AI Paper Atlas

Research map / Label-efficient and robust learning / Knowledge distillation

Prune Your Model before Distill It

Jinhyuk Park, Albert No

ECCV 2022 · Label-efficient and robust learning › Knowledge distillation · approach: Rethinking distillation objectives · 58 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
prune the VGG19 teacher (36% sparsity) before distilling (ablation)vanilla KD, VGG19 to VGG11 · CIFAR100accuracy: 72.02 -> 72.76Improved
prune the ResNet18 teacher (36% sparsity) before self distillation (ablation)vanilla KD, ResNet18 self distillation · TinyImageNetaccuracy: 57.97 -> 59.39Improved
prune the ResNet18DBL teacher (59% sparsity) before self distillation (ablation)vanilla KD, ResNet18DBL · TinyImageNetaccuracy: 61.35 -> 63.19Improved
prune the ResNet18 teacher (79% sparsity) when the student is MobileNetV2 (ablation)vanilla KD, ResNet18 to MobileNetV2 · TinyImageNetaccuracy: 56.10 -> 57.20Improved
prune the ResNet18 teacher (36% sparsity) when the student is VGG16 (ablation)vanilla KD, ResNet18 to VGG16 · TinyImageNetaccuracy: 54.75 -> 56.35Improved
use a student whose layer widths match the 59% pruned teacher instead of hand-designed halved students (ablation)prune-then-distill, VGG19 · CIFAR100accuracy: 71.43 -> 73.81Improved
prune the teacher to 79% for ResNet18-ST79 student (vs unpruned teacher) (ablation)prune-then-distill, ResNet18-ST79 · TinyImageNetaccuracy: 55.65 -> 56.23Improved
prune the ResNet50 teacher (59% sparsity) in self distillation (ablation)vanilla KD, ResNet50 · TinyImageNetaccuracy: 64.54 -> 65.21Improved
prune the ResNet50 teacher to 59% for the constructed student ResNet50-ST59 (ablation)prune-then-distill, ResNet50-ST59 · TinyImageNetaccuracy: 63.84 -> 63.58Hurt
prune the ResNet50 teacher to 36% for ResNet50-ST36 (ablation)prune-then-distill, ResNet50-ST36 · TinyImageNetaccuracy: 64.11 -> 64.12No clear effect
prune the ResNet18 teacher to 36% for ResNet18-ST36 on ImageNet (ablation)prune-then-distill, ResNet18-ST36 · ImageNetaccuracy: 60.93 -> 61.10Improved
prune the ResNet18 teacher to 79% for ResNet18-ST79 on ImageNet (ablation)prune-then-distill, ResNet18-ST79 · ImageNetaccuracy: 50.24 -> 50.14No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy