Discover f-dangel/hbp Open Source project by Felix Dangel (@f-dangel)

Stars
20
Rank 1,121,974 (Top 23 %)
Language
Python
License
MIT License
Created over 4 years ago
Updated over 1 year ago

f-dangel

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Hessian backpropagation (HBP): PyTorch extension of backpropagation for block-diagonal curvature matrix approximations

backpack

BackPACK - a backpropagation package built on top of PyTorch which efficiently computes quantities other than the gradient.

Python

555

cockpit

Cockpit: A Practical Debugging Tool for Training Deep Neural Networks

Python

474

unfoldNd

(N=1,2,3)-dimensional unfold (im2col) and fold (col2im) in PyTorch

Python

phd-thesis

Source code for my PhD thesis: Backpropagation Beyond the Gradient

TeX

singd

[ICML 2024] SINGD: KFAC-like Structured Inverse-Free Natural Gradient Descent (http://arxiv.org/abs/2312.05705)

Python

curvlinops

scipy linear operators for the Hessian, Fisher/GGN, and more in PyTorch

Python

vivit

[TMLR 2022] Curvature access through the generalized Gauss-Newton's low-rank structure: Eigenvalues, eigenvectors, directional derivatives & Newton steps

Python

einconv

Convolutions and more as einsum for PyTorch

Python

sirfshampoo

[ICML 2024] SIRFShampoo: Structured inverse- and root-free Shampoo in PyTorch (https://arxiv.org/abs/2402.03496)

Python

phd-thesis-template

LaTeX template for my PhD thesis at the University of Tuebingen

TeX

backobs

Use DeepOBS with BackPACK

Python

org-export-setup

My org-export settings

TeX

python-utilities

Python utility functions I often use

Python

wandb_preempt

Code and tutorial on integrating wandb sweeps with Slurm pre-emption

Python

vivit-experiments

Experiments for the TMLR 2023 paper "ViViT: Curvature Access Through the Generalized Gauss-Newton’s Low-rank Structure"

Python

f-dangel/hbp

f-dangel

Reviews

Repository Details

More Repositories