1
What Do Students Learn? A Feature-Level Analysis of Dark Knowledge
用Interaction Tensor框架深挖知识蒸馏中学生模型的特征学习机制，揭示黑暗知识的秘密。
arXiv:2606.03052v1 Announce Type: new Abstract: Knowledge Distillation (KD) is a powerful tool for model compression, yet the precise mechanisms by wh…