Fetching the paper…

Self-MoE: Towards Compositional Large Language Models with Self-Specialized Experts · Around