Every Expert Matters: Towards Effective Knowledge Distillation for Mixture-of-Experts Language Models

MoE
Re-assign community