Scaling Synthetic-Image Pre-Training for Federated Fine-Tuning of Large Vision Models
FeDiSyn is proposed, a unified framework that holistically considers the interplay between pre-training and FedFT to minimize the overall LVM training time and introduces a scaling law for FedFT pre-training to determine the optimal number of synthetic images, balancing pre-training benefit against generation/pre-train...
Qianpiao Ma, Xiaozhu Song, Junlong Zhou et al.
· 0 citations