Skip to content

Measurements of Three-Level Hierarchical Structure in the Outliers in the Spectrum of Deepnet Hessians (ICML 2019)

Notifications You must be signed in to change notification settings

stjordanis/DeepnetHessian

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Measurements of Three-Level Hierarchical Structure in the Outliers in the Spectrum of Deepnet Hessians (ICML 2019)

For inquiries, please contact papyan@stanford.edu

Features

  • Download 240 models from a GCP bucket:
    • VGG11, ResNet18, DenseNet40
    • MNIST, FashionMNIST, CIFAR10, CIFAR100
    • Various sample sizes
  • Approximate spectrum using the method presented in https://arxiv.org/abs/1811.07062
  • Compute three-level hierarchical structure using the method proposed in this paper

Usage

See Jupyter notebook main.ipynb

Resources

arXiv paper

About

Measurements of Three-Level Hierarchical Structure in the Outliers in the Spectrum of Deepnet Hessians (ICML 2019)

Resources

Code of conduct

Security policy

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Jupyter Notebook 83.7%
  • Python 16.3%