FedTreeLoRA: Reconciling Statistical and Functional Heterogeneity in Federated LoRA Fine-Tuning

ArXi:2603.13282v1 Announce Type: cross Federated Learning (FL) with Low-Rank Adaptation (LoRA) has become a standard for privacy-preserving LLM fine-tuning. However, existing personalized methods predominantly operated under a restrictive Flat-Model Assumption: they addressed client-side \textit{statistical heterogeneity} but treated the model as a monolithic block, ignoring the \textit{functional heterogeneity} across LLM layers.