SpectR: Dynamically Composing LM Experts with Spectral Routing

William Fleshman, Benjamin Van Durme

Keywords: MoE, routing, merging, LoRA, adapters, experts

TL;DR: SpectR is an approach for routing and merging existing LoRA models per-token and per-layer, without any additional training or data.

Abstract: Training large, general-purpose language models poses significant challenges. The growing availability of specialized *expert* models, fine-tuned from pretrained models for specific tasks or domains, offers a promising alternative. Leveraging the potential of these existing expert models in real-world applications requires effective methods to select or merge the models best suited for a given task. This paper introduces SpectR, an approach for dynamically composing expert models at each time step during inference. Notably, our method requires no additional training and enables flexible, token- and layer-wise model combinations. Our experimental results demonstrate that SpectR improves routing accuracy over alternative training-free methods, increasing task performance across expert domains.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the COLM Code of Ethics on https://colmweb.org/CoE.html

Author Guide: I certify that this submission complies with the submission instructions as described on https://colmweb.org/AuthorGuide.html

Submission Number: 118