Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Conjugate-gradient and Hessian-free optimization #67

Open
SamDuffield opened this issue Apr 16, 2024 · 2 comments
Open

Conjugate-gradient and Hessian-free optimization #67

SamDuffield opened this issue Apr 16, 2024 · 2 comments
Labels
enhancement New feature or request (beyond just a new method) new method New algorithm

Comments

@SamDuffield
Copy link
Contributor

With efficient fvp and ggnvp #63 it might be nice to add the Hessian-free optimization algorithm of https://www.cs.toronto.edu/~jmartens/docs/Deep_HessianFree.pdf

This algorithm runs a short conjugate gradient run at each iteration, so it might be worth implementing posteriors.cg similar to the jax version
@SamDuffield SamDuffield added enhancement New feature or request (beyond just a new method) new method New algorithm labels Apr 16, 2024
@SamDuffield SamDuffield mentioned this issue Apr 18, 2024
Merged
@SamDuffield
Copy link
Contributor Author

SamDuffield commented Apr 18, 2024 •
edited
Loading

Conjugate gradient (posteriors.cg) added in #73 🥳

@KaelanDt
Copy link
Contributor

Goal is to have a hessian_free_opt.py module that includes build, init, and update functions similarly to other modules, which uses posteriors.cg.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
enhancement New feature or request (beyond just a new method) new method New algorithm
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@SamDuffield @KaelanDt