S'MoRE: Structural Mixture of Residual Experts for Parameter-Efficient LLM Fine-tuning

Qiang Zhang (Dalian Martime University) · Jiayi Liu (Kwai Inc.) · Zhuokai Zhao (Meta) · Lizhu Zhang (Meta) · Hanqing Zeng (Meta AI) · Yinglong Xia (Meta) · Chuan Jiang (Facebook) · Qunshu Zhang (Meta) · Xiangjun Fan (Bytedance Inc.) · Benyu Zhang (CloudBrain)

empirical results fine-tuning graph neural networkhierarchical low-rank decompositioninput token routingllm adaptationlow-rank adaptationsmixture-of-expertsmulti-layer structureparameter efficiencyresidualsstructural flexibilitystructural mixture of residual expertstheoretical analysis

Fine-tuning pre-trained large language models (LLMs) presents a dual challenge of balancing parameter efficiency and model capacity. Existing methods like low-rank adaptations (LoRA) are efficient but lack flexibility, while Mixture-of-Experts (MoE) enhance model capacity at the cost of more & under-utilized parameters. To address these limitations, we propose Structural Mixture of Residual Experts (S’MoRE), a novel framework that seamlessly integrates the efficiency of LoRA with the flexibility of MoE. Conceptually, S’MoRE employs hierarchical low-rank decomposition of expert weights, yielding residuals of varying orders interconnected in a multi-layer structure. By routing input tokens through sub-trees of residuals, S’MoRE emulates the capacity of numerous experts by instantiating and assembling just a few low-rank matrices. We craft the inter-layer propagation of S’MoRE’s residuals as a special type of Graph Neural Network (GNN), and prove that under similar parameter budget, S’MoRE improves structural flexibility of traditional MoE (or Mixture-of-LoRA) by exponential order. Comprehensive theoretical analysis and empirical results demonstrate that S’MoRE achieves superior fine-tuning performance, offering a transformative approach for efficient LLM adaptation. Our implementation is available at: https://github.com/ZimpleX/SMoRE-LLM.