Towards Robust and Efficient Federated Low-Rank Adaptation with Heterogeneous Clients

TL;DR

This paper introduces LoRA-A$^2$, a novel federated fine-tuning method for LLMs that enhances robustness and reduces communication costs in heterogeneous, resource-limited settings.

Contribution

It proposes LoRA-A$^2$, an adaptive low-rank fine-tuning approach that maintains performance under heterogeneity and low-rank constraints in federated learning.

Findings

LoRA-A$^2$ achieves significant parameter reduction.

It maintains high performance under extreme heterogeneity.

The method enhances robustness and communication efficiency.

Abstract

Federated fine-tuning for Large Language Models (LLMs) faces significant challenges due to the heavy communication overhead of transmitting large model updates. Although Low Rank Adaptation (LoRA) has been proposed as a solution, yet its application in federated learning is complicated by discordance in aggregation. Existing methods addressing this discordance often suffer from performance degradation at low ranks in heterogeneous data settings. In response, we introduce LoRA-A$^2$ (Low Rank Adaptation with Alternating freeze and Adaptive rank selection), which demonstrates robustness in challenging settings with low ranks and high data heterogeneity. Our experimental findings reveal that LoRA-A$^2$ maintains performance even under extreme heterogeneity and low rank conditions, achieving up to a significant reduction in uploaded parameters compared to full fine-tuning without compromising performance. This adaptive mechanism increases robustness and communication efficiency in federated fine-tuning, enabling the practical deployment of LLMs in resource-constrained environments.