Conference
Open access
Aug 2026
Accelerating Federated LoRA Fine-Tuning for LLMs with Adaptive Ranks at Network Edge
FLFAR reduces federated LoRA fine-tuning time by up to 40% while maintaining comparable accuracy, and at the beginning of each global round, the LoRA rank of each device is adapted according to its computational and communication capacity.
Zhi-Hao Zeng, Yu-Xiang Lu, Xiao-Ning Zhang et al.
· 2026 IEEE/CIC International... · 0 citations