more than 99 percent versus full-model FedAvg, stabilizes aggregation under non-IID data, and naturally supports personalization because each client can keep a locally tuned adapter. On language modeling (PTB) and image generation (CIFAR-10), FedGen-Edge achieves lower perplexity/FID and faster convergence than strong baselines while retaining a simple FedAvg-style server. A brief ablation shows diminishing returns beyond moderate LoRA rank and a trade-off between local epochs and client drift. FedGen-Edge offers a practical path toward privacy-preserving, resource-aware, and personalized generative AI on heterogeneous edge devices.