标签
This paper proposes FedSLM, a parameter-centric framework for federated fine-tuning of foundation models with heterogeneous compressed clients, using SVD-based decomposition and a weak-to-strong elicitation step to handle resource asymmetry. Experiments show it outperforms existing federated baselines while reducing client GPU memory by ~50%.
论文介绍了MESH-FL，一种用于边缘设备上多模态联邦学习的熵引导矩阵乘积态压缩框架。它自适应地分配每层和每种模态的压缩秩，在异构树莓派集群上实现了高达56.8倍的压缩，并在最终准确率上比未压缩的FedAvg提升了2.01%。