1. Self-MoE: Towards Compositional Large Language Models with Self-Specialized Experts
- Author
-
Kang, Junmo, Karlinsky, Leonid, Luo, Hongyin, Wang, Zhen, Hansen, Jacob, Glass, James, Cox, David, Panda, Rameswar, Feris, Rogerio, and Ritter, Alan
- Subjects
Computer Science - Computation and Language ,Computer Science - Machine Learning - Abstract
We present Self-MoE, an approach that transforms a monolithic LLM into a compositional, modular system of self-specialized experts, named MiXSE (MiXture of Self-specialized Experts). Our approach leverages self-specialization, which constructs expert modules using self-generated synthetic data, each equipped with a shared base LLM and incorporating self-optimized routing. This allows for dynamic and capability-specific handling of various target tasks, enhancing overall capabilities, without extensive human-labeled data and added parameters. Our empirical results reveal that specializing LLMs may exhibit potential trade-offs in performances on non-specialized tasks. On the other hand, our Self-MoE demonstrates substantial improvements over the base LLM across diverse benchmarks such as knowledge, reasoning, math, and coding. It also consistently outperforms other methods, including instance merging and weight merging, while offering better flexibility and interpretability by design with semantic experts and routing. Our findings highlight the critical role of modularity and the potential of self-improvement in achieving efficient, scalable, and adaptable systems.
- Published
- 2024