Skip to content
Measurements of Three-Level Hierarchical Structure in the Outliers in the Spectrum of Deepnet Hessians (ICML 2019)
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
README.md
hessian.py
main.ipynb
models.py
network.py
requirements.txt
subset_dataset.py
utils.py

README.md

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

You can’t perform that action at this time.