PPFedIT: Towards Privacy-Preserving Federated Instruction Tuning with Few-shot Local Examples

Name: PPFedIT: Towards Privacy-Preserving Federated Instruction Tuning with Few-shot Local Examples
Uploaded: 2026-10-06T20:54:25.084668+00:00
Duration: 958 s
Description: PPFedIT is a federated algorithm designed to improve instruction tuning while protecting privacy in federated learning. It uses synthetic data generation, parameter isolation training, and local aggregation sharing to enhance model performance on few-shot tasks. This method successfully boosts model accuracy by 6% to 13% while reducing privacy leakage by about 20%.

PPFedIT is a federated algorithm designed to improve instruction tuning while protecting privacy in federated learning. It uses synthetic data generation, parameter isolation training, and local aggregation sharing to enhance model performance on few-shot tasks. This method successfully boosts model accuracy by 6% to 13% while reducing privacy leakage by about 20%.

Open the full episode