FedTLU: Federated Learning with Targeted Layer Updates

TL;DR

This paper introduces FedTLU, a federated learning method that selectively updates critical layers of language models during fine-tuning, improving convergence and performance in non-IID data scenarios.

Contribution

It proposes a targeted layer update strategy using a scoring mechanism to enhance federated language model fine-tuning under non-IID data conditions.

Findings

Improved convergence in non-IID federated settings

Enhanced model performance through targeted layer updates

Reduced noise and poisoning effects during training

Abstract

Federated learning (FL) addresses privacy concerns in training language models by enabling multiple clients to contribute to the training, without sending their data to others. However, non-IID (identically and independently distributed) data across clients often limits FL's performance. This issue is especially challenging during model fine-tuning, as noise due to variations in clients' data distributions can harm model convergence near stationary points. This paper proposes a targeted layer update strategy for fine-tuning in FL. Instead of randomly updating layers of the language model, as often done in practice, we use a scoring mechanism to identify and update the most critical layers, avoiding excessively noisy or even poisoned updates by freezing the parameters in other layers. We show in extensive experiments that our method improves convergence and performance in non-IID settings, offering a more efficient approach to fine-tuning federated language models.