Prune Your Model Before Distill It

Abstract

Knowledge distillation transfers the knowledge from a cumbersome teacher to a small student. Recent results suggest that the student-friendly teacher is more appropriate to distill since it provides more transferable knowledge. In this work, we propose the novel framework, "prune, then distill," that prunes the model first to make it more transferrable and then distill it to the student. We provide several exploratory examples where the pruned teacher teaches better than the original unpruned networks. We further show theoretically that the pruned teacher plays the role of regularizer in distillation, which reduces the generalization error. Based on this result, we propose a novel neural network compression scheme where the student network is formed based on the pruned teacher and then apply the "prune, then distill" strategy. The code is available at https://github.com/ososos888/prune-then-distill

Accepted at ECCV2022

References in corpus (11)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Distilling the Knowledge in a Neural Network
• Bootstrap your own latent: A new approach to self-supervised Learning
• FitNets: Hints for Thin Deep Nets
• Rethinking the Value of Network Pruning
• Scalable Methods for 8-bit Training of Neural Networks
• Comparing Rewinding and Fine-tuning in Neural Network Pruning
• Carbon Emissions and Large Neural Network Training
• DeepCABAC: A Universal Compression Algorithm for Deep Neural Networks
• Knowledge Transfer with Jacobian Matching
• Training Shallow and Thin Networks for Acceleration via Knowledge Distillation with Conditional Adversarial Networks