Federated Parameter-Efficient Selective Fine-Tuning for Vision-Language Models via Global-Local Masking
This paper proposes a personalized Federated Mask-based Selective PEFT framework, which significantly outperforms the existing federated PEFT method on both seen and unseen classes, and achieves balanced performance on all clients.
Wen-Tao Gu, Xin-Yang Jiang, Jun-Yao Gao et al.
· 0 citations