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

KFAC factors are twice as large as BackPACK's and ASDL's #82

Closed
wiseodd opened this issue Mar 4, 2024 · 1 comment
Closed

KFAC factors are twice as large as BackPACK's and ASDL's #82

wiseodd opened this issue Mar 4, 2024 · 1 comment

Comments

@wiseodd
Copy link
Collaborator

wiseodd commented Mar 4, 2024 •
edited
Loading

Check the following test cases. They're for KFACLinearOperator with fisher_type='mc' and fisher_type='type-2', but the same issue also exists for fisher_type='empirical' when compared with ASDL's Empirical KFAC.

`main...wiseodd:curvlinops:compare-kfacs-backpack

Run them using:

pytest -s test/test_kfac.py -k "test_kflr_vs_backpack"                                            
pytest -s test/test_kfac.py -k "test_kfac_mc_vs_backpack"
@wiseodd
Copy link
Collaborator Author

wiseodd commented Mar 8, 2024

Bug found: aleximmer/Laplace#146 (comment)

@runame runame closed this as completed Mar 10, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@wiseodd @runame