Preprint
Aug 2026
FedPA-LoRA: Product-Aligned Framework for Mitigating Aggregation and Initialization Errors in Heterogeneous Federated LoRA
FedPA-LoRA is proposed, a product-aligned federated LoRA framework that jointly addresses limitations and provably converges under both homogeneous and heterogeneous client ranks, with up to a $6.82$ percentage-point improvement in average GLUE accuracy under heterogeneous client ranks.
Juseok Jeon, Ramy E. Ali, Doyun Kwon et al.
· 1 citation