Search papers, labs, and topics across Lattice.
FedProxy addresses the challenge of federated fine-tuning of LLMs while protecting IP, ensuring privacy, and mitigating performance loss on heterogeneous data by introducing a Proxy Small Language Model (SLM) as a surrogate for collaborative fine-tuning. The framework consists of server-guided compression to create the proxy, an interference-mitigating aggregation strategy to handle data heterogeneity, and a training-free "plug-in" mechanism to integrate learned knowledge back into the LLM. Experiments demonstrate that FedProxy significantly outperforms Offsite-Tuning methods and approaches centralized performance, setting a new benchmark for secure and high-performance federated LLM adaptation.
Achieve centralized-level performance in federated LLM fine-tuning without compromising IP, privacy, or performance on heterogeneous data by using a compressed "proxy" model.
Federated fine-tuning of Large Language Models (LLMs) is obstructed by a trilemma of challenges: protecting LLMs intellectual property (IP), ensuring client privacy, and mitigating performance loss on heterogeneous data. Existing methods like Offsite-Tuning (OT) secure the LLMs IP by having clients train only lightweight adapters, yet our analysis reveals they suffer from a fundamental performance bottleneck, leaving a significant gap compared to centralized training. To bridge this gap, we introduce FedProxy, a new federated adaptation framework. FedProxy replaces weak adapters with a unified, powerful Proxy Small Language Model (SLM), compressed from the proprietary LLM, to serve as a high-fidelity surrogate for collaborative fine-tuning. Our framework systematically resolves the trilemma through a three-stage architecture: (i) Efficient Representation via server-guided compression to create a resource-friendly proxy; (ii) Robust Optimization through an interference-mitigating aggregation strategy to handle data heterogeneity; and (iii) Effortless Fusion via a training-free"plug-in"mechanism to integrate learned knowledge back into the LLM. Experiments show FedProxy significantly outperforms OT methods and approaches centralized performance, establishing a new benchmark for secure and high-performance federated LLM adaptation.
