Every Expert Matters: Towards Effective Knowledge Distillation for Mixture-of-Experts Language Models
With the emergence of Mixture-of-Experts (MoE), the efficient scaling of model size has accelerated the development of large language models in recent years. However, their high memory requirements prevent their use in resource-constrained environments. While knowledge distillation (KD) has been a proven method for model compression, its application to MoE teacher models remains underexplored. Through our investigation, we discover that non-activated experts in MoE models possess valuable knowledge that benefits student models. We further demonstrate that existing KD methods are not optimal for compressing MoE models, as they fail to leverage this knowledge effectively. To address this, we propose two intuitive MoE-specific KD methods for the first time: Knowledge Augmentation (KA) and Student-Aware Router (SAR), both designed to effectively extract knowledge from all experts. Specifically, KA augments knowledge by sampling experts multiple times, while SAR uses all experts and adjusts the expert weights through router training to provide optimal knowledge. Extensive experiments show that our methods outperform conventional KD methods, demonstrating their effectiveness for MoE teacher models.
Code (0)
등록된 구현이 없습니다.
Tasks
Knowledge DistillationMixture-of-ExpertsModel CompressionMethods 이 논문이 사용한 방법론
Similar Papers 제목 키워드 기반
Tailoring Instructions to Student's Learning Levels Boosts Knowledge Distillation
It has been commonly observed that a teacher model with superior performance does not necessarily result in a stronger student, highlighting a discrepancy between current teacher training practices and effective knowledg…
Knowledge Distillationtext-classificationText ClassificationTransfer LearningCoupleFace: Relation Matters for Face Recognition Distillation
Knowledge distillation is an effective method to improve the performance of a lightweight neural network (i.e., student model) by transferring the knowledge of a well-performed neural network (i.e., teacher model), which…
Face RecognitionKnowledge DistillationRelationLLAVADI: What Matters For Multimodal Large Language Models Distillation
The recent surge in Multimodal Large Language Models (MLLMs) has showcased their remarkable potential for achieving generalized intelligence by integrating visual understanding into Large Language Models.Nevertheless, th…
Knowledge DistillationHighlight Every Step: Knowledge Distillation via Collaborative Teaching
High storage and computational costs obstruct deep neural networks to be deployed on resource-constrained devices. Knowledge distillation aims to train a compact student network by transferring knowledge from a larger pr…
Knowledge DistillationEvery Coin Has Two Sides: On the Dual Nature of Generalization in On-Policy Distillation of Large Language Models
On-policy distillation (OPD) transfers teacher capabilities by supervising trajectories sampled from the student's own policy, yet its generalization behavior remains poorly understood, as most studies evaluate OPD on a …