wiseodd / natural-gradients
Collection of algorithms for approximating Fisher Information Matrix for Natural Gradient (and second order method in general)
☆137Updated 5 years ago
Alternatives and similar repositories for natural-gradients:
Users that are interested in natural-gradients are comparing it to the libraries listed below
- paper lists and information on mean-field theory of deep learning☆75Updated 5 years ago
- Convolutional Neural Tangent Kernel☆109Updated 5 years ago
- Hypergradient descent☆145Updated 8 months ago
- NTK reading group☆88Updated 5 years ago
- Limitations of the Empirical Fisher Approximation☆47Updated 4 years ago
- ☆132Updated 7 years ago
- This repository is no longer maintained. Check☆81Updated 4 years ago
- Repository containing Pytorch code for EKFAC and K-FAC perconditioners.☆141Updated last year
- pyhessian is a TensorFlow module which can be used to estimate Hessian matrices☆24Updated 3 years ago
- Geometric Certifications of Neural Nets☆41Updated 2 years ago
- hessian in pytorch☆186Updated 4 years ago
- ☆121Updated 8 months ago
- Experiments for the Neural Autoregressive Flows paper☆123Updated 3 years ago
- Sample code for running deterministic variational inference to train Bayesian neural networks☆100Updated 6 years ago
- Understanding normalizing flows☆131Updated 5 years ago
- ☆156Updated 2 years ago
- Hessian spectral density estimation in TF and Jax☆121Updated 4 years ago
- Code for "A Spectral Approach to Gradient Estimation for Implicit Distributions" (ICML'18)☆32Updated last year
- PyTorch-SSO: Scalable Second-Order methods in PyTorch☆145Updated last year
- a python implementation of various versions of the information bottleneck, including automated parameter searching☆122Updated 4 years ago
- Computing various norms/measures on over-parametrized neural networks